Do New Forms of Scholarly Communication Provide a Pathway to Open Science?

A thesis submitted to The University of Manchester for the degree of Doctor of Philosophy in the Faculty of Humanities

2015

Yimei Zhu

School of Social Sciences 2

Table of Contents

Table of Contents ...... 2

List of Tables ...... 7

List of Figures ...... 8

Abstract ...... 9

Declaration ...... 10

Acknowledgements ...... 11

Peer Reviewed Publication ...... 12

1. Chapter One: Introduction ...... 13

1.1 Research problems and rationales ...... 13

1.2 Structure of the thesis ...... 16

2. Chapter Two: Background and Literature Review ...... 19

2.1 Introduction ...... 19

2.2 Traditional forms of scholarly communication ...... 19

2.2.1 Reward system and recognition ...... 20

2.2.2 Discipline, seniority and gender differences ...... 22

2.2.3 Peer review and ‘black box’ ...... 23

2.2.4 Communicating academic research ...... 25

2.3 New forms of scholarly communication ...... 27

2.3.1 Open science ...... 27

2.3.2 Academic use of social media ...... 53

2.4 Conclusions and research questions ...... 61

3. Chapter Three: Research Methodology and Scoping Study Findings ...... 63

3.1 Introduction ...... 63

3.2 The rationale of the mixed-methods approach ...... 63

3.3 Scoping studies ...... 64 3

3.3.1 Observation of social media sites ...... 65

3.3.2 Exploratory interviews ...... 65

3.3.3 A case study of #phdchat on Twitter ...... 67

3.3.4 Key findings from scoping studies on social media use ...... 68

3.4 Online survey of academics ...... 72

3.4.1 Initial questionnaire design ...... 72

3.4.2 Developing and piloting the survey ...... 73

3.4.3 Sample justification and strategies ...... 74

3.4.4 Email address harvesting and data cleaning ...... 75

3.4.5 Distribution of the survey ...... 76

3.4.6 Closing of the survey and responses rate ...... 76

3.4.7 Coding of survey responses ...... 78

3.4.8 Data analysis methods ...... 81

3.4.9 Limitation of the survey ...... 82

3.5 Ethical issues ...... 82

4. Chapter Four: To What Extent do Academics Support Open Science? ...... 84

4.1 Research question A: To what extent do academics support and use open access publishing? ...... 84

4.1.1 Overall attitudes and experience of OA publishing ...... 84

4.1.2 Differences in OA publishing by academic discipline ...... 90

4.1.3 Differences in OA publishing by gender ...... 92

4.1.4 Differences in OA publishing by age, job grade and research experience ...... 93

4.2 Research question B: To what extent do academics share primary research data online? ...... 94

4.2.1 Overall attitudes and experience of sharing data online...... 94

4.2.2 Differences in using and sharing data by academic discipline...... 99

4.2.3 Differences in using and sharing data by gender...... 100

4.2.4 Differences in using and sharing data by age, job grade and research experience . 101 4

4.3 Research question C: To what extent do academics publish ongoing research updates on social media?...... 101

4.3.1 Overall attitudes and experience of publishing ongoing research updates on social media ...... 101

4.3.2 Differences in publishing ongoing research updates on social media by academic discipline ...... 105

4.3.3 Differences in publishing ongoing research updates on social media by gender .... 105

4.3.4 Differences in publishing ongoing research updates on social media by age, job grade and research experience ...... 106

4.4 Conclusion and discussion ...... 107

5. Chapter Five Research Question D: To What Extent do Academics Support the Use of Social Media for Research? ...... 113

5.1. To what extent do academics use social media services in their research work? ...... 113

5.1.1 Overall experience of using social media in research work ...... 113

5.1.2 Differences in using social media in research work by academic discipline...... 116

5.1.3 Differences in using social media in research work by gender ...... 118

5.1.4 Differences in using social media in research work by age, job grade and research experience ...... 118

5.2 To what extent do academics use social media to promote their publications? ...... 119

5.2.1 Overall experience of using Twitter, blogs and social networking sites to promote publications ...... 119

5.2.2 Differences in using social media to promote publications by academic discipline 121

5.2.3 Differences in using social media to promote publications by gender ...... 122

5.2.4 Differences in using social media to promote publications by age, job grade and research experience ...... 122

5.3 Academics’ views on the benefits and risks of using social media ...... 123

5.3.1 Overall attitudes towards the benefits and risks related to the use of social media ...... 123

5.3.2 Differences in attitudes towards using social media by academic discipline ...... 127 5

5.3.3 Differences in attitudes towards using social media by gender ...... 127

5.3.4 Differences in attitudes towards using social media by age, job grade and research experience ...... 128

5.4 Conclusion and discussion ...... 129

6. Chapter Six Research Question E: Understanding Factors Associated with Social Media Use for Research and the Support for Open Science...... 133

6.1 Factors associated with the likelihood of using social media for research ...... 133

6.1.1 Introduction and hypotheses ...... 133

6.1.2 Dependent variable ...... 136

6.1.3 Independent variables ...... 136

6.1.4 Logistic regression results ...... 140

6.2 Factors associated with the likelihood of open access publishing ...... 144

6.2.1 Introduction and hypotheses ...... 144

6.2.2 Dependent variables ...... 146

6.2.3 Independent variables ...... 147

6.2.4 Logistic regression results ...... 148

6.3 Factors associated with the likelihood of sharing primary research data ...... 150

6.3.1 Introduction and hypotheses ...... 150

6.3.2 Dependent variable ...... 152

6.3.3 Independent variables ...... 153

6.3.4 Logistic regression results ...... 153

6.4 Factors associated with the likelihood of publishing ongoing research updates on social media ...... 156

6.4.1 Introduction ...... 156

6.4.2 Factors associated with the likelihood of publishing research updates on social media ...... 157

6.4.3 Factors associated with the likelihood of being a super user publishing ongoing research updates on social media...... 159 6

6.5 Conclusion and discussion ...... 160

6.5.1 Gender differences ...... 161

6.5.2 Disciplinary differences ...... 161

6.5.3 Age differences ...... 162

6.5.4 OA publishing ...... 162

6.5.5 Sharing primary research data ...... 163

6.5.6 Using social media for research ...... 163

6.5.7 Limitations ...... 165

7. Chapter Seven: Conclusion and Discussion ...... 166

7.1 Key findings ...... 166

7.1.1 OA publishing ...... 166

7.1.2 Sharing primary research data ...... 168

7.1.3 Using social media for research ...... 169

7.1.4 Disciplinary differences ...... 170

7.2 Limitations ...... 172

7.3 Implications and ways forward ...... 173

7.4 Future work ...... 176

References ...... 177

Appendix 1 Questions for interviewees who were users of social media for research ...... 193

Appendix 2 Questions for interviewees who were non-users of social media for research ..... 195

Appendix 3 Survey invitation email ...... 196

Appendix 4 Survey questions ...... 197

Word count: 62,837

List of Tables

Table 3.1 Details of Interviewees...... 66 Table 3.2 Responses table to the universities questions ...... 78 Table 3.3 Summary of background characteristics of survey respondents (N=1,829) ...... 79 Table 4.1 Attitudes towards the importance of making research articles freely accessible online ...... 84 Table 4.2 Experience of Gold OA (n=1,601 as a subgroup of respondents who had published research articles) ...... 85 Table 4.3 Experience of Green OA (n=1,614 as a subgroup of respondents who had published research articles) ...... 85 Table 4.4 Experience of Gold or/and Green OA (n=1,585 as a subgroup of respondents who had published research articles) ...... 87 Table 4.5 Author fee in last OA publication (n=625 as a subgroup of respondents who published research articles in Gold OA journals) ...... 87 Table 4.6 Preference of Gold OA (n=1,626 as a subgroup of respondents whose work involved publishing research articles) ...... 88 Table 4.7 Experience of Gold OA by academic discipline (n=1,596 as a subgroup of respondents who had published research articles) ...... 90 Table 4.8 Experience of Green OA by academic discipline (n=1,609 as a subgroup of respondents who had published research articles) ...... 91 Table 4.9 Experience of Gold OA by gender (n=1,587 as a subgroup of respondents who had published research articles) ...... 92 Table 4.10 Experience of Green OA by gender (n=1,600 as a subgroup of respondents who had published research articles) ...... 93 Table 4.11 Attitudes towards the importance of making research data available online for reuse ...... 95 Table 4.12 Experience of depositing primary data in online repositories ...... 95 Table 4.13 Experience of using secondary data from online repositories ...... 96 Table 4.14 Use of secondary data by depositing data ...... 96 Table 4.15 Experience of publishing ongoing research updates on Twitter ...... 102 Table 4.16 Experience of publishing ongoing research updates on research blogs ...... 102 Table 4.17 Experience of publishing ongoing research updates on social networking sites (SNs) ...... 103 Table 4.18 Experience of publishing ongoing research updates on social media ...... 103 8

Table 5.1 Experience of using eight social media services for research ...... 113 Table 5.2 Experience of gathering research information and publishing ongoing research updates on Twitter ...... 115 Table 5.3 Frequency of reading, commenting and posting updates on research blogs ...... 115 Table 5.4 Use of six social media services in research work by academic discipline ...... 118 Table 5.5 Experience of promoting recent peer-reviewed publication (n=1,545 as a subgroup of respondents who had published research articles) ...... 120 Table 5.6 Experience of promoting recent peer-reviewed publication regrouped (n=1,545 as a subgroup of respondents who had published research articles) ...... 121 Table 5.7 Experience of promoting recent peer-reviewed publications on social media by academic discipline (n=1,486 as a subgroup of respondents who had published research articles) ...... 121 Table 6.1 Factor loadings of eleven attitudes items on three rotated factors...... 139 Table 6.2 Logistic regression analysis on the likelihood of using social media for research ..... 141 Table 6.3 Logistic regression analysis on the likelihood of publishing in Gold OA and Green OA ...... 148 Table 6.4 Logistic regression analysis on the likelihood of sharing primary research data ...... 154 Table 6.5 Logistic regression analysis on the likelihood of publishing ongoing research updates on social media ...... 158 Table 6.6 Logistic regression analysis on the likelihood of being a super user compared to an occasional user publishing ongoing research updates on social media ...... 160

List of Figures

Figure 5.1 Use of Twitter in research work by academic discipline ...... 116 Figure 5.2 Use of research blogs in research work by academic discipline ...... 117 Figure 5.3 Attitudes towards the positive effects of using social media in research work ...... 124 Figure 5.4 Attitudes towards the negative effects of using social media in research work ...... 125

Abstract

Do New Forms of Scholarly Communication Provide a Pathway to Open Science? Yimei Zhu Doctor of Philosophy University of Manchester 9 June 2015

This thesis explores new forms of scholarly communication and the practice of open science among UK based academics. Open science broadly refers to practices that allow cost-free open access to academic research. Three aspects of open science are examined in this study: open access to research articles; open access to research data; and publishing ongoing research updates using social media.

The study employs a mixed-methods approach, combining a series of scoping studies using qualitative methods followed up by an Internet survey of 1,829 UK academics. Overall this thesis has shown that whilst there is support for open science, the use of open science by academics was limited. Many academics were not aware of RCUK’s open access policy and had limited experience of making their research articles freely accessible online. Most academics did not share their primary research data online. Although some academics had used a range of social media tools to communicate their research, the majority had not used social media in their research work. Overall, male, older and senior academics were more likely to use open access publishing and share primary research data, but were less likely to use social media for research. Academics based in Medical and Natural Sciences were more likely to use open access publishing and share research data, but less likely to use social media for their research compared to academics from Humanities and Social Sciences. Academics who were aware of RCUK’s open access policy and who recognised the citation advantages of open access were more likely to publish in open access journals. Academics that were aware of RCUK’s open access policy and had used social media for research were more likely to self-archive research articles. Academics that had used secondary data collected by others and self- archived research papers were more likely to share their own primary research data. Academics seemed to be strongly influenced by their colleagues’ recommendation for the adoption of social media in research. Those who considered that the general public should know about their research findings were more likely to share their research on social media. A group of academics were identified and described as super users who frequently communicated ongoing research on social media. These super users were more likely to use tablet computers and have received social media training organised by their institutions.

It is clear that open science is going to be a major factor in future academic work and in relation to building an academic career. Many academics have recognised the importance of open science. However to date the use of the tools for open science has been limited. With the right guidance and reinforcement of relevant policies, the new forms of scholarly communication can provide a pathway to open science which would serve to benefit individual academics, research communities and the public good.

Declaration

I declare that no portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning.

Copyright Statement i. The Author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the “Copyright”) and she has given the University of Manchester certain rights to use such Copyright, including for administrative purposes. ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issues under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must for part of any such copies made. iii. The ownership of certain Copyright, patents, designs, trade marks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the Author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. iv. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy (see http://documents.manchester.ac.uk/DocuInfo.aspx?DocID=487), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s regulations (see http://www.manchester.ac.uk/library/aboutus/regulations) and in The University’s policy on Presentation of Theses.

Acknowledgements

I am grateful to my parents and family who have been very supportive and tolerating my long absence in pursuing my PhD.

I am heartily thankful to my supervisors, Kingsley Purdam, Martin Everett and Rob Procter, whose encouragement, guidance and support from the initial to the final stage enabled me to complete this thesis. Without their guidance and support this thesis would not have been possible.

I would like to express my gratitude to Maria Pampaka, Elisa Bellotti, Mark Brown, Nick Crossley, Helene Snee and my other colleagues in Sociology and Social Statistics who offered feedback and support at different stages of my PhD. I want to further thank my friends and fellow PhD colleagues including Kevin, Gagun, Adi, Tessa, Mollie, Ana, Jimmy, Iheanyi, Gunan, Huixin, Wen, Manxu and the many others for their help and support in both my work and life.

I give my regards to all the participants in this study who selflessly gave their time and insights in response to my survey and pilot studies.

Finally to the memory of my dear grandparents and my friend John. My grandparents were both doctors and brilliant intellectuals. Through their example my grandparents have always been a great inspiration to me. And to my best friend John. John, you were like family to me and helped me through those difficult times in my early years living in the UK, you are always missed.

Peer Reviewed Publication

Zhu, Y. & Procter, R. (2015). Use of blogs, Twitter and Facebook by UK PhD students for scholarly communication. Observatorio (OBS*) Journal, Vol 9, no2, p. 029-046. http://obs.obercom.pt/index.php/obs/article/view/842

Zhu, Y. (2014) ‘Seeking and sharing research information on social media: A 2013 survey of scholarly communication' in Rospigliosi, Asher & Greener, Sue (ed.) Proceedings of European Conference on Social Media ECSM 2014, 10-11 July 2014, University of Brighton. Brighton: Academic Conferences & Publishing International, p. 705-712. https://www.escholar.manchester.ac.uk/item/?pid=uk-ac-man-scw:229261

1. Chapter One: Introduction

This thesis explores new forms of scholarly communication and the practice of open science among UK based academics. Open science is the idea of ‘making the whole research process as transparent and accessible as possible’ (Scheliga and Friesike 2014). A systematic review of open science will be conducted in Section 2.3. Traditional forms of scholarly communication heavily rely on formal peer-reviewed publication which is time consuming and can take over a year for an article to be processed and published in an academic journal. Moreover, these publications are usually behind a pay-wall requiring subscription fee for access. New forms of scholarly communication rely on digital technologies which provide online access to research articles, data, methods and tools potentially free of charge. These new forms include open access publishing, sharing primary research data, publishing research updates online as well as using social media for various reasons in research. However, new forms of scholarly communication can challenge the traditional reward systems and individuals’ old habits. This thesis will investigate to what extent academics support open science and using social media in research as well as what factors are associated with such activities.

1.1 Research problems and rationales

The Internet has reduced the constraints of time and space, and changed the way people communicate with each other (Hewitt 1998). These changes have also made their way into science and scholarly communication. Scholarly communication, has been used as a broad term to cover all the activities and norms of academic research related to producing, exchanging and disseminating knowledge (Rieger 2010; Hahn et al. 2011). It is often used to refer primarily to the process of peer-reviewed publication, which used to depend on scholarly publication in print but has now largely been transformed to relate to electronic formats on the World Wide Web and access to those online contents (Nicholas et al. 2009). In recent years, subject-based and institutional repositories, data centres, open source software and public copyright licenses have emerged along with online publishing. Many scientific records, including published articles, data, presentation slides, methods and tools have been made freely available to all internet users via those new platforms (Björk 2004; Wilbanks 2006; De Roure et al. 2010; RIN 2011a). While scholars previously depended on browsing the most relevant published papers to stay current with the research in their field, and communicating face-to-face in meetings and conferences to exchange new information and research findings, the adoption of online social media, such as academic blogs, Twitter and social networking sites have enabled real-time communication and dissemination of scientific contents (Maron and Smith 2008; Neylon and Wu 2009a). In the new 14

media age, scientists no longer have to travel to another country for conferences to hear the latest news or wait for a long period of time to read the findings of new research. Academic journals are published online and thus search engines and online databases are deployed to search for research articles (Meyer et al. 2011). In recent years, talks of weaknesses in the traditional peer review systems and the publishing crisis have developed and led to wider debates in scholar communities (Procter et al. 2010b). The subscription-based publishing models and the long peer-review processes both hinder timely dissemination of research outputs to wider audience. The notions of ‘open science’ emerged in this context.

The historical origins of open science can be traced to the early seventeenth century and refers to the disclosure of new knowledge in contrast to private property and secrecy (David 2008). Digital technologies have opened up new opportunities for the contemporary form of open science (Scheliga and Friesike 2014). Nielsen (2011) defines open science as ‘the idea that scientific knowledge of all kinds should be openly shared as early as is practical in the discovery process’. The core ideas of open science are openness and sharing. As Neylon and Wu (2009b: 540) point out, openness is arguably the ‘great strength of the scientific method’. It makes it possible to provide scientific information, articles, data, methods and tools available online to everyone with internet access. This novel practice of open science has been welcomed by many scholars as an approach to advance science and improve efficiency because in this way, scientific information and records can be quickly picked up and used by other scientists, thus avoiding duplication (Piwowar et al. 2007; Bukvova 2011). For example, in Medical Science, a project called AllTrials calls for all past and present clinical trials to be registered online and their full methods and summary results reported so that trials would not be repeated unnecessarily1.

The rise of public interests and engagement in science, especially publicly funded research, calls for a more open environment within the scientific research process and the sharing of outputs (Gray 2009). Gray’s rationale is that the paywall of research articles prevented the public from accessing good information in the form of peer reviewed academic research. Instead, the public may turn to unreliable information from the Internet. This is an inefficient use of public money.

In academia, open access (OA) policy has been discussed among various academic communities and a number of research councils in the UK have had policies on open access since 2005. From April 2013, Research Council UK’s (RCUK) open access policy came into effect which

1 http://www.alltrials.net/ accessed 30 December 2014 15

required RCUK-funded research to be published either through open access journals or self- archiving (RCUK 2013). RCUK arranged to pay block grants to universities and other institutions to support the cost of Article Processing Charges (APC) for publishing in Gold OA journals. Many universities also started institutional repositories to support Green OA model (e.g. PURE for Bristol University2 and ORA for Oxford University3 ). In July 2014, the Higher Education Funding Council for England (HEFCE) also introduced an open access policy which applies to research outputs accepted for publication after 1 April 2016 in relation to the research assessments after the 2014 REF4.

Along with the open access movement for publishing research articles, a growing chorus of voices have advocated open access to research data, namely the Open Data Movement (Nicholson and Bennett 2011). A study found that by 2011, eight of the nine major UK research funders5 had a data policy, with seven of them preparing to fund data sharing activities within a project bid (RIN 2011a). Mover, social media tools have provided novel distribution channels for research outputs. Rather than waiting for the long process of publishing in peer-reviewed journals, academics may share ongoing research updates on research blogs, such as the Open Notebook Science project (RIN 2010). Social media enable almost real-time science communication through the World Wide Web and new media platforms such as smartphones and tablet computers. Academics can use social media not only to disseminate research outputs to a wider audience but also to search for research information, promote their publications and for building contacts and networking with peers. Social media tools such as Twitter have become popular among academic users and are found to be effective information resources as well as dissemination channels (Gu and Widén-Wulff 2011).

However, open science practice faces barriers and risks. The high cost of APCs for publishing in OA journals is a barrier for non-funded academics (Björk 2004). The author-pays model can also lead to quality control problems for OA journals. The peer review process of OA journals has been questioned and many OA journals were found to accept seriously flawed papers (Bohannon 2013). Academics are also concerned that OA journals are not established in their field and have not enough impact (Swan and Brown 2004). Authors can face legal restrictions to self-archive published papers, as the publishers of traditional journals usually own the copyrights to the

2 http://www.bristol.ac.uk/library/support/research/rcuk.html accessed 30 December 2014 3 http://www.bodleian.ox.ac.uk/ora/deposit-in-ora accessed 30 December 2014 4 http://www.hefce.ac.uk/whatwedo/rsrch/rinfrastruct/oa/policy/ accessed 8 December 2014 5 Arts and Humanities Research Council (AHRC); Biotechnology and Biological Sciences Research Council (BBSRC); Cancer Research UK (CR-UK); Engineering and Physical Sciences Research Council (EPSRC); Economic and Social Research Council (ESRC); Medical Research Council (MRC); Natural Environment Research Council (NERC); Science and Technology Facilities Council (STFC) & Welcome Trust. 16

articles (Howard 2013). The barriers to data sharing include the maintenance and improvement of privacy, technology, and standards (Gardner et al. 2003).

Moreover, the traditional practice of science is based on the academic reward system that emphasises the priority of discovery based on the quality and quantity of formal publication (Merton 1957;Cole and Cole 1967). Under this competitive system, researchers are inclined to be careful or sceptical towards openness and sharing as they strive to be the first to publish new knowledge and to receive more citations. For example, academics may be reluctant to share their research data because of the lack of incentives, issues of data ownership and lack of trust in other users (Zimmerman 2008). Cox and Forshaw (2011) criticized scientists who blog their research as being unfair to collaborators because this could leak results and circumvent the peer review process. A researcher may prefer not to be open in sharing information to avoid the possibility of a rival researcher getting ahead in the research race or worse, taking the credit for their work.

Thus, research communities face the challenge of encouraging openness and sharing of scientific research without undermining competition. With this in mind, the following questions should be considered:

 What are the benefits of practising open science? And what are the concerns of those who are not practising open science?

 To what extent do academics support open science and what factors are associated with these attitudes and activities?

 To what extent do academics use social media to communicate research and what factors are associated with these practices?

In order to answer the above questions, this thesis first reviews relevant literature and then employs a mixed-methods approach to explore the current practice of open science among UK academics, their attitudes and experiences towards various aspects of open science and to what extent academics use social media in research work.

1.2 Structure of the thesis

There are seven chapters in this thesis.

Chapter Two provides the background of scholarly communication from the traditional forms to the new forms. Traditional forms of scholarly communication are based on an academic reward system that emphasises the priority of discovery and blind peer review. The thesis will discuss the 17

academic reward system, discipline, seniority and gender differences, peer review system and communicating academic research. New forms of scholarly communication emerged with the development of the internet. These new forms include open access publishing of research articles, open access to research data, publishing research updates online as well as using social media for various reasons in research work. An analytical review of these new forms, including rationales, target audience, barriers, policies, and discipline difference will be discussed with examples of relevant policies and empirical studies. The emerging research questions will be identified at the end of this Chapter.

Chapter Three offers a detailed outline of the overall research design and methodology. It first explains the rationale behind the employment of a mixed-methods approach, which incorporates qualitative scoping studies and an internet survey. It will then introduce specific techniques and procedures of the research methods, followed by study approval and ethical issues.

Chapters Four, Five and Six present the findings of the study. Chapter Four presents the results of the survey respondents’ attitudes towards and experience of the three aspects of open science: allowing open access to research articles; sharing research data online; and publishing ongoing research updates on social media. Each aspect will first discuss the overall attitudes and experiences based on the responses to the survey. Some key issues will be illustrated with examples of quoted comments from survey respondents. It will then examine whether there are any differences in terms of academic discipline, gender, age, job grade and research experience using descriptive analyses of the survey data.

Chapter Five presents the survey results of the respondents’ attitudes towards and experience of using various social media services for research. It will first discuss the overall experience of using social media in research work. It will then discuss respondents’ experiences of using social media to promote publications and their attitudes towards the benefits and risks related to the adoption of social media. To be consistent with Chapter Four, each aspect of social media use will be examined using descriptive analyses to determine whether there are any differences in use between respondents by discipline, gender, age, job grade and research experience.

Chapter Six explores the factors associated with academics’ practices in supporting open science and using social media for research. Logistic regression analysis of the survey data aims to determine what factors are associated with the likelihood of using social media in research work, publishing in Gold and Green OA channels, sharing primary research data and publishing ongoing research on social media including the likelihood of being a super user. Background characteristics 18

discussed in Chapter Four and Five are then examined after taking account of other factors to determine whether there are statistically significant differences between different groups by gender, discipline and age. Various factors are examined to determine their relevance with the likelihood of various open science practices after controlling for background characteristics.

Finally Chapter Seven draws the main conclusions that have been identified in the research.

2. Chapter Two: Background and Literature Review

Summary: Traditional scientific practice is based on an academic reward system that emphasises the priority of discovery. Under this principle, academics tend to share their outputs only after securing the priority of discovery by putting them in print first. Moreover, scholarly publications usually include only significant findings and successful results rather than failures, detailed research process, tools or datasets. However, open science challenges the traditional reward system by welcoming openness and sharing, and opening up the research process for wider academic and public participation. Open science broadly refers to practices that allow cost- free open access to academic research. Three aspects of open science are identified: open access to research articles, open access to research data and publishing ongoing research updates online. Social media have been used to not only communicate research, but also search for information, promote publication and facilitate networking with peers. After reviewing the traditional and new forms of scholarly communication, five research questions are identified for this study.

2.1 Introduction

In this chapter the traditional forms of scholarly communication will first be reviewed, utilising American Sociologist Robert Merton’s theory of ‘priority’ and ‘reward system’ (Merton 1957: 635) to examine scholarly communication within traditional science. New forms of scholarly communication include the use of open science and various uses of social media for research. Three aspects of open science practice – open science to research articles, open access to research data and publishing ongoing research updates online – will be discussed along with their advantages, barriers, relevant policies and strategies. The academic use of social media will include a discussion of blogs, Twitter, social networking sites and findings from empirical studies. After reviewing the existing literature, the emerging research questions will be identified at the end of this chapter.

2.2 Traditional forms of scholarly communication

Traditional scientific practice is based on an academic reward system that emphasises the priority of discovery. Since the nineteenth century, the publication of articles in journals and the relative prestige of the journals in which they are published have become predominant indicators of professional performance for researchers and the institutions that employ them (Merton 1957; 20

Schauder 1993; Correia and Teixeira 2005). Moreover, scholarly publications usually include only significant findings and successful results rather than failures, detailed research processes, tools or datasets. In Section 2.2, the history of scholarly communication will be reviewed.

2.2.1 Reward system and recognition

Over many centuries, alongside the development of science, the scientific reward system has become one of the most important norms for academics to follow. Merton (1957: 635) reviewed the history of science and discussed this reward system in his prominent paper ‘Priority in Scientific Discovery’. Merton noted that during the last three centuries, since modern science developed, disputes over the priority of discovery have been highly debated by numerous scientists. For example, Galileo campaigned to fight for his rights to priority to a number of discoveries and inventions, claiming that four of his rivals had taken credit for things he had discovered first. Newton battled with Robert Hooke, Leibniz and Wallace over priority in discoveries over a long period of his career. Merton criticised explanations of human nature and egotism and blamed the conflict on the institutional norms of science. These norms place emphasis on originality – ‘a major institutional goal of modern science’ (Merton 1957: 645). The desire for recognition of originality pressures scientists to assert claims on their scientific work through publication.

Scholarly journals were introduced in Western Europe in the 1600s (Hahn et al. 2011). The earliest scholarly journals, such as the Philosophical Transactions of the Royal Society created by Henry Oldenburg in 1665, facilitated communication and dissemination of important scientific discoveries to the wider research community (Guédon 2001; Correia and Teixeira 2005). Meanwhile, publication has become a solution for scientists’ frustration because scholarly journals could operate as ‘a public registry of discoveries’ and hence secure the ‘ownership of ideas’ (Guédon 2001: 6). For most researchers, getting their work published ‘becomes a symbolic equivalent to making a significant discovery’ (Merton 1957: 655). Since recognition became the symbol and reward for excellence in science, the race to be published became a critical objective in scientific research.

The cultural emphasis on originality and what can be termed the publication race occasionally led to negative responses and inappropriate behaviour such as fraud and forgery (ibid.). Some academics faked scientific evidence in order to obtain the reward of fame and recognition. Others were suspicious towards their rivals and falsely accused them of plagiarising. Fraud, fakery and plagiarism, as LaFollette (1992: 1) argues, ‘contradict every natural expectation for how scientists 21

act’, and they challenge the trustworthiness of scientific communities. However, even scientists who were employed by the leading universities, such as Harvard and Yale, and published in top- ranked journals, were accused of ‘forging, faking or plagiarising their way to success’ (LaFollette 1992: 1). A survey study with early- and mid-career Health scientists (n=3,247) in the United States found that 0.3% of the respondents admitted to falsifying or ‘cooking’ research data, while 1.4% admitted to plagiarism (Martinson et al. 2005: 737). It is common for high-profile journals such as Nature to have more retractions of research articles (Nature 2014). Steen (2011) found that fraudulent authors were more likely to target journals with higher impact factors.

Although sharing one’s knowledge is seen as a moral imperative of science, scientists only tend to share after securing the priority of discovery by proclaiming it to the world first. This competition in science has intensified alongside the expansion of scientific knowledge and increase in research funding over the past centuries. In most academic institutions, the measure of scholarly accomplishment has been dependant on the quantity and quality of publication (Merton 1957). The quality of publication is usually associated with the rank of the journal and the citation rates of the publication. That is to say, priority competition among scientists occurs in traditional science where the person who is the first to make a significant discovery is rewarded. Scientists are able to claim the priority by getting their discoveries into academic publications. Hence, publication has become a symbolic measurement for academic success and a standard assessment for researchers’ careers.

Scientific recognition usually depends on work being ‘published in a reputable scientific journal’ and published articles being ‘cited in papers by other scientists’ (Ziman 1987: 70). First of all, a published article is more significant when it is cited by other scientific papers and its significance increases with the number of citations. In the 1960s and 1970s, many institutions and research centres started to build an objective scale to measure the performance of researchers (Guédon 2001). Garfield proposed a ‘citation index’ as a new bibliographic tool in 1955, and in 1960 he founded the Institute for Scientific Information (now Thomson Reuters, Philadelphia, US), which started a bibliographic indexing service called the Scientific Citation Index (Garfield 2006). The Citation Index provides bibliometrics, which is a quantitative analysis of the units of scholarly communication (e.g. published papers, book chapters, etc.) and the citations that connect them. The current most well-known online citation index is Thomson Reuters’ Web of Science, which contains approximately 40 million bibliographic records and over 550 million citations from the past century (Craig et al. 2007). Several websites also provide a digital indexing service, including 22

GoogleScholar, Scopus and NASA’s Astrophysics Data System. The Science Citation Index has become a career management tool for scholars following its development and popularity.

Secondly, academics aim for their research to be published in high standard journals. Thus the quality of journals needs to be evaluated. In 1975, Garfield (1976) introduced Journal Citation Reports (JCR), as an instrument to evaluate the significance of scientific journals. The best-known metric produced by the JCR is the Journal Impact Factor (JIF) – the ‘ratio of the current year citations to the source items published in the journal during the previous two years’ (Alexandrov 2011: 1). Over the years, JCR has become an authority for evaluating scholarly journals predominantly through the JIF (Bornmann et al 2011). The journals with a higher impact factor in each discipline were identified as core journals in which scientists, especially those in their early career stage, must publish in order to secure professional status and advance their career (Guédon 2001). Rowlands and Nicholas (2006) conducted a survey with researchers from various countries across the world and found that early and mid-career researchers specifically regard impact factors as highly influential in choosing where to publish their work. A publication in a high-rank journal (with a high impact factor) is considered to be a sign of distinction and may have more chance of being cited (Alexandrov 2011).

2.2.2 Discipline, seniority and gender differences Publication competition varies in its intensity and impact in terms of different discipline areas and seniority. Competition can be more intense in fields such as Chemistry, Physics and Biology than in Social Sciences (Hagstrom 1974). Over half of the respondents in Hagstrom’s survey study with natural scientists had been anticipated by another scientist in the publication of a discovery in their careers. Compared to Natural Sciences, in Humanities, there is less collaboration and academics tend to work alone (Stone 1982). Academics in Humanities and Social Sciences were reported to be much less likely to have co-authorship for publications compared to those in Medical and Natural Sciences (Kyvik 2003). While scientific journal articles are the dominant publication form for academics in Medical and Natural Sciences, monographs were found to have great value for career advancement in the Arts and Humanities (Williams et al. 2009). A number of studies have found that monographs were cited far more frequently than journal articles in Humanities (Thompson 2002). Larivière et al. (2006) also found that journal literature accounts for less than half of the citations in several disciplines of the Humanities and Social Sciences, whilst journal articles dominate the core literature for the dissemination of knowledge in Natural Sciences and Engineering. Overall, academics in Medical and Natural Sciences have higher productivity than those in Humanities and Social Sciences in terms of the numbers of publication 23

including journal articles, books and reports (Kyvik 2003). In some Humanities Disciplines such as Arts, Music Composition and Drama, academics conduct practice-based research and submit performance, composition or artefact for Research Assessment Exercise (Oppenheim and Summers 2008). Hence, journal articles may not be relevant for academics in these disciplines in terms of career advancement.

In many Sciences Disciplines, journal articles are written in scientific language which can be difficult to understand by non-specialist readers (Bucchi 2004). However, in some Humanities disciplines, research outputs may target public audience and they may be in forms such as museum exhibitions, paintings & music other than academic writing.

Young and junior academics have the disadvantage of getting disproportionately little credit for their work when they collaborate with eminent academics who get disproportionate great credit for their contributions to published work (Merton 1968). This cumulative advantage was also confirmed by Cole and Cole (1967) who found that academic reward system operated to the advantage of established scientists while discouraging those less established from further research.

In the last twenty years, a number of empirical studies were carried out to study gender differences in higher education and science (Dehdarirad et al. 2014). Gender differences were evident in job status, job mobility and academic achievements (Fox 2001; Børing et al. 2010; Hopkins et al. 2013). Gender differences in science careers and productivities were evident that men usually perform better than women (Hermanowicz 2012). Xie and Shauman (1998) found that gender differences in publication productivity could be explained by personal characteristics such as academic field and years of research experiences, types of universities, teaching loads, research grants and assistance. Fox (2010) suggests structural barriers such as the gendered expectations of women being in households and families rather than in science keep female scientists from accomplished like men. Sonnert and Holton (1996) found that women were less like to consider themselves as self-confident and more likely to have unclear career aspirations when starting out in science. These gender differences in job status, mobility, achievements and productivity may have different impact on female and male academics’ opportunities to publish.

2.2.3 Peer review and ‘black box’ In traditional science, the institutional norm and standards have developed a structure for researchers to follow. All the publications of scientific knowledge need to be peer reviewed and follow a certain structure and format. The peer review process has functioned as the ‘gatekeeper 24

of science’ – the articles submitted to an academic journal have to be evaluated by experts in those specific fields to ensure their quality (Watts 2007: 2). The beginnings of the scientific peer review process are often attributed to the 18th century, when the Royal Society of London took over editorial responsibility for Philosophical Transactions in 1752, at which time it adopted a review procedure for its publications (Kronick 1990; Spier 2002). However it has been suggested that peer review did not become institutionalised as the accepted practice in industrialised countries until after the Second World War (Popescu 2007). It is now common practice that all academic papers must go through a ‘double-blind’ peer review process (or ‘single-blind’ in some discipline areas such as Computer Science and Material Science), in which the reviewer and author’s identities are concealed from one another, in order to be published in scholarly journals.

However, publications usually include only significant findings and successful results rather than failures (Merton 1957). Neither the data nor research tools used are routinely contained in publications. This raises concerns about the peer review process itself: without primary research data and metadata peer reviewers are unable to run the data to test the validity and credibility of research results. This type of traditional practice has put science into a ‘black box’ in which only the outputs can be studied rather than the processes through which scientific knowledge is produced (Whitley 1972). Moreover, in traditional science, publications are often the commercial property of publishers and, as such, are not usually openly accessible to the public. In order to access those publications, researchers or their institutions must pay subscription fee (Björk 2004). The general public and tax payers, who have paid for the publicly-funded research, have little power over the access to those research outputs.

There is also a critique of how peer review works in practice. Academic journal’s peer review process could fail to identify flawed research articles. There was evidence of peer reviewed articles being faked and retracted, such as Wakefield’s fraudulent articles linking the MMR vaccine and autism (Deer 2010). The shortcomings of journal peer review system has led to ‘reproducibility crisis’ when scientific experiments sometimes yield results that turn out to be incorrect and irreproducible (Johnson 2015). One of the cornerstones of science’s claim to objective truth requires the same experiments result in same findings regardless of the scientists who conduct them; otherwise, either the original research or the replications are flawed (The Economist 2013). Publishers including those well-known legitimate ones such as Elsevier, Springer and IEEE, were criticised of faking peer review because they accepted articles written by SCIgen, an algorithm that ‘strings together nonsensical phrases larded with computer science buzzwords’ and ‘some phony references’ (Seife 2015). It was likely that these publishers did not send these 25

articles for peer review by other computer scientists because a real computer scientist can easily spot these errors. Some journal editors were reported to be reluctant to overturn favoured scientists’ reviews to ensure that they would not stop submitting their work to that journal (Henderson 2010). The peer review scam also exposed individual author scammers. A number of academic authors were caught reviewing their own papers by recommending themselves as reviewers by a different surname, creating bogus email addresses or hacking the publishers’ computerized reviewing systems (Ferguson et al. 2014).

This has led to the debate over a possible future of open science which calls for a greater transparency of the whole research processes (Gezelter 2009).

2.2.4 Communicating academic research In the 16th and 17th centuries in Europe, the scientific revolution changed the dominant norm of ‘secrecy in the pursuit of nature’s secrets’ to the rapid disclosure and broader dissemination of scientific discoveries (David 1998: 16). Researchers are required to accumulate knowledge over time in their fields in order to extend that knowledge, and reduce excess duplication of research efforts. Openness in David’s Open Science theory refers to knowledge dissemination in contrast to private property and secrecy of knowledge or discoveries. This interpretation of ‘open science’ is often adopted in industrial organisation studies. For example, Ding (2011) investigated the For- Profit Biotechnology firms’ adoption of open science by studying the firms’ published papers in scientific journals. In these studies, open science is a system for cumulative knowledge production by the disclosure of knowledge through publication, while organisations and their researchers pay the expense of accessing the knowledge (Mukherjee and Stern 2009). This type of definition of ‘open science’ refers to providing access to research outputs with a subscription fee (David 1998; Willinsky 2010); however, research findings being published is now the norm in academia and David’s definition of open science is in fact traditional scholarly communication. Today open science means no limitation of paywalls and more detailed discussion will be carried out in Section 2.3.

Scholarly communication, has been used as a broad term to cover all the activities and norms of academic research related to producing, exchanging and disseminating knowledge (Rieger 2010; Hahn et al. 2011). Scholarly communication traditionally relies on publishing peer-reviewed articles and books by scholars and for scholars. Science communication, on the other hand, refers to public communication of science or the ‘popularization of science’ to the lay audience from the public (Bucchi 2004). However, public science communication is not necessarily a distinct 26

different activity from scholarly communication. Cloitre and Shinn (1985) argue that scientific popularisation can offer to inform both scientists and lay public audience of recent discoveries and advances. Scholarly communication can also involve disseminating knowledge to non- scientists with interests and concerns. Moreover, in Humanities and Social Sciences, the public are often the target audience and service users of academic research while scholarly communication may aim to disseminate research findings to the public to enhance their well-being or to influence policy making.

Communicating academic research to the public has been included in policies by UK’s major funders. UK based academics are now asked about their public engagement plans when applying for research funding; however, there is no formal requirement if not completed (Pearson 2001). The rationale of public science communication is explained by the Wolfendale Report, ‘Scientists, engineers and research students in receipt of public funds have a duty to explain their work to the general public’ (Wolfendale Committee 1995). Poliakoff and Webb (2007) investigated the factors that influence scientists’ decision to take part in public engagement activities. They found four factors including past behaviour in participation, positive attitude toward participation, confidence in their ability to participate and belief of colleagues’ involvement. Besley (2014) explored factors influencing scientists’ online engagement in science communication and found significant factors including past engagement behaviour, funding source, perceptions of efficacy, and a belief in the need for scientists to contribute to public debates.

However, the dominant view of scientific popularisation see itself as a one-way process of simplification which views scholarly articles as the originals of scientific knowledge being debased by translation and mediation for the public who are ignorant of such matters (Hilgartner 1990). Davies (2008) also found that majority of scientists and engineers in her study view the constructions of science communication as one-way and negative. This dominant view of one- way transfers of information as ‘education’ has been criticised as it fails to take into account the interaction between the public and scientists (Myers 2003). These discourses have raised a question of how science should be communicated to the public.

Before the age of internet, academics either go through the traditional mass media such as television, radio, and newspaper or organise public events and workshops to communicate their research to the public. In both US and France, approximately 80% of scientists reported experience of popularising science through the mass media (Bucchi 2004). The Internet and new media have offered new opportunities for the scientists to engage public in science and raise public awareness which will be discussed in Section 2.3. 27

The next section will review open science and new forms of scholarly communication.

2.3 New forms of scholarly communication

New forms of scholarly communication emerged with the development of the Internet and digital technology which enabled more open practices within the academic community. These new forms include open access publishing, sharing primary research data, publishing research updates online as well as using social media for various reasons in research work, such as searching for research information, promoting publications and networking with peers. The Open Science Movement emerged in this new age. Various aspects of open science will be reviewed in Section 2.3.1. These aspects are：(i) the open access movement of publishing – enabled by many open access journals and online repositories; (ii) the Open Data Movement – free access to data, tools and methods. This form has become more common in recent years under the emergence of data policies and data centres. It has been commonly practiced in some disciplines, such as the biomedical field, more than others; and (iii) publishing ongoing research updates online – a new phenomenon of research practice enabled by social media, which calls on real-time communication of academic research. Section 2.3.2 will discuss other uses of social media in research work.

2.3.1 Open science Nielsen (2011) defines open science as ‘the idea that scientific knowledge of all kinds should be openly shared as early as is practical in the discovery process’. Open science supports the concept of Open Access publishing of research articles and there is extensive literature in relation to OA publishing. However, open science is not limited to OA publishing, but extends this open practice to publishing datasets, workflow, methods, details of ongoing research processes and so on (Grand et al. 2010; De Roure et al. 2010). Internet and new media have opened up opportunities for this contemporary form of open science (Scheliga and Friesike 2014). Having reviewed the relevant literature, open science is defined as the practices that allow cost-free open access to academic research. Academic research is a broad term including the whole research process from idea collection, primary data production, to publication of finished articles.

Next, three various aspects of open science practice are reviewed in more details. The first aspect refers to the open access movement for publishing, enabled by many open access journals and online repositories (Björk 2004). It is the most discussed mechanism for an open science 28

approach and is limited to free access to scholarly papers. This aspect of open science allows research articles to be accessed freely at no cost by other academics and members of the general public. The second aspect enables the exploration of how scientific knowledge is produced by making primary research data, metadata, methods and tools freely available on the internet (Lyon 2009). Data are essential to research within most academic communities, but were not traditionally shared systematically until recently with the emergence of the Open Data Movement (Nicholson and Bennett 2011). The second aspect of open science allows the sharing of primary research data at free cost that can be reused by other researchers or amateur scientists from the members of the public who have the ability of understanding those data. The third aspect refers to publishing ongoing research updates online enabled by social media services in a new media age. The last aspect of open science allows cost-free open access to research updates on social media sites and possibly before the research findings get officially published as journal articles. This aspect of open science is closer to Nielsen’s (2011) version as he advocates that scientific knowledge should be freely shared as early as is practical in the discovery process6. Tacke (2011) also suggests that open scientists inform about their scientific activity and invite participation of other professionals, amateurs and students to be part of the problem-solving process in constructing knowledge together. This type of open practice would rely on Web 2.0 and new media technology.

2.3.1.1 Open access to research articles Open access publishing generally refers to the free and unrestricted availability of scientific articles which can be accessed over the internet from anywhere in the world (Björk 2004; Pinfield 2005; Davis et al. 2008; Davis and Walters 2011). Academic journal articles and conference papers are still regarded as the most important information sources by UK researchers (Nicholas et al. 2010). Section 2.3.1.1 will review the rationales and purposes of open access (OA) publishing, business models, target audience, barriers, OA policies and discipline differences. Then it will critically review empirical studies about the citation impact of OA publishing.

Rationales and purposes of OA publishing The rationales of open access repositories are related to poor accessibility for publicly funded research to the public, publishing crisis of price rise and the issues of traditional publishing that it

6 Although peer review system of academic publishing has been criticised, peer review is still at the heart of all academic publishing. There are debates of whether it is appropriate to discuss about research through press release or online platforms before formal peer review process. The type of open science might be too radical for some scientists as they would question the reliability and credibility of non-peer reviewed materials. 29

can take a long period of time to publish findings which may delay the dissemination of new knowledge.

Over the last decade, academic publication in the UK has shifted from hard copy print to online journals, which has liberated researchers from the physical library and shifted the potential barrier to the access to those online journals (Nicholas et al. 2009). The traditional business model of publishing requires readers or institutions to submit a subscription fee which may prevent the public and researchers with limited resources from accessing the information. This has become a major issue in publicly funded research. Jim Gray gave a talk to the American National Research Council in January 2007, suggesting that publicly funded science literature should be online for free. The rationale is that a paywall leads to a situation where the public cannot use restricted reliable information in the form of peer reviewed academic research and instead may trust potentially unreliable information from the Internet, which Grey believes is an inefficient use of public money (Gray 2009). His view has been well supported by other scholars. There have been debates about whether lay audiences are able to understand scientific journal. This provokes discussions (that we return to below) of how science can be communicated to the public through other media such as science blogs.

The purposes of OA publishing are also related to improvement of accessibility, efficiency, publicity and recognition as well as science development especially in poorer countries. Both open access journals and self-archiving articles in a repository can lead to a wider dissemination of information to readers who are normally blocked from the subscription-based journals. It is notable that in the 2011 UK Public Attitudes to Science survey, more than half of the respondents (56%) felt that they were not informed about science, or scientific research and developments (IpsosMORI 2011). Open access to scientific publications can help the public access reliable information from research outputs. Masnick (2015) argues that open access to research articles might have prevented much of the Ebola outbreak in Liberia because Liberian doctors were not aware of the Ebola studies published by European researchers behind paywalls.

Another rationale of OA publishing is related to the publishing crisis and the criticism of publishers making profit from academic publishing. Academic libraries face a crisis that has continued for many years, as pointed out by McGuigan and Russel (2008). Academic authors and scholars usually provide the content as well as peer reviewing and editorial services to scholarly journals and they are usually unpaid by the journals. The universities purchase the journals and pay salaries to academic authors and editors whilst the publishers make profit on providing access to the journals. After the shift of paper format to electronic format, publishers of academic 30

journals started to offer site licenses for online access to academic institutions (Björk 2004). Since there is very little competition in academic publishing, the subscription price continues to rise. In protest against the high subscription costs of mainstream publishers, a growing number of researchers, institutions and funding bodies have started open access journals and online repositories which are free to all (Björk 2004).

One of the purposes of self-archiving to online repositories is the early and rapid dissemination of research findings. Open access increases visibility to a wider audience and improves research efficiency by avoiding duplication (Correia and Teixeira 2005). Some open access journals incorporate the open comment function as a new peer review model, which can help authors to take advice from other researchers and correct errors even after the article is published online (Gould 2009). Some open access journals that have adopted the open comment function, such as Shakespear Quartley, have asked reviewers to register so that authors know the status and reputation of their reviewers (Hahn et al. 2011).

Willinsky (2010) argues that open access repositories appear to co-exist well with independent publishing and commercial publishing, without disturbing current economic models or practices. Repository services also benefit the research communities by supporting digital content preservation activities, including not only scientific papers, but also data sets, multimedia data, tools and technical support (Park and Shim 2011). Self-archiving initiatives benefit researchers specifically from developing countries or poorly resourced organisations and improve their visibility by enabling them to distribute their local research to a world-wide audience. This bridges the knowledge gap between developing and developed countries without the bias from traditional journals which may prioritise more well-known scholars or those from more established institutions (Chan and Kirsop 2001; Chan and Costa 2005). In this way, disadvantaged academic communities can now take advantage of open access scientific knowledge which they might not have been able to afford to access in the past. Thus, e-print archives help researchers from developing countries enjoy global participation without further delay (Chan and Kirsop 2001).

Another purpose of OA publishing is to increase the visibility and readership of research papers. Davis et al (2008) found that open access articles seemed to be downloaded more often in the short term. The Research Information Network (2014) analysed articles in Nature Communications and found that open access articles were viewed three times more often than articles that were only available to subscribers. OA may also improve the impact of the papers and enhances the recognition of the authors. Academic recognition is closely associated with the citations impact of a work. Several studies on different disciplines showed evidence of more 31

citations and wider readership for open access articles (Pinfield 2005; Gould 2009). Citation impact will be discussed in more detail in later section of this chapter.

Gold/ Green OA and their business models The two major approaches of OA publishing are commonly referred to as Gold OA and Green OA. Gold OA refers to online journal articles which are either totally or to some extent made freely accessible to the public by the publishers (Laakso et al. 2011). The earliest open access journals were founded in the early 1990s (Björk 2004). Björk et al. (2010:7) studied scholarly articles that were published in 2008 and divided the Gold OA journals into three categories: Direct OA, Delayed OA and Hybrid OA. Direct OA refers to journals that are published as open access without any limitations and these accounted for 62% of all Gold OA in the sample. Delayed OA journals (14% of all Gold OA) preserve the most recent documents available only to paying subscribers and after a delay of certain period, the embargo is lifted and the content are free to all. Hybrid OA (24% of all Gold OA) gives authors the choice to pay for their articles to be made freely accessible within an otherwise subscription-based online journal (Björk et al. 2010; Laakso et al. 2011).

Green OA refers to the self-archiving of an author’s work (Crow 2008). These open access contents are supplied by the authors on a web site that is freely available without publisher mediation (Pinfield 2003). The open access channels for self-archiving include subject-based repositories, institutional repositories, social media sites and authors’ own homepages. This can include the university’s archive where the academic works. Björk et al. (2010) found that approximately 11.9% of all scholarly papers published in 2008 were freely accessible through some form of Green OA. It is estimated that 43% of all Green OA articles are published in a subject-based repository, 24% in an institutional repository and 33% on other websites. Most early OA journals were started by pioneer researchers and institutions; hence the major business model has been relying on free institutional resources and academics volunteering for the peer- review process (Björk 2004). These open access journals often use open source tools such as the Open Journal System, a low cost journal software platform (Edgar and Willinsky 2010). A survey of 998 scholarly journals that were built upon the Open Journal System found that the majority of these open access journals rely on subsidies, such as grants from institutions, the government or a funding body (Edgar and Willinsky 2010).

In more recent years, a new business model of open access journals has emerged requiring author charges, which, for instance, was adopted by the BioMed Central journals in 2000 (Björk 2004; Correia and Teixeira 2005). In this model, authors or their institutions pay a fee to have an 32

article published after it is accepted following the peer-review process. This fee is also called the Article Processing Charge (APC). In 2008, after BioMed was bought by Springer, authors were given the choice of paying up to USD 3,000 to allow their specific articles to be open access published in an otherwise subscription-based journal (Willinsky 2010). North American universities started to establish author funds for Gold OA article charges (Brown 2009). From April 2013, RCUK’s Open Access policy arranged to pay block grants to universities and other institutions for setting up publication funds for APC if RCUK funded researchers choose to publish in Gold OA journals (RCUK 2013). However for disadvantaged institutions and non-funded researchers, this model brings financial challenges and perhaps heralds a two-tier information system.

Subject-based repositories are usually associated with pre-existing research communities whose members talk with each other on a frequent basis, and are often used for conference proceedings and pre- and post-conference communication (Björk 2004). The systems, such as arXiv server, have been made by academics in Physical and Mathematical Sciences and are based on low running costs. They cannot charge subscription fees or author fees; however, they still cost money to run (Björk 2004). The institutional repositories typically use free open source software packages to run open archives, such as DSpace developed by MIT, and ePrints developed by the University of Southampton (Swan et al. 2005). By 2004, there were already over 100 universities worldwide who had Open Access Eprint Archives (Harnad et al. 2004). The institutional repositories will depend on the universities’ policies and decisions regarding the future roles of their libraries and publishing departments (Björk 2004).

Target audience of OA publishing As discussed earlier, communicating academic research can inform both scientists and public members of recent discoveries and advances. Thus the target audiences of OA publishing could be academics and members of the public who have interests and concerns about certain areas in academic research. Academics based in prestigious institutions may already have the access to many online subscriptions of scholarly journals. The vast majority of general public and disadvantaged researchers who are blocked out by the pay wall of conventional subscription- based journals can now access academic articles that are published in OA journals or deposited on OA repositories.

However, since the early 20th century, the idea has been established that science is too difficult and complex for the lay public audience to understand (Bucchi 2004). Scholarly articles are not usually written in lay language and thus might be misunderstood by non-specialist 33

audience. Hence the target audience of OA publishing are mainly academics and the members of the public who have scientific literacy and ability to understand academic research articles. The public’s ability to understand academic can vary for different disciplines. In some disciplines in Humanities and Social Sciences, the research outputs are often targeted to public audience as to influence public policies, improve participation or improve public well-being.

The assumption of public’s low level of scientific understanding resulted in translation and mediation of scientific knowledge (Shanahan 2011). The emergence of social media tools has offered potential for translation and mediation of academic research, which will be discussed in a later section.

Barriers to OA publishing Open access journals and self-archiving repositories face a number of barriers, including financial difficulties, quality control, copyright issues, referencing standards and OA journal’s lack of impact. Other barriers include individuals’ reluctant attitudes and resistance of OA publishing and a lack of knowledge of OA publishing and the existence of OA repositories.

Open access journals which depend on voluntary editing work and small funds require long term financial solutions. The author-pays model which requires on average 3,000 USD per article potentially discriminates against academics without a sufficient research grant, such as self- funded PhD students and those from less advantaged backgrounds. Swan and Brown (2004) found that only 4% of the authors of open access journal articles in their survey sample paid the author fee themselves, and over half of the respondents had not paid to have their paper published. There is also issue with the hybrid model of OA publishing which authors are given the choice to pay for open access for their articles to be published in an otherwise subscription-based online journal. From October 2009 to January 2012, among major academic publishers, the number of journals offering hybrid OA option has more than doubled, from around 2,000 to over 4,400, while the majority of authors did not pay the charges of 3,000 USD APC price to make their articles open access (Björk 2012). Many authors and institutions are facing difficulties of paying the expensive author fee and subscription fee, whilst the publishers are criticised of ‘double dipping’ that they make extra profits by charging both the subscribed libraries and the authors who are willing to pay for the APC (Cressey 2009). Academic institutions not only need to pay for the journal subscription fee and article processing charge if they wish to access the journals and comply to RCUK’s OA policy, but they also need to pay salaries to those academics who edit journals or peer review articles voluntarily for the journals. For academics based in Humanities 34

and Social Sciences who publish monographs, a special challenge arises since the business model of OA publishing are based around open access journal articles or preprints.

The author-pays model can also lead to quality control issues and poor or fake peer review process with open access journals. Bohannon (2013) questioned the peer review process of open access journals with his experiments of sending a flawed paper to 204 open access journals. More than half of those journals, including those major publishers Sage, Elsevier and Wolters Kluwer accepted the paper, failing to notice its shortcomings, which suggests serious quality control issues. However, Bohannon’s experiment was criticised as lacking a control group. Because he failed to submit his study to any traditional non-OA journal, his experiments results were unable to confirm that open access journals had poorer peer review quality controls than traditional journals (The Daily Texan 2013).

Institutional repository models have few or no financial barriers for authors. However, authors can face legal restrictions, as the publishers of traditional journals usually own the copyrights to the articles, so authors may be restricted to self-archiving their papers. It was reported that Elsevier, one of the leading academic publishers, sent thousands of takedown requests to Academia.edu users who self-archived their research articles which were published by Elsevier’s journals (Howard 2013). Some of the major publishers, however, are becoming more flexible and open about this issue (Lucas and Willinsky 2009) and many allow authors to deposit e-prints in an online repository (Pinfield 2005). According to the SHERPA/RoMEO7 database of academic publisher archiving policies published on 12 February 2015, out of a sample of 1,796 international publishers, 69% allow authors to self-archive the published version of their paper and an additional 7% permit posting of the pre-peer review version, while 24% of publishers fail to support the self-archiving of journal articles.

By 2010, over 1,200 online repositories were publically accessible world-wide, which potentially allows free access to at least half of current academic literature; however, authors have not participated in self-archiving in large numbers (Willinsky 2010). An international survey of 3,787 authors from 97 countries more than a decade ago in 2004 found that only around one fifth (21%) of respondents said that they had self-archived research papers in an institutional

7 SHERPA are based at the Centre for Research Communications, University of Nottingham and maintain on behalf of the open access community a portfolio of services: RoMEO, JULIET and OpenDOAR. http://www.sherpa.ac.uk/romeo/statistics.php (accessed 12 February 2015)

repository and 15% said that they had not practised this, and had no further intention to do so (Rowlands et al. 2004). Another survey study found that most authors in their sample failed to negotiate rights with publishers before signing the copyright transfer agreements (Hahn et al. 2011). Confusion over copyright issues and uncertainty about publishers’ self-archiving policies are among the barriers of Green OA found in a mixed-methods study of 17 Carnegie doctorate universities with institutional repositories in the United States (Kim 2010).

Back in 2006 many academics seemed to be unaware of the concept of open access or chose to remain ignorant of its implications in spite of having heard of the term (Swan 2006). A number of studies of institutional repositories across the United States, West Europe and Australia suggest that academics had little awareness of opportunities for self-archiving in digital repositories, and that a large percentage of documents on those repositories were deposited by a librarian or administrative staff (Xia 2007; Xia and Sun 2007; Kim 2011). Many scholars may be reluctant to spend the additional time and effort necessary for self-archiving practice (Kim 2010). Others are reluctant to post drafts in a repository because they are concerned whether readers are capable of accurately citing the final published article (Willinsky 2010).

In the case of open access journals, a large number of researchers, especially those who have not published in this way previously, are concerned that OA journals have not enough impact or are not as highly ranked as traditional journals (Swan and Brown 2004). An international survey of senior researchers found that authors’ major concerns regarding which journals to publish in are highly associated with the ‘reputation of the journal’, its ‘readership’ and its ‘impact factor’. In contrast, this study found the level of open access to a journal and the ability to deposit preprint or post-print in a repository is considered to be of relatively low importance (Rowlands and Nicholas 2006). Of course many of these studies were conducted more than a decade ago. Moreover, many OA journals were only recently established and hence many have not yet obtained the levels of impact factor or reputation sought by authors.

OA policies OA policies have been discussed and adopted to certain extent by various academic researchers, funders and institutions in western countries. According to Pinfield (2005: 30), the best and quickest way to accomplish major improvements in scholarly communication in the short and medium term is to ‘make it mandatory to deposit research papers in open access institutional repositories’. This view has been echoed by other commentators such as Swan et al. (2005) and Gargouri et al. (2010). Swan and Brown (2005) conducted survey questionnaires with researchers across various nations and disciplines and found that the vast majority of respondents (81%) 36

would willingly make their work publicly accessible by self-archiving their articles in an open access repository if required to do so.

The University of Southampton’s School of Electronics and Computer Science (ECS) was the first to implement an official mandatory self-archiving policy in 2002. This policy was effective in achieving a considerable increase of deposits in the school repository (Gargouri et al. 2010). The Queensland University of Technology in Australia became the first university in the world to require deposits from a faculty in an institutional repository. Although this mandatory policy was passed by the administration in September 2003, in practice, it took some time to implement and the faculty encountered difficulties due to continued concerns from academics (Crow 2008). Moreover, publishers may ‘see mandated archiving as an incursion on the exclusive property rights they have secured from authors (in exchange for publication)’. This situation can lead to a decrease in the author’s archiving rights to the post peer-review draft, while also imposing an embargo period that prohibits archiving until 12-24 months after original publication (Willinsky 2010:13).

More recently in the UK, major funders have started requirements for funded projects to be published with open access. On 1 April 2013, RCUK’s Open Access Policy came into effect which required RCUK funded research outputs to be published either through Gold or Green OA channels (RCUK 2013). These research outputs include any research articles and conference proceedings. RCUK further announced to pay block grants to universities and other eligible research organisations for supporting article processing charges. Other funding bodies such as charities also started to adopt the open access model. For example, an alliance of leading UK medical research charities launched a new fund to help make charitably-funded research freely available as soon as it is published (WellcomeTrust 2014). In July 2014, the Higher Education Funding Council for England (HEFCE) introduced OA policy for research outputs accepted for publication after 1 April 2016 in relation to the research assessments after the 2014 REF. Since these are all very recent policies, the effects of these policies on academics’ publishing practices are not yet reported in the literature.

Disciplinary differences The discipline differences in OA publishing are related to the modes of work and types of publication among different disciplines. As discussed earlier, academics in Humanities and Social Sciences do not produce as many journal articles as Medical and Natural Sciences. Monographs can be essential for the Arts and Humanities research community in terms of career progression. This difference would have influenced the development of OA publishing. 37

Open access journals have been well-developed in Medical and Life Sciences comparing to other discipline areas (Björk et al. 2010). In Humanities, however, there were lower levels of availability for OA journal (Darley et al. 2014). Compared to the availability of OA journals, open access monograph is far less common and is in great needs of a sustainable business model to get through the monograph publishing crisis (Look and Pinter 2010). In the Netherlands, a pilot project OAPEN-NL explored open access monograph publishing and found that the average cost of creating a monograph were a little more than 12.000 Euros (Kirchner 2013).

Open access repositories are more established in Natural Sciences. In Physical and Mathematical Sciences, a subject-based repository called arXiv8, is frequently used for depositing research articles (Björk 2004). There are a number of established subject-based repositories in Social Sciences, such as the Social Science Research Network (SSRN) and Social Science Open Access Repository (SSOAR). Open access repositories in Humanities are less established or ran at a small scale by individual libraries or departments.

Citation impact of OA articles Open access journals emerged in the early 1990s, long after the dominant model of journal evaluation (journal impact factor) was born. Thus little is known whether this model is able to evaluate OA practices. Many OA journals have only been in existence for a few years and some only publish a low number of articles per year. Therefore, many OA journals have not been indexed or have received no rank from JCR and hence face difficulties in attracting authors who presumably favour publishing in more highly-ranked journals (Alexandrov 2011). Self-archiving is seen as neutral or positive in terms of career advancement, as it does not conflict with researchers’ choice of which journal to publish in. According to a study of self-archiving practices by Kim (2010), many academics regarded self-archiving positively as it raised the profile of their research, influenced name recognition and enhanced their reputation.

In recent years, an increasing amount of research has been carried out into the effects of OA publishing on scholarly communication. One of the most frequently asked questions is, ‘does open access practice have a citation advantage?’ Since citation impact is so closely related to the academic reward system and career advancement, scholars are often concerned about whether increased free access to scientific articles has a citation benefit. The first study that suggested an OA citation advantage was conducted by Lawrence (2001), who compared conference

8 arXiv.org started in 14 August 1991 as [email protected]. See http://arxiv.org/abs/1108.2700 (accessed 18 May 2015)

proceedings articles listed in the Digital Bibliographic Library Browser (DBLP) Computer Science Bibliography and found that the availability of an online full-text paper has a positive effect on the number of citations received. Many studies on the same topic emerged afterwards and the methods for conducting these studies have been developed alongside the new OA models (Craig et al. 2007). These studies contained contradictory findings with some supporting the hypothesis of open access citation advantage, and others challenging or refuting it. The association between the two forms of OA publishing and citation impact will be discussed next with critical evaluation of relevant studies.

Citation impact of Green OA The citation advantage of Green OA remains controversial in the existing literature. Earlier studies of citation impact have mostly focused on the effect of Green OA and generally indicated that self-archiving of scientific publications can lead to increases in the number of article citations. Some studies investigated the effects of depositing in a subject-specific repository, such as arXiv (Schwarz and Kennicutt 2004; Metcalfe 2005; Metcalfe 2006). The methodology of these studies typically uses citation counts for materials from certain journals deposited in arXiv, termed as ‘open access’, to compare with those for all other articles within the same journal and year with ‘toll access’. However, this method misses articles that are deposited in other repositories or on personal websites.

A number of studies of Green OA define open access as any version of publication freely available on the web, in the form of self-archiving documents in repositories or other web sites (Antelman 2004; Harnad et al. 2004; Hajjem et al. 2005; Norris et al. 2008). Harnad and colleagues used computer algorithms to identify self-archiving OA versions, excluding articles in OA journals, from millions of scholarly articles across various disciplines and over a long period of time (Harnad et al. 2004; Hajjem et al. 2005). Others chose specific subject areas and selected a relatively small quantity of articles, then manually identified open access version papers and their respective citation counts (Antelman 2004; Norris et al. 2008). Antelman (2004) selected a sample of four disciplines, namely philosophy, political science, electrical and electronic science and mathematics, and manually searched article titles from the sample (n=2,017) in Google to identify OA articles. Norris et al. (2008) used four search engines – OAIster, Open-DOAR, Google Scholar and Google – to determine the open access status of articles (n=4,633) in non-OA journals in the subjects of ecology, applied mathematics, sociology and economics.

These earlier studies of Green OA mostly used basic citation comparisons without controls and all of them found a positive OA effect on citation increase. However, they suffer from some 39

methodological problems; as Craig et al (2007) pointed out, the authors failed to control for confounding factors, such as article and author characteristics. Hence the OA advantage could be due to the quality of article or reputation of the authors rather than Green OA status. Moreover, these studies failed to provide evidence for a causal relationship other than correlation.

Later studies of Green OA employed different methodologies and suggested that early view effects and self-selection of posting high quality work are responsible for the citation effect rather than OA status (Kurtz et al. 2005; Henneken et al. 2006; Davis and Fromerth 2007; Moed 2007). Kurtz et al. (2005) were among the first to investigate these possible explanations for citation effect. After controlling for the early access probability (preprints appearing sooner than formal publication), the result showed that there is no general OA status effect. However, the early view effect is criticised as not significant because posting of preprints are only popular with some disciplines, such as physics and economics, while most authors appear to favour posting a published version of their work (Antelman 2006; Xia et al. 2011).

To investigate the selection bias, Kurtz et al. (2005) compared citations of self-archiving articles deposited in arXiv with those non-deposited articles in the same journal (APJ) and in the same year (2003). The results of this study indicated that the citation counts of the deposited articles were two times more than non-deposited ones. This finding suggests a selection bias, in that a quality differential between self-archiving articles and non-deposited ones could be responsible for the citation effect. The effect of selection bias was confirmed by follow-up studies using more advanced statistical analysis methods (Henneken et al. 2006; Davis and Fromerth 2007; Moed 2007). However, other studies have provided opposing evidence to challenge the finding that selection bias is responsible for OA citation effect (Metcalfe 2006; Gargouri et al. 2010). Gargouri et al. (2010) compared self-selective self-archiving with mandatory self-archiving articles in non- OA journals using ratio analysis and logistic regression with controls, and found a similar pattern of OA citation advantage for mandatory self-archiving and for self-selected self-archiving. Thus Gargouri et al. (2010:11) concluded that the OA advantage is ‘real, independent and causal’.

A more recent study by Xia et al. (2011) of multiple open access availability in non-OA journals and its impact on citation deployed a different method from other similar studies. This study counted the number of Green OA availability by searching the article in three web search engines (Google, Google Scholar and Yahoo) and used the OA availability (number of 1, 2, or 3 search engines that can find an online free full-text) as the explanatory variable, instead of binary OA status (yes or no to a free text) as in previous studies. Multiple linear regression analysis (n=486) with controls for number of authors, references and pages, confirmed a citation advantage of the 40

easier availability of Green OA publications. While the findings of Xia et al. appear to be fairly robust, the sample for the regression analysis is rather small and it is questionable to include the Yahoo search engine in the study given the dominant status of the Google search engine. The extent to which the findings of this study are applicable in countries with a different search engine culture is also uncertain.

Citation impact of Gold OA Studies of Gold OA have provided evidence either to support or oppose OA citation advantage. Eysenbach (2006) was the first to conduct a longitudinal study of a cohort of OA and non-OA articles from the same journal (Proceedings of the National Academy of Sciences [PNAS]) in the latter half of 2004 to investigate the Gold OA effect on citation. This approach eliminated the possibility of an early view effect as both OA and non-OA articles in the same edition were published at the same time. Authors of PNAS had the choice of paying to make their work open access from the start. All the non-OA articles also became freely available to non-subscribers six months after publication. Logistic regression analysis was conducted after controlling for number of authors, days since publication, funding, subject area and author status and confirmed citation advantage. Eysenbach (2006) argued that the advantage of OA status may be even greater in fields where articles from the control group remain toll-access six months after publication. The selection bias effect, however, was not explored in this study.

Another study of eleven biological and medical journals from 2003-2007 that employed author-choice OA models found a small citation advantage (17%) for OA articles in all journals using linear regression models. Only two of the eleven journals showed positive and significant OA effects after controlling for other explanatory predictors of citations such as, number of authors per paper and the length of paper (Davis 2009). The potential causes of the citation impact were discussed but not tested in the analysis. Gaule and Maystre (2011) carried out a similar study to Eysenbach (2006) which analysed citations of author-pays articles in PNAS from 2004-2006 using linear regression models. According to Gaule and Maystre (2011), the OA citation effect becomes non-significant after taking into account a full set of control variables which may be explained by selection bias.

With most previous studies being observational, a novel research design of a randomised controlled trial was proposed by Davis and Walters (2011). The authors obtained permission from a professional association to conduct several controlled experiments by randomly assigning OA status to articles in various online scholarly journals (Davis and Walters 2011). This approach allows investigators to control for potential differences at the start of the experiment and 41

eliminate the influence of the early view effect and selection bias. Studies of eleven journals published by the American Physiological Society found that OA availability had led to more downloads, but not more citations (Davis et al. 2008; Davis 2010). A similar study of 36 journals across the Sciences, Social Sciences, and Humanities, with randomly assigned OA method, confirmed no citation impact (Davis 2011).

Davis et al. (2008) calculated the citation counts of articles published in journals in their first year of publication. Such journals are considered not to have had sufficient time for high citation counts to accrue (Xia et al. 2011), since the peak in citation frequency for scholarly articles usually occurs around the third year after publication (Harnad and Brody 2004; Norris et al. 2008). This factor was taken into account by Davis (2010), whose study used the same dataset as Davis et al. (2008), but studied citation within the first three years and still found no citation advantage.

Although Davis and colleagues’ research design seems to be advanced compared to previous studies, their work also suffers from limitations. In their study, the majority of non-OA articles became freely available to non-subscribers six to twelve months after publication; therefore there is considered to be no difference between subscription and OA publication after that period. The lack of citation impact in Davis and Colleagues’ study can be due to the delayed access of only six months to one year not being long enough to differentiate OA and non-OA to those scientific materials. In other words, writing up and redrafting a scientific paper can take more than a year, while toll-access has already been removed after the original publication date, therefore the OA effect no longer exists in this type of journal.

Calver and Bradley (2010) investigated the citation effect of both open access journal articles and self-archiving documents in repositories or other websites by comparing the citations of both Gold OA and Green OA articles in biology journals and books using linear regression. They found that neither Gold OA nor Green OA status had a significant citation effect having taking account of the type of paper, its length, the number of authors, the authors’ citation profiles, and whether the author or publisher released the paper in OA. Nevertheless, author-archived OA book chapters accumulated up to eight times more citations than non-OA chapters in the same book.

In short, various studies of Green OA and Gold OA have provided evidence to support or oppose OA citation advantage. Some studies suggested that author self-selection of high quality work or the early view effect was the cause of the citation effect rather than OA status. However, the benefits of OA publishing which are related to the increase of visibility and readership of OA articles are generally acknowledged in the literature. The OA citation effect remains controversial 42

and can vary by subject matter. Further research on this subject is needed with a more carefully designed methodology and a more advanced statistical analysis.

Section 2.3.1.2 will discuss the second aspect of open science.

2.3.1.2 Open access to research data Borgman (2007: 115) regards data as the foundation of scholarship – ‘data are outputs of research, inputs to scholarly publication, and inputs to subsequent research and learning.’ Scientific data is referred to as a resource and is now largely produced in digital form and consequently storage (European Commission 2013). Compared to the standardization of research publication as in article format, the format of research data vary largely between different disciplines across Sciences and Humanities.

In this study, open access to data refers to practices that allow cost-free access to primary research data that can be reused by others. Some of these cost-free accesses may be limited access, which requires users to register an account or provide their institution’s login information, while others are publicly accessible. Section 2.3.1.2 will review rationales and purposes of sharing data, target audience, barriers, data policies, discipline difference and strategies for future practices.

Rationales and purposes of sharing data The rationales and purposes of open access to research data are related to the accessibility to public goods, validation of findings, reducing duplicate data collection and accelerating scientific progress (Piwowar et al. 2007; RIN 2011a).

Data sharing policies were brought to the fore in the Bermuda Principles in 1996 when the genomics community argued for free access to the genomic sequencing as soon as it became available in order to encourage further research and development, and to maximise its benefit to society (Rodriguez et al. 2009). The rationale behind this is that knowledge and creativity can be furthered by access to openly available data, so that data produced in projects by public funding should be freely available in the research community and to the public (Kaye et al. 2009). This is similar to the rationale of OA publishing that publicly funded research data is a public good and thus should be made available on the World Wide Web. The field of genomics is the first subject area to adopt the data-sharing policy and develop infrastructure and resources to support data sharing. 43

The Berlin Declaration on Open Access to Knowledge in the Science and Humanities (2003) supported an open access paradigm and called for the setting up of regulation to promote open access to scientific contributions, including ‘original scientific research results, raw data and metadata, source materials, digital representations of pictorial and graphical materials and scholarly multimedia material’. They called for free access to a complete version of research findings and all supplemental materials. The Declaration was echoed by scientific communities in North America and Western Europe. The policies and practices of data sharing started to emerge in various disciplines, including in the Medical Sciences, Environmental Sciences, Physical Sciences and Social Sciences. Open access data centres have been set up to support the curation of research data by stakeholders. For example, the Research Collaboratory for Structural Bioinformatics (RCSB) Protein Data Bank have scientists across the world collecting data on protein structures before submitting to the data bank, which is often required by funding agencies. The Protein Data Bank had over 30 US-based full-time staff to review and validate the submitted data, and then made it freely available online to the public (Maron and Smith 2008).

Sharing research data can validate research findings, reduce duplicate data collection, accelerate scientific progress through re-analysing existing datasets, increase visibility and citation, and provide financial benefits (Piwowar et al. 2007; RIN 2011a). Researchers may use existing results as a context for their own data and to test the replicability of certain results as a means of quality assurance. New analysis can also be performed on existing data to undertake new and original research (Boulton et al. 2011). Re-analysis and meta-analyses can build on the value of the research of the original data collectors and lead to both current and new hypotheses being tested on new data, as well as new analytic methods or technology being developed and validated (Gardner et al. 2003), especially when combined with other studies with publicly accessible data sets (Piwowar 2010). Open access to research data also benefits educators and employers by providing resources to train new researchers on the exposure of real-world data as well as supporting commercial innovation outside academia.

Similarly as the purpose of OA publishing, sharing research data may increase the visibility and citation for the related research papers. For individual researchers, sharing data may increase online visibility and provide opportunities for collaboration (Piwowar et al. 2007; 2008). Piwowar and colleagues found that sharing biomedical research data on a publicly accessible website was significantly associated with an increase in citations for the related publications, independently of journal impact factor, date of publication, and author based in the United States. Belter (2014) 44

also found that in oceanography, data sets are more highly cited than most journal articles published during the same years as the data sets.

Piwowar et al. (2011) found potential financial benefits of open access to data – they estimated an annual cost of around USD 400,000 for data preservation in Dryad, an international open data repository for the biological sciences. According to an analysis of the reuse of data deposited in 2007 in the Gene Expression Omnibus (GEO), a database at the US National Centre for Biotechnology Information, third-parties contributed to 1,150 published articles by the end of 2010 by using the 2,711 data sets in the GEO, and reuse continued to accumulate rapidly. Similarly, it is estimated that an investment of USD 400,000 on Dryad in one year will contribute to over 1,000 papers in the next four years, which is far more than the accepted value for the research fund.

Target audience of OA to research data The target audience of open access to research data are mainly scientists and specialists who have scientific literacy of understanding and reusing research data in the specific fields. Although the rationale of Open Data Movement is that data produced by public funding should be freely available to the public, in reality, it requires great scientific literacy to understand raw data, metadata and other specialist source materials.

Barriers to sharing data Research by Nelson (2009) suggests that very few academics share their data with other scientists or the public more widely. Hannay (2009) suggests that the barriers to the full-scale adoption of data sharing are not only technical (as infrastructure needs to be built to store the data), but also psychological and social.

One of the major barriers of data sharing is the lack of recognition incentive for sharing research data (Piwowar et al. 2008). Sharing data, methods or tools, brings few rewards to the original data creators or gatherers; indeed such generosity could allow credit to be inappropriately allocated to re-users, which is a clear disincentive to share (RIN 2011a). An anecdotal example of this is provided by Cragin et al. (2010), who report that a re-user of data from a geological study published his findings first while the original author was still working on his analysis. In that case, the original researcher who collected the data lost his ‘priority’ in that specific scientific discovery. Junior academics, who have not yet secured a professional position or reputation, particularly fear having data or findings, which can generate multiple reports over time, taken by others, as the academic reward system is associated with publication and its 45

impact (Gardner et al. 2003). As such academics can be reluctant to share their data—unenforced mandates are often ignored as data sharing is not regarded as a part of academic evaluation and there are no penalties for noncompliance (Tucker 2009).

Another barrier is the technical challenge. Public repositories still differ in the formats and kinds of data deposited. Researchers may have to learn entirely different systems in order to access information in different repositories. Central repositories also need quality metrics for assessing data quality, and enabling it to coordinate with researchers and other repositories (Rodriguez et al. 2009). A large scale dataset may require cataloguing or indexing which can be a technical challenge for researchers (Gardner et al. 2003). Scientists have reported difficulties of sharing their data, such as the datasets being too large to store in the file-sharing service provided by their universities, and local technical infrastructures that do not support the format of their datasets (Cragin et al. 2010). The lack of standard for data formats may prevent data from being reused. Even when the data is publicly accessible, without standards for data formats, descriptive information and measurement scales, data may not be usable by others, or may be misinterpreted which could result in the publication of unwanted results that can challenge the credibility of the original work and harm the reputation of the original investigators (Gardner et al. 2003). In a study of information practices in various physical sciences, little was agreed on how to cite databases, or allocate credit to researchers and technicians who generated and maintained the database (Meyer et al. 2011).

Privacy issues and a lack of trust in other users are also among the barriers. For datasets collected with human participants, it is important to protect personal information and avoid individuals being identified. For example, for clinical data sharing, a central challenge in research practice is ‘how to put a human face on the data while still protecting individual privacy’ (Anderson and Edwards 2010: 16). As misuse of research data has been reported, many academics may find it hard to trust potential users. In a number of cases publicly available scientific data has been misinterpreted by industry groups in order to promote their products (Cragin et al. 2010). A lack of standards is also one of the major impediments to the sharing and re-use of scientific data (Zimmerman 2008). Other potential barriers can be lack of time including time to format, document and maintain the data, and time to deal with requests for information, confusion in where to archive the data, fear of being challenged and errors identified, and objections from industrial or other sponsors to the release of data and other information (Piwowar et al. 2007; RIN 2008).

Data policies 46

It is notable that although the principles for open data sharing among the genomics community were agreed in Bermuda in 1996, in practice, it was difficult to convince researchers to follow the principles. Data policies are more established in some subject areas than others, such as in Biomedical Sciences. Many journals in this area have started to request the sharing of primary datasets from authors. In a study of journal policies and data sharing, Piwowar and Chapman (2008: 2) studied biomedical journals (n=70) that published articles on ‘gene expression profiling’ in 2006, and analysed journal data sharing policies applicable to microarrays. They found that academic publishers are more likely to have stronger data sharing policies than commercial publishers, while all four of the open access journals in the sample had a data sharing policy. Policy strength was associated with impact factor – journals with a higher impact factor were more likely to have stronger data sharing policy. Policy strength was also positively correlated with measured data-sharing submission into the GEO database, in that the journals with a stronger policy had higher median data-sharing prevalence than journals with no data-sharing policy. However, data-sharing prevalence in general was quite low, even for journals with very strict sharing requirements (29%).

In the United States, the National Institutes of Health (NIH) have a data-sharing policy; however, being subject to the NIH data-sharing mandate requirement are not necessarily more likely to share their own datasets. Piwowar and Chapman (2010) analysed 397 biomedical microarray articles and found that researchers were more likely to practise data sharing when their work was published in a high-impact journal and when the first or last authors were highly experienced in their careers with a high level of professional impact. In another study, Piwowar (2011) used similar methods and found that authors were more likely to share data if related articles were published in an open access journal or a journal with a stronger data-sharing policy, if they had prior experience sharing or reusing data, or if their research received a large amount of NIH funding. However, the worrying result is that the data relating to studies on human subjects and cancer were least available while these areas could make an important impact in the field of study and bring about many health benefits.

In the field of Environmental Sciences, Weber et al (2010) analysed the policies of funding agencies, repositories and journals. Their findings suggest that very few stakeholders in this field had policies for data sharing, archiving or citation, including less than half of the funding agencies, and only around one in every eight journals. Also, instructions showing how to cite data sources were especially rare. 47

In the UK, the Research Information Network (RIN) conducted several studies on data sharing and the use of academic data centres with research councils and major funders as well as academics that used data centres. They found that five of the seven research councils had data sharing policies in place by 2008 (Sherpa/Juliet 20119; RIN 2008). By 2011, eight of the nine major UK research funders had a data policy, with seven of them preparing to fund data sharing activities within a project bid (RIN 2011a). Seven out of the nine funders require researchers to include a ‘data-sharing plan’ when applying for funding, while eight funders require maximum time limits for depositing data after the projects finish (RIN 2011a). These studies also found that many researchers, although registered users, do not submit their own new data to the data centres.

Disciplinary differences Disciplinary culture may influence the practice of sharing research data. Some disciplines may have a long history of a data sharing culture while others may not produce primary research data. In those disciplines with more established data policies such as Biomedical Sciences, Environmental Sciences, Physical Sciences and Social Sciences, researchers may have more resources and skill sets to store their primary research data (e.g. the availability of repositories and standardisation of datasets). In the Biomedical Sciences, journals often have data sharing policies especially those open access journals and hence authors are more likely to deposit research data when they publish articles to comply with the journals’ requirement (Piwowar 2011). In Humanities, some disciplines do not produce research data that can be reused by others as discussed earlier. However, a study found that a key challenge for Humanities scholars is their lack of ability with tools and methods to link data housed in different archives (RIN 2011b).

Ethical issues involve human subjects may affect academics’ willingness and ability to share research data. Disciplines in Medical and Social Sciences with human participants may be reluctant to share research data because of confidentiality or anonymity reasons (Anderson and Edwards 2010; Grand et al. 2014). Natural Sciences and Engineering have no human participants and thus would not face such issues.

Strategies for future practices A number of studies suggested strategies for sharing research data, including when and with whom to share research data, what not to share and what needs to be shared.

9 SHERPA/JULIET directory service available at http://www.sherpa.ac.uk/juliet/ (accessed 12 Jan 2012) 48

The 2008 International Summit on Proteomics Data Release and Sharing Policy in Amsterdam agreed on the principles of when to release data. The summit stated that data generated by individual researchers should ‘be released into the public domain at the latest upon publication while data generated by community resource projects should be released upon generation following appropriate QA/QC procedures’ (Rodriguez et al. 2009: 3689). In this way, sharing would be accomplished after the academic rewards had been secured by the researchers who collected the data.

In a study with 20 researchers, 60% (12/20) pointed out a need for restrictions on publically sharing some or all of their data for a certain length of time, and many of the participants agreed that distributing any data before publication should be restricted to trusted acquaintances, such as collaborators or close colleagues (Cragin et al. 2010). Another study with cancer researchers suggests that trust in the person or community being shared with is a specific condition to motivate data sharing (Tucker 2009). Some data centres require users to register with institutional or professional email addresses to enhance trustworthiness.

In relation to clinical data sharing, a strategy to build trust for inter-investigator data sharing is for the ‘original researcher [to] remain the primary data steward and gatekeeper for future use’ (Anderson and Edwards 2010: 17). In geology, a strategy is suggested to gain strict control of any data provided to anyone beyond the scientist’s immediate collaborators (Cragin et al. 2010). The current dependence on the de-identification processes should remain ‘a component of future technically leveraged data sharing processes’, and regulatory guidance is needed to support new models of governance that can ‘adapt and respond to the unforeseen data management challenges facing biomedicine’, and maximise the utility of clinically derived data (Anderson and Edwards 2010: 20). In other disciplines, safeguards should require that re-analysis of data ‘be limited to that which can be meaningfully derived, given restrictions, parameters, or boundaries inherent to the original hypotheses, protocols, and techniques for acquisition and processing’ (Gardner et al. 2003: 292).

It is also important to develop strategies to foster trust in using others’ shared data. Zimmerman (2008) interviewed 13 ecologists who successfully reused secondary data for their publication to investigate the role of standard methods in the data gathering process. They found that the ‘ability to comprehend data collected by others’ is the key to their reuse (Zimmerman 2008: 21). To achieve this ability, the essential condition is the public availability of the metadata. The accessibility to this additional information and details about methods and tools used, and how the data was generated can enhance trust in reusing others’ data. 49

In medical research such as genomic study, policy calls for public data release, but clinical data has often been collected without including data sharing information in the participants’ informed consent process (McGuire et al. 2008). McGuire and colleagues’ study used the results from focus group interviews with researchers to suggest that data sharing information should be included in the consent form for all genetic research and participants preferred multiple data sharing options. However, at present it is not possible for the data generators to fully inform participants, how their data will be shared and processed in the data collection stage, which is required under the principles of data protection (Zika et al. 2008). Therefore, new approaches to informing participants about the use of their personal information for various potential research purposes should be developed, especially in a context of worldwide data sharing (Kaye et al. 2009). These approaches should also apply to Social Sciences and other disciplines that involve human objects.

For sensitive information, such as DNA data, which can be used to identify individuals, only aggregated data should be shared on the Internet and more detailed data should be obtained only through request to the principle investigators (Kaye et al. 2009). Additionally, appropriate approaches are needed to make sure potential users are informed about the availability of data sharing, and the details of when, where and how the data has been selected, collected, and used, in order to ensure the maximum benefit of data sharing and reuse is gained (Gardner et al. 2003).

Many scholars have proposed to develop an incentive system and approach to the citing of data and databases. There are two reasons for doing so. First, without means to be allocated credit for their effort, scientists and research groups who are responsible for creating and maintaining the data have fewer incentives and motivation to conduct such work; and second, in some scientific disciplines, such as physical sciences, the version of the data used needs to be identified in order to replicate scientific work, in cases where databases continue to grow (Meyer et al. 2011).

All research institutions are advised to ‘develop and track metrics for data-sharing contributions as part of their academic research environment’ (Piwowar et al. 2008: 1315). Faculty leaders should encourage their staff to monitor the purposes for which their data are reused and adopt a data sharing citation index (ibid). According to Kaye et al. (2009), indices to measure the effort of data generators and their contribution to establishing a resource for other researchers must be included in national assessment schemes. Gardner et al. (2003) also suggests that the reluctance of sharing data might be minimised or even eliminated when a similar culture of rewards and safeguards for publication apply to data. A citation and credit paradigm must be applied in order to promote data sharing, that is, to require the acknowledgement of the sources 50

and collectors when using shared data. Another solution is enforcing rewards for data contribution. This approach was developed by the Genetic Association Information Network (GAIN), to give the data generators a certain time period to establish a publishing lead on their competitors, even though the data is open to all. However, this approach can put great pressure on the research team to rush the analysis process to meet publishing deadlines (Kaye et al. 2009).

Although researchers rarely volunteer to share their primary research data (Eysenbach and Sa 2001), they may be willing to do so if requested by a funding agency or journal policy, or if made aware of potential benefits such as increased citation impact. Therefore, stakeholders should do much more to encourage data sharing practices (Weber et al. 2010).

Section 2.3.1.3 will briefly introduce the third aspect of open science.

2.3.1.3 Publish ongoing research updates on social media New forms of scholarly communication not only allow academics to provide open access to research articles and data, but also informal communication and dissemination of ongoing research. The last aspect of open science allows cost-free open access to research updates on social media sites possibly before the research findings get officially published as journal articles. Grand et al. (2014) suggest that how and when information should be made open remains under discussion. A broad term for this online medium which provides a social function for interpersonal communication is called ‘social media’. Social media refers to a group of online applications ‘that build on the ideological and technological foundations of Web 2.0, and that allow the creation and exchange of User Generated Content’ (Kaplan and Haenlein 2010: 60). Web 2.0 is seen to offer a technical platform for its users to interact and collaborate with each other in a social media dialogue as creators of user-generated content in a virtual community, in contrast to websites where users are limited to the passive viewing of the content that was created for them (Thanuskodi 2011). Examples of Web 2.0 applications include social networking sites, blogs, micro-blogs, wikis, and photo and video sharing sites.

Internet and new media can be used for different purposes in research life. Social media can also be used to search for research information, promote publication and networking with peers, which will be discussed in more details in Section 2.3.2. In the literature, most studies in relation to communicating research on social media focus on science blogging.

Blogs are web pages whose content can be filled in by dated entries, usually displayed in a reverse chronological order; and often combining text and images. The pages also usually display links to other blogs or web pages relevant to the topics contained in them, and a comments 51

function allows interaction between the authors and readers. This function is usually monitored and can be altered by the blog host (Bukvova 2011). The configurations of blogs can range from individual blogs with a sole author to a variety of network structures, such as those hosted by a faculty or a research community over a broad topic, or a set of carefully selected blogs aggregated and presented in a single web space (Tatum and Jankowski 2010). Science blogs/research blogs publish blog posts related to science and review scientific developments (Shema et al. 2014). Scientific blogs can be found in popular academic journals, such as Nature and Science; from specialised web-portals such as scienceblogs.com or researchblogging.org (Kjellberg 2010). Section 2.3.1.3 will review the rationales and purposes of communicating research on research blogs, target audience, barriers, policies and discipline difference.

Rationales and purposes of publishing ongoing research updates The main rationales of publishing ongoing research updates are to increase the efficiency of scientific knowledge production and improve the understanding of science. Since the publication of research findings is the output of a black box process, in order to understand the process of scientific discoveries, it is necessary to open that black box. Blogging about ongoing research updates could bring transparency to research process. One example is the Open Notebook Science in Chemistry and Chemical Biology, whose participants use a web blog to record day-today laboratory work within which data can be linked and open to the public (RIN 2010). This involves real-time scholarly communication at all stages of the scientists’ work and also invite comments from readers of their blog. This approach was motivated to maximise the reuse, and analysis, of data, but the volumes of data that were generated exceeded the limit of capacity of the peer review system.

The purposes of publishing ongoing research updates include keeping a research diary, getting timely feedback and helping other researchers avoid same mistakes. Formal publication process can take a long time and feedback can only be gained from limited numbers of peer reviewers beyond immediate colleagues. By recording process at the lab on a daily basis and sharing methods, data and results online with free access, open scientists may gain timely feedback to improve their research. In this way, they open up the peer review process to a wider audience and reviewers early on in their research cycle. Meanwhile, other researchers are able to test and reproduce the same experiments (McDaniel 2013). Moreover, sharing failed experiment results helps other scientists avoid duplicate mistakes and thus improve efficiency.

Another rationale is similar to the rationale of OA publishing which is related to poor accessibility for publicly funded research to the public. Publishing ongoing research updates on 52

social media could provide opportunities for academics, policy makers, service users and members of the general public to access research contents without the paywall and participate in discussion.

Target audience Similarly as OA publishing, the target audiences of blogging about ongoing research could be academics and members of the public who have interests and concerns about certain areas in academic research. The target audience may need to have scientific literacy to understand certain scientific subjects and this requirement would vary in different disciplines.

As discussed earlier, the emergence of social media may offer potential for translation and mediation of complicated academic research (Shanahan 2011). Research blogs offers academics new opportunities to communicate their research in a lay language other than formal writing of journal articles. In order for the lay audience to have a better understanding of a research topic, it raises questions for academic writers of how to write research blogs and what kind of language to use. These questions also apply to those who blog about their ongoing research updates, especially in some disciplines in Humanities and Social Sciences, when the research outputs are often targeted to public audience to enhance their learning and well-being or to influence policy making. As one of the main purposes of publishing ongoing research is to getting feedback, it is not a one-way process as Hilgartner (1990) suggested, but rather a two-way communication, discussion and participation between academics, specialists and interested public members.

Barriers and policies Unlike OA publishing and sharing data, there is no formal policy on sharing ongoing research on the social media. Publishing ongoing research on social media is a novel practice. It is not a norm in academia and thus there is no formal incentive related to such practice. Moreover, blogging about ongoing research can challenge the academic reward system by leaking research results before formal publication (Cox and Forshaw 2011). Academics may be reluctant to share research updates online for fear of competitors knowing too much about their ongoing research. This raises the question of how to discuss new scientific findings that have not yet been peer reviewed in an online public space. Barriers of using social media for other purposes will be discussed later in Section 2.3.2.

Disciplinary differences Blogging about ongoing research has been reported and studied more often in Natural Sciences disciplines than in Humanities and Social Sciences. American astronomy professor David 53

Hogg reported on his daily research work on his blog on a regular five days per week basis (Hogg 2012). Another example is Canadian microbiologist Rosie Redfield’s blog that she described as an ‘open science’ research blog, in which she wrote her daily findings in the lab, as well as her comments on open science practice and scholarly communication (Redfield 2012). Visitors of Redfield’s blog can choose to leave an open comment with the blogger’s name or anonymously to interact with the author and other commentators.

In Humanities, McDaniel (2013) uses Gitit as an Open Notebook for History. Unlike in Natural Sciences when experiments can be reproduced, historians may use Open Notebook to develop specific historical arguments and communicate the general argument with each other. More studies will be required to explore how academics in Humanities and Social Sciences publish ongoing research updates online and what benefits and difficulties they have encountered.

Section 2.3.2 will discuss other kinds of use of social media tools in academic work.

2.3.2 Academic use of social media

2.3.2.1 Introduction Many new media applications have become popular among academic users in the recent years and were found to be effective information resources as well as dissemination channels (Gu and Widén-Wulff 2011). The current top two most popular social media tools with the largest number of monthly visitors are Facebook and Twitter (eBiZMBA 2014). Research blogs and other social networking sites, such as Academia.edu and ResearchGate, have also been commonly used by academics to communicate research in general (Kjellberg 2010; Thelwall and Kousha 2014a, 2014b). Thus research blogs, Twitter and social networking sites will be considered in detail below.

2.3.2.2 Research Blogs As discussed, some academics have used research blogs to publish their ongoing research updates. There are also other functions and purposes for use research blogs. Academic blog users may write blogs to communicate their research of published articles, share their ideas and interests on various topics or simply read blogs for learning or entertaining. Academic publishers such as the Nature Publishing Group (NPG) and Public Library of Science (PLoS) also started to write blogs to promote scholarly articles (Stewart et al. 2013).

Rationales and purposes of writing and reading research blogs The rationale of writing research blogs can be similar to the rationale of public science communication as discussed in Section 2.2.4 that academics in receipts of public funds have a 54

duty to explain their research to the general public (Wolfendale Committee 1995). This is related to the debates that lay public audience are not able to understand scientific journal articles or not have the access to traditional subscription based journals while science can be communicated to the public through other mediums such as blogs. Blogging provides a medium for scholars to present ideas in a far less formal setting than traditional publication channels, which has made blogging an informal method of scholarly communication with free and open access (Priem and Hemminger 2010). The accessibility of scholar contents on research blogs provides new ways of creating scholarly culture in the digital age, fostering wider participation and improving public engagement. It creates ‘an amiable digital ivory tower spearheaded by the open access movement, a movement that presents fresh opportunities to educate or to influence public participation’ (Virdi 2010). The general public are able to access scientific blogs without the paywall and may be able to comment and interact with the authors. Moreover, blogs are often written in a more informal language that is easier to understand than technical journal papers.

The purposes for those who write research blogs include functions of expressing opinions, disseminating scholarly material, gaining feedback, keeping abreast of new discoveries in the field, and interacting with others, as suggested by a study of 12 academic bloggers in Northern Europe (Kjellberg 2010). The participants in Kjellberg’s study were motivated by the opportunity to share and create an audience in order to build a reputation and feel connected with peers and the academic community. Another purpose of writing research blogs is to promote publication, increase readership and build personal reputation. Shema et al. (2014) reviewed journal articles blogged in researchblogging.org and found that blogged articles attracted more citations.

The purposes and advantages for those who read academic blogs include finding useful research information, discovering peers in other institutions and networking with colleagues and potential collaborators (McGrath 2010). Blogs not only benefit the search of specific scientific information, but also have social and economic benefits. Blogs provide benefits similar to an academic conference – academics can participate in blogging by giving a paper, receiving comments and responding to questions (Kosmopolit 2009). Blogs can also lead academics to other sources of information, through reading posts and talking to people who do similar research. Bloggers often become familiar with and communicate with one another. Therefore, blogging can become a useful networking tool for academics without the need to travel physically to a conference.

Barriers to blogging 55

One of the biggest barriers of writing research blogs may be the lack of formal incentives. It takes a great amount of time and effort to write, read and comment on others’ research blogs whilst the precise benefits of blogging are unknown or uncertain. Academics may not get paid to write a research blog and hence those efforts are often based on using their personal time. Potentially negative consequences associated with new forms of scholarly practice can be a concern for academics. A study of digital scholarly communication in North America found a case of rumours about young researchers being denied interviews or jobs when employers mistakenly assumed that the informal, unpolished ideas that they had published on their blogs entirely represented their formal scholarly output (Maron and Smith 2008). Academic bloggers may also be concerned about anonymity, their authority being questioned and threat of libel actions (Riesch and Mendel 2014).

Another barrier to academics writing blogs to be taken seriously as a form of scholarly publication is that quality control is lacking. Since there are no required credentials to write blogs, bloggers may publish content that is not tested against the quality standards of the research community, or even use false claims to attract a larger audience (Bukvova 2011). It is notable that a survey study of academics in a Finnish university found that almost half of the respondents believed that it would be more difficult to evaluate the reliability of future digital research due to the ease of spreading information online (Gu and Widén-Wulff 2011). A strategy to overcome the problem of reliability proposed by Bukvova (2011) is for scientific communities to apply quality control by reading, evaluating and commenting on published online content. In this sense, the comment and response function of blogs is similar to the traditional peer review approach for scholarly publications provided by academic journals, but in a manner which is less structured and much faster (Powell et al. 2011). However, a survey study of UK academics found that it was not yet common practice to comment on other people’s blogs (Procter et al. 2010a). Moreover, studies found that feedback for academic blogs tended to be friendlier than those from the blind peer reviews of formal publications. In other words, researchers would often receive confirmation and appreciation from the comments on their blog, in contrast to the more stringent criticism received in traditional peer reviews (Kjellberg 2010).

2.3.2.3 Twitter Twitter is the most popular public micro-blog, which is a weblog that consists of short messages (Java et al. 2007). Other popular micro-blogs include FriendFeed, Jaiku, Tumblr and Plurk (Murthy 2011). Twitter was first launched by a San Francisco-based team of 10 members in October 2006 and soon became an international phenomenon, which is popular in many parts of 56

the world, including North America, Europe and Japan (Honeycutt and Herring 2009). It allows registered users to share short messages of up to 140 characters, which are posted in reverse chronological order, to any other registered members. Such messages are called ‘tweets’. Users ‘follow’ another member in order to subscribe to that user’s posted message feeds. Over the past five years, Twitter has been adopted for scholarly activities, such as sharing information and resources, asking for advice, promoting work, and networking with peers (Veletsianos and Kimmons 2012). Twitter enables scholars across the globe to spread scientific materials to reach different communities, including their peers, students and general public (Ebner et al. 2010). Next , four rationales and purposes of using Twitter in research work will be discussed.

Information seeking and dissemination In an early study of Twitter use, Java et al. (2007: 62-63) identified three main types of user, including ‘information source’, ‘friend’ and ‘information seeker’. ‘Friend’ is a broad category and friendships can be with family, colleagues or strangers on their follower lists. ‘Information source’ users post news and tend to have a large number of followers. For academic Twitter users, the information source can be publishers, research centres, well-known scientists or any active academic users. According to a survey study, academics found Twitter to be one of the most popular digital tools to disseminate information, such as their publications, their research projects or conference promotion (Letierce et al. 2010a). Users are able to disseminate publications by including a hyperlink (URL) in their tweets. An ‘information seeker’ is a user who might rarely post anything, but follows others to search for information regularly. For an academics information seeker, Twitter is valuable for helping keep up to date on new literature in their field (Bonetta 2009). They may find new published papers from the posts by their peers, which they would otherwise not have heard about. On the other hand, academic ‘information sources’ promote themselves and enhance their visibility through Twitter. The convenience of using Twitter has been enhanced by the popularisation of smartphones, on which one can download the Twitter application without charge. Many basic mobile phones also have an internet function which can access Twitter easily. This convenience enables users, including those from impoverished countries, to search for and disseminate information rapidly and at all times (Murthy 2011).

Promotion of publications Twitter has been found to effectively promote publication and increase readership. A study found that Twitter citations were significantly faster than citations in traditional media, as 39% of citations in their sample referred to articles less than one week old, and 15% of them referred to papers published that same day (Priem and Costello 2010). Another study found that 57

disseminating publication information on Twitter either increases citations or reflects the underlying qualities of the published paper which also predicts citations (Eysenbach 2011). Academic Twitter user Melissa Terras (2012) observed her own papers’ download rate and found that seven out of ten of the most downloaded articles from her department in her institution’s repository were authored by her. She suggested that the reason was not because her papers had better quality but because she blogged or tweeted about these articles.

Usefulness in conferences Twitter has also become more and more popular for use in academic conferences to benefit the organisers and participants in many aspects. Twitter was first used as a conference communication tool as an experiment by early adopters and later became a common feature (Ebner et al. 2010). By using hashtags, conference organisers are able to disseminate information about the conference and facilitate communication between participants. In micro-blogs, hashtags were introduced to be used by a community of users interested in, and discussing, a specific topic (Laniado and Mika 2010). By placing a hash sign in front of a word (e.g. #openaccess) to represent a specific topic, it can help users search and aggregate messages related to that topic. Several studies of tweets from academic conferences used official hashtags of those specific conferences to analyse the usage of Twitter, the information flows and the usefulness of adoption. Letierce et al. (2010b) analysed tweets from ISWC 2009, a conference of Semantic Web, by crawling messages tagged #iswc2009. They found that Twitter was mainly useful to disseminate the announcement to participants of the conferences or those who were already following the technology. They noted that in this conference, although participants were willing to share information from the conference, their current tagging habits meant that the messages reached only their followers because only a low percentage of participants included hashtags related to ‘domains’, ‘applications’ or ‘documentation’ which could potentially reach a broader audience. Ebner et al. (2010) conducted a similar analysis of conference tweets with the official hashtag (#ec10hh) of EduCamp 2010 in Hamburg. Their study similarly found that the Twitter stream had a limited usefulness for external participants. Moreover, their study found that the Twitter live stream was only useful if the tweets included additional material, such as pictures, videos or slides, to help external audiences to understand the current conference situation. With a very low percentage of tweets offering such a possibility, it was problematic for non-participants.

Another study analysed information flows and citations in Tweets by using hashtags of two academic conferences, and found that a considerable percentage of users (40% and 27% of the two conferences) use URLs in their tweets, in which some of them were directed to publications 58

(Weller et al. 2011). They also found that users who had been retweeted most often were the ones who were most active with the most tweets in the dataset. However, this study lacks discussion and interpretation of their results from the analysis of URLs and retweets.

Creating support network The use of hashtags on Twitter can provide a platform for informal support networks and communities without the constraints of time and space. There are discipline-specific communities, such as #twitterstorians for History scholars, and position-specific networks, such as #phdpostdoc (Regis 2012). In the UK, #phdchat was created in 2010 to hold discussions between doctoral students and later becomes a regular live chat event on Twitter every Wednesday 7.30-8.30pm at British Standard Time (Riazat 2011). Ford et al. (2014) analysed public tweets that contained ‘#phdchat’ over a period of 39 days of 11,184 users and found that #phdchat is situated around a core group of users between 32 to 41 who attended the weekly discussion and frequently participated by sharing resources, offering advice and providing social and emotional support to one another. This core group of users can be seen as frequent users.

2.3.2.4 Social networking sites According to eBiZMBA (2014), Facebook is by far the most popular social networking site with the highest traffic rank. Other commonly used social networking sites by academics include Linkedin, Academia.edu and ResearchGate.

Facebook Facebook was created by two Harvard students in 2004 for other university students in the United States, but rapidly became a global phenomenon. Users can create online profiles with information about themselves, link with other people’s profiles as their ‘friends’, comments on friends’ profiles and send private messages to other users. Additionally, users are allowed to create groups and events that they can invite others to join (Hodge 2006). Facebook profiles have different degrees of privacy settings so that users have control over who can see them. For example, a user may choose her profile to be accessible only to ‘friends’. Facebook started the feature of the Facebook page in 2010, which enabled the creation of pages for companies, brands, persons and so on.

Individual Facebook members can join in by ‘liking’ the specific page and participating in discussions. The owners of these pages can make use of these online spaces to share photos, videos and massages with ‘fans’ without revealing any private information from their own profiles. Academics may create a Facebook page as an individual or for the courses they teach to 59

communicate with students. Academics can also create Facebook groups for their research centres or disciplinary areas to disseminate research information and interact with peers. By 6 June 2012, over 85% of the higher education institutions in the UK had both an official Facebook page and Twitter account (Kelly 2012a). Facebook is often used in higher education for teaching and learning purposes (e.g. Selwyn 2009; Roblye et al. 2010; Moran et al. 2011; Grosseck et al. 2011). To date, the literature on how academics use Facebook for communicating research has been extremely limited.

Other social networking sites There are social networking sites specifically designed for academic communities, such as Academia.edu and ResearchGate. Both ResearchGate and Academia.edu have been adopted on a large scale by researchers from the world’s elite academic institutions (Thelwall and Kousha 2014a, 2014b). Both sites provide a combination of communication and dissemination fuctions allowing users to post public questions to the community, list or upload research articles into their profiles and provide publication analytics (Ovadia 2014). ResearchGate is popular within the Sciences community for finding collaboration opportunities and discovering research articles, jobs and conferences, whilst Academia.edu enables users to share work and track the publications of academics they follow (Gross and Suttor 2013). Lupton (2013) suggests that Academia.edu functions as an open access repository with the potential to increase academic citation of those articles deposited on Academia.edu. LinkedIn, along with Academia.edu, was also found to have enhanced access to online publications and content hosted in repositories (Kelly and Delasalle 2012).

2.3.2.5 Existing evidence on academic use of social media Academic use of social media use has become a popular topic for empirical research in the past five years. Some conducted quantitative studies with survey methods (Gu and Widén-Wulff 2011; Nicholas and Rowlands 2011); some implemented qualitative approaches with interviews, case studies or activity theory (Kjellberg 2010; Coverdale 2011; Regis 2012), and others deployed web metrics analysis or content analysis (Neylon and Wu 2009a; Groth and Gurney 2010; Bar-Ilan et al. 2012; Shema et al. 2012).

In general, it is not yet the norm to use social media tools for academic research. Procter et al. (2010a; 2010b) conducted an internet survey in 2009 with a large sample of UK academics (n=1,477). Their survey results indicated that only 4% of academics wrote a research blog and 6% post slides, text and videos publicly as frequent users, while 39% of UK academics were non-users 60

of social media. Gu and Widén-Wulff (2011) found that academics in a Finnish universities (n=126) were more likely to use multimedia sharing and social networks in their daily life than in research. Nicholas and Rowlands (2011) surveyed academics all over the world (n=2,414) and found that there are two broad types of academic social media users: the ones who make joint use of blogging, microblogging and social networking, and those who focus on scheduling meetings and sharing documents. They found that the most popular social media tools for research purposes are those for collaborative authoring, conferencing, and scheduling meetings, whilst blogging and microblogging were among the least popular.

Discipline difference Findings from research by Procter et al. (2010a; 2010b) suggest that current forms of scholarly communication among UK scholars are strongly influenced by disciplinary and institutional norms. They found that academics from Computer Science and Mathematics were more likely to be frequent users than those from Medical and Life Sciences. Nicholas and Rowlands (2011) found that academic users of social media are 1.27 times more likely to be found in the Arts, Humanities and Social Sciences and 0.67 times less likely to be found in Biosciences and Health. Similarly, Mas-Bleda et al. (2014) found that social scientists were the most likely to have both personal webpages and social media presences. It is common for academic social networking sites to have disciplinary bias. For example, ResearchGate is more popular within Science disciplines (Gross and Suttor 2013). Academia.edu seems to be more popular with academics in the Social Sciences and Humanities (Thelwall and Kousha 2014a). Almousa (2011) compared Academia.edu users from four disciplines which had the widest coverage of total users and found that Philosophy and Anthropology seemed to have more active users than Chemistry and Computer Science. Sociologist Lupton (2013) suggests that based on her own experience, ResearchGate is not as beneficial as Academia.edu in terms of disseminating her deposited papers to peers because Academia.edu has far more sociologist members and thus provides more opportunities for networking and interaction in her field.

Gender difference A number of studies suggested that men were more likely to accept the use of new technology than women (Shema et al. 2012; Procter et al. 2010b). Shema et al. (2012) studied science blogs and found that men were more likely to write research blogs than women in their sample. However, a survey study of internet users found that women (75%) were more likely to use social networking sites than men (63%) (Rainie et al. 2012). Another study of Twitter users suggested that women were more likely to use Twitter (Smith and Rainie 2010). On the other hand, Hall 61

(2014) suggested that female scientists were not as successful as their male peers in gaining followers on Twitter after taking account of their citations. Hall proposed a ‘Kardashian index’ which compared scientists’ Twitter followers with their citations and very few women had a highly inflated Twitter following in his sample. Following up Hall’s work, the journal Science conducted a survey and found that 46 of 50 top Twitter scientists were men (You 2014). In response to the Top 50 Science list, a number of authors criticised the methods used by Hall and Science and suggested that many female scientists who had thousands of Twitter followers were missed out in those Twitter Science star lists (Wing 2014; Jarreau 2014). Therefore, gender difference in using social media remains controversial in the literature.

Age/seniority difference A number of quantitative studies suggest that age is inversely associated with Internet and other new media use (Dutton et al. 2005; Helsper and Eynon 2010; Smith and Rainie 2010; Rainie et al. 2012). However, empirical studies found contradictory findings about age/seniority differences in academic social media use. Nicholas and Rowlands (2011) found that younger academics were more likely to make use of blogging and microblogging than older ones. However, Priem et al. (2011) studied scholarly use of Twitter and found no evidence that rank (full-time faculty, post docs, or PhD students) disproportionally associated with Twitter use or presence. Similarly, Procter et al.’s (2010b) survey results suggested that PhD students and professors both had the highest percentage of frequent users of Web 2.0 (20%) comparing to research fellows (18%), senior lecturers (15%), lecturers (13%) and reader (6%).

In general, all the previous empirical studies suggest that it is not yet the norm for academics to use social media tools in their research work. Evidence suggests that disciplinary norms and culture may influence academics’ social media use in research work. However, findings as regards gender, age and seniority differences in social media adoption remain contradictory. Moreover, much of the existing evidence is now outdated.

2.4 Conclusions and research questions

This chapter has reviewed literature related to the traditional and new forms of scholarly communication. The new forms of scholarly communication include three aspects of open science and academic use of social media. Open science broadly refers to practices that allow cost-free 62

access to academic research. The three aspects of open science – open access publishing, sharing primary research data, publishing ongoing research updates online – were discussed with rationales, purposes, target audience, barriers, relevant policies, debates and discipline differences. Academics may use social media not only to publish research updates but also to search for research information, promote publications, popularise science and network with peers.

Gaps have been identified in the evidence on the attitudes and use of open science and social media. To date, most empirical studies on open access publishing investigated the citation impact of open access articles using citation analysis of published articles. Very little was known about to what extent UK based academics support and use open access publishing. With the recent OA policy by the RCUK came into effect at April 2013, this thesis also aims to explore whether the awareness of OA policy is associated with OA publishing practice. Moreover, what other factors can be associated with the support to OA publishing. Similarly, very limited studies have been conducted on sharing primary research data for disciplines other than Biomedical Sciences, Environmental Sciences, Physical Sciences and Social Sciences as discussed earlier. This thesis aims to explore to what extend academics share primary research data online and what factors can be associated with their sharing.

At the time of research, using social media was neither a norm nor a policy for academics. Little was known to what extent UK academics have adopted various social media tools or published ongoing research updates on social media. As a number of studies have pointed out, there were discipline, gender and age differences among various open science practice. These characteristics along with other possible factors need to be explored for their association with open science practice and other forms of social media use in academic work.

Hence, five research questions are identified as follows:

A. To what extent do academics support and use open access publishing?

B. To what extent do academics share primary research data online?

C. To what extent do academics publish ongoing research updates on social media?

D. To what extent do academics support the use of social media for research?

E. What factors are associated with the reported experiences of questions A, B, C and D?

3. Chapter Three: Research Methodology and Scoping Study Findings

3.1 Introduction

This chapter explains the research methodology. It first explains the rationale behind the methods used, then introduces the specific techniques and procedures of the research methods and finally briefly discusses the ethical issues.

This study adopted a mixed-methods approach, combining a series of scoping studies using qualitative methods followed up by an internet survey with UK academics. Such an approach was required given the limited existing research in the area. The scoping studies informed the development of the survey questions. Details of the different exploratory scoping studies are explained in Section 3.3 along with the findings from these scoping studies. Section 3.4 explains the survey and data collection process including questionnaire development, sample design, distribution of questionnaires, response rate, coding of responses and data analysis methods, as well as acknowledging the limitations of the survey. The survey results were analysed using SPSS. The ethical issues raised by the research methods are discussed in Section 3.5.

3.2 The rationale of the mixed-methods approach

The research questions as raised in Chapter Two aim to explore the attitudes and reported behaviours of academics towards open science and academic use of social media. Though there are limitations, a quantitative survey is a robust method for social researchers to gather original data to describe a population as well as measuring attitudes and orientations (Babbie 2004). Therefore, the survey method is the main method used for this study. However, the use of social media in academia was relatively new when this study started in 2011 and there were very limited studies regarding academics’ use of social media in the literature. Thus a series of qualitative scoping studies were conducted to improve the author’s understanding of this topic and develop the focus and the questions for the survey.

In theory, quantitative and qualitative research methods are from different philosophical stances. Quantitative methods are based on positivism, which bases knowledge on observable facts and implies that an objective reality exists independent of human perception (Sale et al. 2002). Therefore, investigators are capable of studying a phenomenon without influencing it or 64

being influenced by it. On the other hand, qualitative methods are based on interpretivism which emphasise the ‘interpretive, value-laden, contextual and contingent nature of social knowledge’ (Greene et al. 2005: 274). Findings of qualitative studies are created by the investigator and the study object together as they are interactively linked (Lincoln et al. 2011). Although both quantitative and qualitative methods have their weaknesses, in practice they can be used altogether as so called mixed-methods (Tashakkori and Teddlie 2003).

Greene et al. (1989) categorised the rationales of conducting mix-methods research and this current study fits into the category of ‘development’ – the results from one method can be used to help develop or inform the other method to increase the validity of constructs and inquiry results. In this case, qualitative research can be employed to develop questionnaire and scale items with better wording and more comprehensive closed answers (Bryman 2008). Thus an exploratory design was employed to gather the qualitative data from scoping studies and the results were analysed to improve the design of the quantitative study (Creswell et al. 2008). Moreover, a number of survey questions are open-ended or provide comment options to generate qualitative data. This mixed-methods approach can corroborate methodological ‘triangulaton’ to enhance the quality of data (Mason 2002). The findings from the qualitative approach offer convergence to the quantitative findings (Seale 1999).

One aim of this study is to identify to what extent academics use various social media tools. Social media tools that were discussed in published papers could have been outdated; for example, MySpace used to be ranked as the second most popular social networking site following Facebook (Stonehill-Publicity 2010). Khan (2012) listed MySpace as an example of a social networking site used in Academia. However, MySpace lost its popularity with users in general and was not in the top 15 list in October 2014 while Twitter, LinkedIn and Pinterest became second, third and fourth place respectively (eBiZMBA 2014). New social networking sites specifically designed for academia or specific academic disciplines have also emerged, such as ResearchGate and Academia.edu. Therefore, it was necessary to interview academics from different discipline areas to learn a variety of popular social media tools to form appropriate survey questions.

3.3 Scoping studies

A number of scoping studies were conducted, with the broad question of how academics publish and communicate their research to others. These scoping studies include observation, interviews and a case study of online Twitter live chat #phdchat. 65

3.3.1 Observation of social media sites After reviewing the literature on the topic of social media, a number of popular social media tools were identified. Various social media tools, such as Twitter, Facebook, WordPress, Delicious, Pinterest and LinkedIn, were explored in the author’s own research work. The author created profiles on various social media platforms which seemed to be popular for academic users – namely Twitter, Facebook and Academia.edu – and started writing a research blog on WordPress. The author also identified and read through a number of popular research blogs through the links recommended by colleagues and Twitter networks.

3.3.2 Exploratory interviews

Interview questions Between May 2012 and March 2013, thirteen interviews were conducted using a convenience sample of eight academics who had used social media for scholarly communication and five who chose not to use social media in their academic life. The interviewees who were social media users were asked when and why they started to use social media, whether and/or how they used blog and Twitter to promote their research, and whether their colleagues also used social media (see Appendix 1 for questions for users). The questions for non-user participants were shorter and they were asked to comment on using social media in research work and what would convince them of adopting social media tools in the future (see Appendix 2 for questions for non-users). All participants were asked in relation to their experiences and attitudes towards open access publishing and data sharing.

Sampling The main sampling strategy was purposeful sampling – a mixture of convenience and quota sampling. Convenience sampling involves the selection of the most accessible subjects which is the least costly in terms of time, effort and money (Marshall 1996). It allows the data collection to be completed quickly; however, it cannot enable reliable generalisations to a wider population. Quota sampling is designed to reflect a wider population in terms of important categories. It aims to ensure the inclusion of diverse elements of the population in the proportions in which they occur in the population (Selltiz et al. 1959). As in this study, these elements are gender, academic discipline and seniority (see Table 3.1). Interviewees were selected from various discipline areas to explore any disciplinary significance. Seven respondents were from Humanities and Social Sciences. Six respondents were based in Medical, Life Sciences and Engineering. The respondents were based in six different universities in various areas of England, two of which were in northern 66

England, two in the Midlands and two in southern England. Among these six universities, three of them belonged to the Russell Group, one belonged to the 1994 Group and one belonged to the University Alliance group.

Table 3.1 Details of Interviewees

Users Non-users Female PhD student in Education Female PhD student in Material Engineering Female PhD student in Biological Science Male Senior lecturer in Music Composing Female PhD student in Sociology Male PhD student in Politics Female Research assistant in Medical Science Male Research Fellow in Computer Science Male Professor in Politics Male Professor in Computer Science Male Lecturer in Social Science Male PhD student in Education Male PhD student in Biological Science

Recruitment of participants Based on the observations and literature review, Twitter was identified as the key social media tool for academic use at the time of the pilot studies. Thus the criterion for choosing interviewees who were users depended on whether they had some experience using Twitter in their research work. From November 2011 to May 2012, the author attended a number of workshops in Manchester and Oxford that were related to academic use of social media which helped recruit potential participants. Six interviewees were recruited from those acquaintances met at those workshops. In one of the social media training sessions organised by the University Library of Manchester, the tutor used an example of a blog by a PhD student from another university. After contacting this person, he agreed to participate in the interview. One interviewee was recruited through the author’s personal network. Most of the eight respondents who were social media users were fairly experienced users as the scoping studies aimed to explore their insights into the benefits and difficulties of using social media tools for research as well as the strategies used to maximise the impact of their practice.

One non-user was recruited as a participant in a workshop and two were met through social meetings. Another two non-users were recruited through the author’s personal network. All five non-users had experience of using social media tools for leisure purposes, such as Facebook and YouTube. All thirteen respondents including the five non-users were familiar with the internet and had used computers in their daily work.

Interview methods 67

Interviews were conducted using three different methods – six were face-to-face, two were by Skype instant messaging and five were via email. Semi-structured interviews were conducted in face-to-face situations and were tape-recorded and transcribed afterwards (Greenleaf et al. 2006). Given the exploratory nature of the investigation, semi-structured interviews with open-ended questions were chosen as the best means of obtaining preliminary data on the basic information to develop the survey questionnaire (Cobbledick 1996). Semi-structured interviews allow questions to be re-worded and re-ordered, and the interviewer is able to freely use prompts to encourage elaboration (Louise Barriball and While 1994). In this study, findings from the first interview helped develop questions for interviews with later interviewees.

Two interviews were conducted via Skype with social media users and five email interviews were carried out mainly with non-users. A number of the potential interviewees were located in different cities from the author. Online interviews benefit from cost and effort savings by avoiding travel to meet in person (Opdenakker 2006). However, online interviews have limitations. As effective data collection through qualitative interviewing depends upon developing rapport with interviewees (Fontana and Frey 1994), online interviews are criticised as not making it easy to build the connections and trust between interviewer and participant to generate rich data (Mann and Stewart 2000). In this study, for example, a number of respondents of email interviews might have skipped questions or given short answers on some questions. In that case, the author would send another email to probe, although probing is more difficult in email interviews in comparison to the face-to-face situation. Another disadvantage of online interview is that the rate of those refusing to participate is higher (Bryman 2012). During the fieldwork, three potential participants who agreed to be interviewed through Skype or email failed to attend or reply to further emails.

3.3.3 A case study of #phdchat on Twitter One interviewee recommended a popular online Twitter live chat and thread forum called #phdchat. The author participated in the live chat on four occasions between May and October 2012 and observed active participation from many individual users. A case study of #phdchat tweets on relevant topics employed thematic analysis.

#phdchat was first set up in 2010 by a UK-based PhD student and has become a regular live chat event on Twitter every Wednesday between 7.30–8.30pm at British Standard Time (Riazat 2011). The organisers usually post a poll of topics before-hand and participants vote to make a decision for a topic for that week. Twitter users may include #phdchat in their public messages any other time during the week in order for their messages to be seen by other users who search 68

for this term. An early participant in the #phdchat group created a wiki which stores some of the highlights from archives of previous live chats. A case study of two cohorts of topic called ‘blogging about your research’ was conducted as the topic fit into the agenda of this study. This topic was discussed as the #phdchat live chat on two occasions at 4 April 2012 and 20 June 2012. All the tweets from the live chat of ‘blogging about your research’ on two occasions were on the public wiki site and analysed for this case study. After assigning a number to the name of each participant, it showed that around 50 people participated in the first live chat and around 60 people participated in the second. The first cohort contains 527 tweets (11,705 words) and the second cohort contains 307 tweets (7,432 words).

Thematic analysis was conducted to analyse the two cohorts of #phdchat tweets deposited on the public wiki site. Thematic analysis, as a method for identifying patterns and themes, is ideal for describing qualitative data in rich detail (Braun and Clarke 2006). The contents were coded by the author into the following themes: function of blogging, blog content, worry about blogging, strategies, share of resource and good practice, sense of community and mentioning of other social media tools. Among these themes, function of blogging, blog content, worry about blogging, and share of resource and good practice were discussed the most frequently. Participants in these two live chats seemed to generally acknowledge the benefits of blogging, but also had some concern about how and what to blog, and how to reach audiences. Some of the tweets agreed that they would rather blog about research processes rather than findings. A number of participants said they would mix everything together and others preferred to keep professional blogs separate from personal ones. The most serious concern was that of ideas on the blog being stolen, but some suggested that blog entry was a way of proving the origins of ideas. Some of the issues and attitudes raised were used to inform the development of the survey questions. The findings from this case study and other pilot studies will be discussed briefly next.

3.3.4 Key findings from scoping studies on social media use

Popular social media sites and their functions The participants from the scoping studies listed a number of social media tools that they used to communicate their research. These included blogs, Twitter, Facebook, Academia.edu, LinkedIn, Scoop.it, Pearltree, Pinterest and Mendeley. Among those being mentioned, the most popular ones were Twitter and research blogs. In relation to research blogs, respondents mainly used blog sites such as WordPress and Blogger or blog sites hosted by their institutions. 69

A main function of using social media is to find information as well as to disseminate information. The participants indicated that they would post links to papers they had published or abstracts that had been accepted by conferences on blogs and Twitter. Some of them also posted links to articles that they found interesting or information about news of their workplace. Participants indicated the usefulness of social media for finding information, such as funding opportunities and conferences details. One interviewee said, ‘Twitter is good to find bits of info quickly and to keep up-to-date regarding events or news items’. A number of interviewees confirmed that they used Twitter during conferences, but the decision to use it would depend on the type of conference and whether they had access to the internet. In some discipline areas, Twitter was not yet widely adopted in conferences.

Benefits One of the benefits of social media use is getting advice and feedback you otherwise would not be able to get quickly. Respondents who had large numbers of followers on Twitter pointed out the benefit of getting timely feedback once they posted a question. Another benefit that was identified was the increase of readership and thus citations. One interviewee believed that the presence and promotion of his work on social media would bring more citations. Research impact as measured by citation rate is closely related with the academic reward system, which can, in turn, heavily influence a scholar’s professional status and reputation.

Potential Problems One of the perceived potential problems of using social media for work purposes is the risk of unwanted content being linked to one’s professional social media account which could hurt one’s reputation. Other potential problems included any inappropriate uses of the content from one’s social media account. For example, an interviewee found out that a picture from his website was used by a company for commercial purposes without consulting him. A number of interviewees expressed worries in regard to their ideas being stolen from contents in their blogs and micro- blogs.

Non-user attitudes The non-users in the scoping studies gave a list of reasons for rejecting or ignoring social media for scholarly communication. The most common reasons were the lack of time and perceived incentives. Other reasons included a lack of trust for online information that is not peer reviewed or is written by anonymous authors, a preference for face-to-face or traditional communication methods, and subject area or professional status reasons. A number of non-users suggested that 70

as content on social media was not peer-reviewed it therefore was not trustworthy. One respondent emphasised the importance of traditional peer-reviewed papers and pointed out that participation in social media was not recognised in the current UK Research Excellence Framework10 assessment criteria and was thus not worth investing time in. Several non-users indicated that once more benefits were confirmed and patterns of wider usage became established, they might adopt social media in the future. They also suggested that peers’ recommendation and their positive experiences were of great encouragement.

Strategies for social media use The participants revealed a number of ways in which they sought to maximise the value to them of using social media to support their research. The most commonly used strategy to maximise the impact of having various social media accounts was to link them together for cross- platform promotion. Linking accounts from different social media sites together was an easy and quick way for scholars to promote their online content. For example, when they uploaded presentation slides on SlideShare, embedded them on their blog, then tweeted about them and published them in their other social media accounts such as LinkedIn and Schoop.it. This cross- platform promotion helped disseminate the presentation slides to more audiences on various social media sites, in contrast to the sole audience on SlideShare.

Researchers may use free web tools to manage their activities on social media. For example, one respondent used a tool called Buffer11 to schedule her tweets to be spread throughout the day, which overcame the problem of flooding her followers. In this way, the respondent only organises her tweets once in the morning via Buffer and let Buffer dispatch the tweets at different times during the day. Buffer helped the respondent improve efficiency and save time. While a big criticism of using social media is the possibility of wasting time and getting distracted, this web technology enables users to dispatch the tweets more effectively and not get distracted throughout the day.

Another strategy identified by the interviewees was to create or join existing groups to form a support network for themselves. On Twitter, many hashtagged keywords had been created and deployed to form discussion of specific topics. For example, the live chat #phdchat on Twitter formed an informal network in which participants could seek and provide advice and information to one another. Similarly, on Facebook many groups and pages have been created to support local

10 http://www.ref.ac.uk/ accessed 8 December 2014 11 http://bufferapp.com/ accessed 12 February 2015

or international research communities. Similarly, on blog sites bloggers can follow other bloggers and comment on their posts, thus attracting others to comment on their posts and gradually forming a network for themselves.

Using copyright licenses was suggested as a strategy to protect one’s online intellectual contents. With the emergence of the open access movement, many scientific records, including published articles, data, presentation slides, methods and tools have been made freely available to all internet users via various forms of online publishing and under public copyright licenses, such as Creative Commons (Wilbanks 2006), which allows the reuse of content as long as the source is acknowledged. A number of interviewees used the Creative Commons license on their own web pages and blog sites. Scholars may also put informal warnings on their social media sites, such as putting a note next to the content they created and take action, such as reporting to website administrators, when plagiarism happens.

One strategy used by interviewees for tackling privacy issues was to have two different identities on Facebook and Twitter; for example, using Facebook for their private identity and Twitter for their professional identity. A number of respondents separated their personal social media profiles from their professional accounts.

Interviewees agreed that they would be cautious about content they posted online to make sure that they could secure academic rewards. Similar to the views from participants of the #phdchat case study, a number of interviewees stated that they would rather blog about the process of writing the articles versus the actual findings. Interviewees all agreed that they would not be happy if their collaborators revealed early research findings on social media without notifying them.

Overall, for these participants, social media seems to be used as a tool to reflect on their work, allowing scholars to ruminate over the process of doing research rather than revealing the findings, which could potentially compromise collaboration with peers and undermine traditional publications.

Disciplinary differences There is some evidence that disciplinary differences do have an impact on the perceived value of social media in scholarly communication. An interviewee in Politics suggested that unlike in the Hard Sciences, there were no ‘results’ in his area, but were analyses and claims of varying degrees of certainty. In addition, the political scientist commented, ‘in political communication there aren’t that many people doing the same kind of research, so it’s not likely anyone would suddenly 72

design a project to take advantage of our research’. In Hard Sciences disciplines, however, scientists might only show results of pre-publication research in order not to reveal findings to competitors.

3.4 Online survey of academics

The evidence from the scoping studies was used to inform the design and focus of the survey. An online survey was conducted to explore academics’ attitudes towards and experiences with open science and using social media for various reasons in research. Respondents were sent an invitation email with a link to the online survey. An online survey is suitable for a PhD project since it has the advantage of saving time and cost compared to post or phone surveys (Wright 2005). Respondents may also take an online survey at his/her convenience, whilst phone surveys require an immediate response (Evans and Mathur 2005). In academia, it is common practice for academics to use Internet and emails for work-related communication (Hole 2008), thus an online survey invited by email is practical and appropriate for this study.

3.4.1 Initial questionnaire design The first draft of the questionnaire was structured upon the 2009 survey questionnaires from the study conducted by Procter et al (2010a). Based on findings from the scoping studies and the research questions, a number of questions were added and some wording of questions was changed. The approach to get more detailed information was to ask a number of open-ended questions and give options for comments with some of the questions. The survey was created online with Survey Monkey.

The questionnaire had four main sections. The first section collected background information including gender, age, institution, discipline, job grade, teaching status and research experience. The second section asked questions about respondents’ attitudes and experiences in disseminating and searching for information in their research work, including their experiences with open access publishing and data sharing, whether they promoted their publication by personal contact or posting on social media, their attitudes towards the importance of open access to research outputs, and the tools they used to search for scholarly articles. The third section focused on the use of social media for research, and asked questions about online services that the respondents used, how frequently they used them, why they started using them and whether they received training or encouragement from their institutions and colleagues. The final section asked questions related to research impact. Respondents were asked whether they 73

would be assessed in the 2014 REF, whether they had a permanent academic job, whether they had created a profile in Google Scholar, their estimate of publications, citation counts and H-index. Finally they were asked to what extent they agreed or disagreed with 12 statements related to the influence of using social media and publishing in open access journals. A full version of the questionnaire is in Appendix 4.

3.4.2 Developing and piloting the survey

Cognitive interviews to test the questions Cognitive interviews are a type of interview for pretesting survey questionnaires, which study the ‘cognitive process that respondents use to answer survey questions; in particular, their comprehension, recall, decisions and judgment, and response processes’ (Willis 2005: 6). The goal is to identify any response problems which might occur in answering survey questionnaires. Respondents are encouraged to think aloud when answering interview questions while responses are probed extensively (Jobe and Mingay 1989). Cognitive interviewing was employed in this current study as a technique that was found to be effective at evaluating specific question wording and reasonably reliable (Campanelli 1997).

Two cognitive interviews were conducted with two colleagues to test the wording of the questionnaires in detail. The first one was conducted with a female who had six years’ research experience in Biological Science. The second interview was conducted with a female who had two years’ research experience and who was based in Sociology. The two interviews were conducted in the author’s office with a quiet environment for approximately one hour each. The two participants were asked to read aloud what they were thinking when they read the questions on the computer screen. They were encouraged to ‘think aloud’ about what was required to answer each question and to point out whether any question was problematic, ambiguous or sensitive. They were also asked whether the wordings of any question were too long, how confident they felt about the rating tasks, whether they had any problems understanding any questions and how they felt about the open-ended questions. The questionnaires were amended afterwards according to the feedback provided by the two interviewees.

Survey pilot The questionnaire was further tested for language, structure, layout, length and wording with eleven colleagues (ten from the University of Manchester and one from another Russell Group university; seven male and four female; title including professor, senior lecturer, lecturer, research associate and PhD student; four from Medical and Natural Sciences and seven from 74

Humanities and Social Sciences). An email with the link to the online survey was sent to each of them and they emailed feedback after completing the survey. The respondents were asked to be honest and comment whether they felt the survey was too long and whether there was any problematic or ambiguous wording. The questionnaire was amended taking account of their comments. For example, the questionnaire was criticised for being slightly too long. Thus a number of questions or items to be chosen in some answers were revised or removed. Several people pointed out double questions and missing response categories for some questions, which were amended accordingly. The format of several questions was changed to make them simple and easy for people to answer. By the middle of June 2013, the questionnaire was ready to be distributed.

3.4.3 Sample justification and strategies The sampling frame of the survey was the population of all academics in the Russell Group universities. The population for the sample was chosen because the Russell Group universities all have a strong research focus. As such the research would be capturing the attitudes and behaviour of some of the academics at leading research focused universities.

Firstly, the Russell Group, ‘represents 24 leading UK universities which are committed to maintaining the very best research’12 and are well-acknowledged in the world as elite universities for the impact of their research. In August 2012, four member universities of the 1994 Group left to join the Russell Group. By 2013, the 24 members of the Russell Group were reported as accounting for around half of the UK higher education sector’s total income and this elite group took £3.3 billion of £4.5 billion sector-wide for research income (Gill 2013). Acquiring the label of a Russell Group university could be seen as adding value since evidence was found that a Russell Group graduate earn a higher income than those graduating from a modern university (Chevalier and Conlon 2003).

Secondly, most UK universities, including those of the Russell Group, had a social media presence on the internet and as such it would be interesting to see whether academics based in these institutions followed the steps of their employers in terms of adopting new media tools. Studies of Russell Group universities found that 19 out of 20 Russell Group universities had a Twitter account by 28 June 2011 (Kelly 2011). By January 2012, 15 out of 20 Russell Group universities posted links to their institutional Facebook and Twitter accounts on their official university home pages (Kelly 2012b). When the author searched for the universities that had no

12 http://www.russellgroup.ac.uk/ accessed 2 November 2013 75

Twitter account or were not included in Kelly’s 2011 study, the results indicated that all 24 Russell Group universities (including the four new ones that joined in August 2012) had a Twitter presence by June 2013. Therefore, it seems that all Russell Group universities had recognised the influence of social media and had adopted them for marketing and publicity in various degrees.

Thirdly, since existing studies found differences in practice between disciplines in open access publishing and using social media for research (as discussed in Chapter Two), this study also explores whether there are significant differences between different academic disciplines. Russell Group universities are all large universities with a broad range of disciplines for both Humanities and Sciences (RussellGroup 2012). Thus the sample was designed to capture a more representative picture of various disciplines in UK higher education.

As the nature of this study is a PhD project with limited funding resources, the strategy was to use clustering to lower the cost of distribution of the survey (Weisberg and Bowen 1977). Each university became a primary sampling unit (PSU) and half of PSUs were chosen in the sample (Moser and Kalton 2004). Ten out of twenty original Russell group universities and two out of four new group members were selected randomly (see Table 3.2 for the selected university list). In theory, all academics in selected PSUs were selected for the sample and all of the twelve units’ email addresses would be harvested from those universities’ websites. However, this method would have excluded academics that had no email address listed on their university website.

3.4.4 Email address harvesting and data cleaning After the sample units of twelve universities were selected randomly, the author worked closely with an IT professional 13 to harvest all the email addresses on these twelve universities’ official websites. The email addresses were captured using a script written in the Perl programming language. Perl programming is commonly used for automatic data mining on the internet (Pham and Bogdan 2010). The harvesting generated a long list of email addresses from each of the twelve universities. As the survey was targeted for academics, irrelevant email addresses for each university were deleted, including email addresses containing admin, service, help, support and email addresses without the universities’ specific domain.

Because of the large number of email addresses in the sample, it was not realistic to check whether each email address belonged to an academic by cross-referencing to the person’s information on his/her university website. Therefore, all the remaining email addresses were used

13 Andy Macheta from the IT service, University of Warwick

for the sample and respondents were notified in the invitation email that the population targeted for this study was academic researchers. The final numbers of email addresses in the sample was 42,008.

3.4.5 Distribution of the survey Hosker (2008) suggests using a covering letter for a postal survey which identifies the researcher, explains the purpose of the survey and why it is important for the reader to participate, and also addresses the issue of confidentiality. A similar approach was conducted by sending an invitation email to each email address in the sample (see Appendix 3 for survey invitation email).

The invitation email was kept simple and short in order to encourage responses and contain only vital information. The invitation letter first introduced the topic of the survey and emphasised the author’s status as a PhD student who needed help to complete a PhD project. It then suggested the time needed to complete this survey as 10 minutes and expressed the author’s gratitude for respondents’ effort in filling in the survey as well as emphasising the confidentiality of the data. There was also the incentive of entering a prize draw for £50 of book vouchers. Finally, it provided the link to the survey with a thank you and the email address of the author in the case potential respondents wanted to make contact. The signature included the author’s name, university addresses and telephone number. The email invitation was first piloted and then revised. A number of lists were created to distribute emails on a few occasions to avoid sending all of them at once where there is a possibility that they could be detected as spam. The distribution of emails first started on 19 June and was completed by 2 July 2013.

Interest in a survey topic is found to have a positive effect on the likelihood of completing an online survey (Galesic 2006). If the invitation email and introductory text of the survey mentioned the study was related to social media use, it might have attracted those who were enthusiastic about social media tools. To avoid social media users being over-represented, the invitation email and introductory text for the survey mentioned nothing about social media.

3.4.6 Closing of the survey and responses rate No reminder emails about the survey were sent in order to avoid being regarded as spamming which could lead to possible negative responses. The survey was closed on Tuesday 6 August 2013 after being live for approximately one month. In total 1,841 responses were received, with a completion rate of 89.7% (1,651). The response rate of the survey was 4.4% (1,841/42,008); however, the email addresses in the sample were for both academics and supporting staff as it 77

was impossible to identify the academics from the non-academics with the methods used. According to the HESA (2013), in 2011-2012, there were 181,385 academics and 196,860 non- academics in all UK higher education institutions. Accordingly, half of the email addresses harvested from those sampled universities might belong to non-academics, who were ‘units outside the population’ (Moser and Kalton 2004: 169). The author also received a number of emails from sampled individuals indicating that they were not based in an academic post. Therefore, the real response rate of this survey could be higher than 4.4%.

In Table 3.2, all sampled universities are listed and ordered from the highest to the lowest in terms of their response rates (response count/harvested email addresses). Newcastle University and the University of Edinburgh had the highest response counts of 240 and 230 respectively, whilst Queen’s University Belfast and the University of Warwick had the lowest response counts of 54 and 45. However, taking account of total harvested emails, the University of York and King’s College London had the highest response rates of 6.7% and 6.1% respectively, whilst the University of Leeds had the lowest response rate of 2.5%.

The harvested email addresses for the University of Glasgow and the University of Leeds were more than twice the number of academic staff, whilst most other universities’ harvested email addresses were not as many as twice the numbers of academics. Thus it was likely that less than half of the email addresses in the sample for the University of Glasgow and Leeds were for academics. This could explain the low response rates of these two universities as their response counts were among the middle range in all sampled universities. As for the University of Warwick, for which the number of harvested email addresses was lower than the number of academics suggested by HESA, the response count and response rate was relatively low compared to other universities. Most email addresses of the University of Warwick were hidden with Javascript on the Warwick website and the addresses captured might be largely made up of auxiliary staff. Thus only 1,400 email addresses were harvested from the Warwick website which was much less than the academic numbers suggested by the HESA.

The non-response could be due to various reasons. The most common reason was refusals, as some people invariably refuse to co-operate and others may be too busy to respond (Moser and Kalton 2004). Email overload related stress which was reported for academics (Jerejian et al. 2013) might also contribute to non-response. Similar to ‘away from home’ and ‘out at time of call’ for post and phone surveys, not receiving or missing the survey invitation email could be the reason for non-response of the online survey. The completion rate of the survey was 89.7%. One of the reasons for non-completion was that respondents who were non-academics broke off after 78

finding out that the questions were designed for academics. Other reasons for breakoff for web surveys include lack of interest, high task difficulty, distractions and technical problems (Roßmann et al. 2014). As discussed in Section 3.4.1, the third section of the questionnaire focused on the use of social media. Those who had no interest in using social media might have dropped out at this stage. High task difficulty, often related to long and open-ended questions perceived as a cumulative burden, could increase the probability of breakoff (Peytchev 2009). Distractions were easier to happen when respondents used mobile devices to answer the survey. Technical problems could also happen. For example, the link to next page of the questionnaire could not be opened due to poor internet access (Roßmann et al. 2014).

Although the sample is not random, the response rate is substantial and one of the largest of its kind to examine the issue of attitudes towards open science amongst academics. The fifth survey question asked how many years the respondents worked as an academic researcher which would have identified those non-academics as well as increasing the breakoff and non-completion.

Table 3.2 Responses table to the universities questions

Sampled Universities Harvested Academic No. Non-academic Total of Response Response Response rate emails 2011-2012 by No. 11-12 by staff by percent count (after (Response HESA HESA HESA coding data) count/harvested University of York 2205 1410 1900 3310 8.1% 148 6.7% King's College London 3193 3920 2405 6325 10.6% 195 6.1% Imperial College London 2287 3690 3210 6900 7.5% 137 6.0% University of Manchester 4033 4415 4820 9235 11.5% 211 5.2% Newcastle University 4743 2360 3070 5430 13.1% 240 5.1% University of Edinburgh 5223 3350 4730 8080 12.6% 230 4.4% Queen's University Belfast 1270 1515 1860 3375 2.9% 54 4.3% Cardiff University 3122 2705 3240 5945 6.6% 121 3.9% Durham University 2822 1505 2475 3980 5.7% 104 3.7% University of Warwick 1400 1795 3065 4860 2.5% 45 3.2% University of Glasgow 5080 2655 3190 5845 8.9% 163 3.2% University of Leeds 6630 2940 4005 6945 9.1% 166 2.5% Other 0.9% 17 Skipped universities Qs 10 Total 42008 32260 37970 70230 100.0% 1841 4.4%

3.4.7 Coding of survey responses The survey responses were downloaded to Excel and saved in a password protected computer at the University of Manchester. Two responses were problematic as their comments suggested that they filled in the survey only to express their anger and frustration in receiving a survey invitation which they regarded as spam. A further ten responses were also excluded because the respondents only filled in one or two questions. The final response count for the analysis of this study was 1,829. 79

The coding process was first conducted in Excel. There were 44 main questions in the survey and 25 questions had an ‘other (please specify)’ option with a comment box. The responses were downloaded as numerical values with answers to comments and open-ended questions in Excel spread sheet format. All the comments for those 25 questions were investigated and coded accordingly. Some comments represented one of the choices given in the question that respondents had missed. In that case, they were coded into the pre-coded categories. For example, several respondents did not realise that their university was in the answer list and wrote ‘Newcastle’ or ‘Glasgow’ in the comment boxes. A number of respondents failed to notice that archaeology was in the discipline list and wrote it in the comment box. Some comments were coded as missing values – 77 for ‘unsure/not specified’, 88 for ‘no answer’ and 99 for ‘not applicable’ – based on each specific case.

There were six open-ended questions. Three of them asked for specific numbers for research outputs, citation counts and H-index and provided a small box for each answer. The other three questions that provided a much bigger box for each answer invited comments regarding blog contents, reasons for using social media in research work and additional comments about the topic. The answers to the three open-ended questions for comments were excluded while coding in Excel as they could not be turned into numerical values for quantitative analysis and would be kept in their original format for qualitative analysis. The answers to three questions that asked for numbers were coded to numerical symbols. If the answer was ‘about 250’, it was coded as ‘250’. If the answer was ‘30-40’, it was coded as the mean of the two estimates (30+40)/2= 35. Answers such as ‘no idea’, ‘don’t know’, ‘don’t care’, ‘not applicable’ and those with no answers were coded as missing values. The coded data were exported to SPSS for quantitative analysis.

Table 3.3 Summary of background characteristics of survey respondents (N=1,829) 80

Variables N % N % Gender Female 836 46% Discipline Medical & Life Sciences 635 35% Male 977 54% Area Natural Sciences & Engineering 415 23% Other 6 0% Social Sciences 490 27% Total 1819 100% Arts & Humanities 279 15% Age group Under 35 633 35% Total 1819 100% 35-44 475 26% Grade Professor/Reader 421 24% 45-54 390 21% Senior lecturer/Senior researcher 286 16% 55 and over 322 18% Lecturer/Research Fellow/Postdoc 681 39% Total 1820 100% Researcher in training 356 20% Research 1-5 years 441 24% Total 1744 100% Experience 6-10 years 399 22% 11-20 years 476 26% over 21 years 481 26% N/A 32 2% Total 1829 100%

Table 3.3 gives a summary of the background characteristics of respondents included in the final analysis. Approximately 46% were female (836) and 54% were male (977). The original question for age had six categories and was recoded into four categories. As there were less than 100 respondents aged ‘under 25’ and ‘65 and over’, they were each combined with their neighbouring group. The four categories of age group were: ‘under 35’ (31%), ‘35-44’ (26%), ‘45- 54’ (21%) and ‘55 and over’ (18%). The original question for job grade had ten options. These have been recoded into four groups in order to reduce the categories and make the comparison between groups more meaningful. The original question for job grade was regrouped to: researcher in training (20%) including PhD candidate, Masters student and research assistant, lecturer/research fellow/postdoc (39%), senior lecturer/senior researcher (16%) and professor/reader (24%). In relation to research experience, respondents were asked to indicate a number (years) on a scale from 1 to ‘45 and over’, with a ‘not applicable’ option. The research experience of the respondents was grouped into 1-5 years, 6-10 years, 11-20 years and over 21 years to represent junior to more senior levels. These four groups were evenly spread across the sample. Respondents who answered ‘not applicable’ (n=32) to research experience were coded as ‘missing’ to ensure that only academics who were involved in research were included in the analysis.

The original question for academic discipline had 36 categories with an ‘other (please specify)’ option. These disciplines were listed in the same order as the official 2014 REF categories with four categories. These four categories were named by the author as four discipline areas— Medical and Life Sciences (35%), Natural Sciences and Engineering (23%), Social Sciences (27%) and Arts and Humanities (15%). Overall this means that 58% of respondents were from Medical 81

and Natural Sciences and 42% were from Humanities and Social Sciences. The divide between Sciences and Humanities subjects is similar to data from 2011-2012 HESA (2013) statistics, which estimated that 53% of UK academics were in Sciences and 47% were in Humanities if part-time staff were included, while 61% of full-time academics were in Sciences and 39% were in Humanities. The number of respondents from each university broadly represented the size of that university.

3.4.8 Data analysis methods The percentages of key questions are reported in Chapters Four and Five.

The aim of the survey questionnaire analysis is to ‘examine patterns among replies to questions and explore the relationship between variables that the questions represent’ (May 1993: 102). In this study, descriptive statistics, inferential statistics and statistical modelling are used to analyse the survey data. Broadly speaking, three levels of measurement can apply to social sciences data – nominal, ordinal and interval – and most survey questions in this study are nominal or ordinal, which can both be called categorical variables. The frequency distributions of two or more categorical variables can be examined using a cross tabulation, as long as each variable does not have too many categories (Lewin 2005). A cross tabulation is a useful procedure or technique for measuring the association between two variables (Zeller and Carmines 1978). Descriptive statistics are applied in Chapters Four and Five using cross tabulations to measure the associations between various open science practice and social media use variables with key background characteristics. In addition, inferential statistics are applied to test the significance level of these associations using Chi-square tests (Barnes and Lewin 2005). The Chi-square test examines whether there is a real association between two categorical variables. If the significance value p is small enough (usually < 0.05), the null hypothesis that the two variables are actually unrelated to each other can be rejected (Field 2009). This method of analysis is used in Chapters Four and Five for exploring whether there are any differences between different groups by discipline, gender, age, job grade or research experience.

Statistical modelling is applied in Chapter Six with logistic regression modelling as the main method. Key variables are described as dependent or independent. A dependent variable is explained by reference to the influence of the independent variable (May 1993). Logistic regression methods are useful in models with a binary dependent variable for determining whether or not a particular exogenous variable exerts a statistically significant effect on the outcome variable (Sanders and Brynin 1998). This modelling method is based on the notion of 82

probabilities – the probability of an event occurring over the probability of it not occurring. Chapter Six reports findings from logistic regression models for exploring what factors are associated with the likelihood of social media use and open science practice.

The comments given for six open-ended questions and a number of 25 single/multiple choices questions were coded using thematic analysis. The main themes were positive compared to negative comments. Some direct quotes are used in analysis in Chapters Four and Five.

3.4.9 Limitation of the survey As only half of the Russell Group universities were selected for the sample, the survey is limited in its representativeness for other UK higher education institutions. Moreover, exclusion bias resulted from exclusion of particular groups from the sample, such as those having no email addresses listed on their university websites. It is also possible that the techniques used failed to harvest certain email addresses from sampled universities’ websites.

Web-based survey was found to have much lower response rates compared to post surveys for the same kind of study (Hardigan et al. 2012). The response rate of this study is similar to other Internet surveys for studying scholarly communication (e.g. Procter et al. 2010a; Nicholas and Rowlands 2011), which indicates that 4.4% of response rate is reasonable for this type of web survey. The sample used in this study is reflective of Russell Group universities because of the selection methods used. Even though it is not a random sample, they survey is the biggest of its kind to date and has strong explanatory power; and thus it is robust enough to conduct analysis.

Comparing to the data sourced from the Higher Education Statistics Agency (HESA 2013), this sample was broadly representative of the wider UK academic population in terms of the weight of responses by our primary background characteristics of gender, academic discipline and age. As such, whilst there are considerable limitations to the data, the large number of responses provides a rich source of data to examine the research questions.

3.5 Ethical issues

This study complied with the University of Manchester Graduate School’s guidelines and the Social Research Association. The ethical approval was granted by the University of Manchester Ethics Committee in May 2012. 83

Interviewees from the qualitative interviews were provided with a Participant Information Sheet and asked to sign a consent form. If they were in a remote location, this was done by emailing an electronic copy. The interviewees were briefed about their rights as research subjects, covering issues such as data protection, anonymity, confidentiality and the ability to terminate the interview at any time if they pleased, with no ramifications. Twitter data from two cohorts of #phdchat live chat were on the public wiki site. As contents on public websites can be regarded as a collection of documents, it is ethical to extract these contents as data for analysis given that anonymity is provided (Wilkinson and Thelwall 2011).

The ethical issues raised by web surveys include how to appropriately obtain informed consent, administer effective withdrawal procedures and ensure anonymity and confidentiality of responses (Hewson and Laurent 2008). In the invitation email that contained the link to the online questionnaire, it was stated that all responses would be treated confidentially. The sampled population were informed that the study had been approved by the Ethics Committee and that they could write to the author directly with enquires or if they wanted to be removed from the sample list. Consent of survey response was indicated by clicking the ‘Done’ button at the end of the survey, however, a withdrawal procedure was not available once the respondents started filling in the survey.

The collection of email addresses from public websites is a commonly used methodological technique for conducting survey (Schaefer and Don 1998; Procter et al 2010a; 2010b); however, their compilation into a database can raise data protection issues. The email addresses were deleted straight after their use. The large-scale emailing of an organisation’s email addresses can also raise ethical concerns including in relation to spam email (Muller et al. 2014). However as the email addresses were publicly accessible, there should be no restrictions on this.

4. Chapter Four: To What Extent do Academics Support Open Science?

Summary: This chapter presents the survey results of respondents’ attitudes towards and experiences of the three aspects of open science: (i) allowing open access to research articles; (ii) sharing primary research data online; and (iii) publishing ongoing research updates on social media. It provides an overall picture and examines whether there are differences between groups defined by gender, discipline, age, job grade or research experience.

4.1 Research question A: To what extent do academics support and use open access publishing?

Section 4.1 explores the first aspect of open science – allowing open access to research articles. It provides an overall picture and examines differences between groups by gender, discipline, age, grade and research experience.

4.1.1 Overall attitudes and experience of OA publishing

4.1.1.1 Attitudes towards OA publishing As discussed in Chapter Two, open access publishing refers to research articles being published either through open access journals (Gold OA) or self-archiving via online repositories (Green OA). All respondents were asked to rate how important they thought it was, in general, to make research articles freely accessible online to everyone, as shown in Table 4.1.

Table 4.1 Attitudes towards the importance of making research articles freely accessible online

How important do you think it is, in general, to make research articles freely accessible online to everyone? N % Very important 987 57% Fairly important 619 36% Not very important 104 6% Not at all important 12 1% Total 1722 100%

The vast majority of respondents (93%) acknowledged the importance of open access publishing. Of the 1,722 respondents, in total 1,606 rated it ‘very important’ (57%) or ‘fairly important’ (36%) for research articles freely to be made accessible online to everyone, while only 85

12 (1%) and 104 (6%) of the respondents rated it ‘not at all important’ and ‘not very important’ respectively.

4.1.1.2 Experience of OA publishing When asked whether they had published articles in a journal that was open access, 108 respondents (6%) indicated that publishing articles was not applicable to them as they had not or did not produce research articles, and these types of responses were filtered out of the frequency table. Of the remaining 1,601 respondents who had already published research articles, 41% (649) answered ‘yes’ to having published an OA journal, 31% (501) stated ‘no, but I plan to in the future’, 10% (163) stated ‘no, I have no plan to publish in an OA journal’, and another 18% (288) answered ‘no, not sure about this’ (See Table 4.2). The respondents who answered ‘no, not sure about this’ might have been unaware of OA journals as well as being unsure of whether they would choose OA journals in the future.

Table 4.2 Experience of Gold OA (n=1,601 as a subgroup of respondents who had published research articles)

Have you published an article in a journal that is open-access? (Gold OA) N % Yes 649 41% No, but I plan to in the future. 501 31% No, I have no plan to publish in OA journal. 163 10% No, not sure about this. 288 18% Total 1601 100%

When asked whether they had deposited research articles (published or working paper) in an open access online repository (Green OA), 94 respondents (5%) stated that they had not or did not produce research articles and were thus filtered out. Of the remaining 1,614 respondents who had produced research articles, 43% (696) answered ‘yes’, 26% (415) stated ‘no, but I plan to in the future’, 8% (126) stated ‘no, I have no plan to do so’, and another 23% (377) responded ‘no, not sure about this’ (See Table 4.3). The respondents who answered ‘no, not sure about this’ might not be aware of the existence of OA repositories as well as being unsure of whether they would self-archive in the future.

Table 4.3 Experience of Green OA (n=1,614 as a subgroup of respondents who had published research articles) 86

Have you deposited your research articles in an open-access online repository? (Green OA) N % Yes 696 43% No, but I plan to in the future. 415 26% No, I have no plan to do so. 126 8% No, not sure about this. 377 23% Total 1614 100% Those who noted having no plan to publish in Gold OA and Green OA tend to be male, aged 55 and over, senior academics and those who have had over 21 years’ research experience. This will be discussed further below.

4.1.1.3 Awareness of OA repositories and OA policy When asked whether they were aware of open access repositories for depositing research articles, out of 1,733 respondents, 26% answered ‘no’ and 74% answered ‘yes’. When asked whether they were aware of RCUK policy on open access to the outputs of RCUK-funded research which came into effect on 1 April 2013, out of 1,716 respondents, 42% (721) answered ‘yes’, 28% (473) answered ‘no’, and another 30% (522) selected ‘heard of, but not sure about the detail’. Thus the majority of respondents were aware of the opportunities of Green OA publishing (74%), but were not aware of the details of RCUK’s Open Access policy (58%). The lack of awareness of OA repositories and RCUK’s Open Access policy might be the reason that a large number of respondents either indicated that they had no plan to publish in Gold or Green OA or were ‘not sure about this’.

Those who were not aware of RCUK’s OA policy tended to be female and younger, less experienced PhD students. For example, over half of the researchers in training (54%) answered ‘no’ to having an awareness of RCUK’s OA policy. Female and senior academics were also more likely to have no awareness of open access repositories for depositing research articles. Again this will be discussed in later sections.

4.1.1.4 Reasons for not publishing through OA channels In order to evaluate the overall experience of using Gold and Green OA publishing, the two questions regarding whether respondents had published with Gold and Green OA were combined together (see Table 4.4). After excluding ‘not applicable’ answers for both variables, there were 1,585 respondents who had published research articles. Approximately 23% (357) had published through both Gold and Green OA channels, 38% (602) had published through one of the OA channels, and another 39% (626) stated they had no experience in this. Thus, within the 87

subsample of academics who had already published articles, 61% had experience of at least one type of open access publishing.

Table 4.4 Experience of Gold or/and Green OA (n=1,585 as a subgroup of respondents who had published research articles)

If you have published research articles , have you had Gold or/and Green OA experience? N % No 626 39% Yes to one 602 38% Both 357 23% Total 1585 100%

After excluding respondents who had no experience of publishing, the percentage of respondents who rated it ‘very important’ or ‘fairly important’ to make research articles freely accessible online was still 93%. As only 61% of those who had published articles used Gold or/and Green OA routes, there was a 32% difference between attitudes and reported experiences of making articles freely accessible online. This difference could be related to the lack of knowledge in relation to OA publishing, OA repositories and funders’ OA Policy.

Of the 625 respondents who had published articles in open access journals (Gold OA), 24 (4%) paid out of their own pockets for the author fees for their last OA publication. Another 364 (58%) stated that their institutions/research funds paid and 47 (8%) claimed that their collaborators paid. In total 30% of them paid no fees as 31 (5%) claimed that publishers waved the fees and 159 (25%) stated that no fee was required (see Table 4.5).

Table 4.5 Author fee in last OA publication (n=625 as a subgroup of respondents who published research articles in Gold OA journals)

If yes, did you or your institution pay for the author fee in your last OA publication? N % Yes, my institution/research fund paid. 364 58% Yes, I paid out of my own pocket. 24 4% No, our collaborators paid. 47 8% No, the publishers waved the fee. 31 5% No, no fee was required. 159 25% Total 625 100%

Among the respondents who paid an author fee out of their own pockets, 67% of them were professors or senior academics. PhD students and junior academics might be less likely to pay APCs out of their own pockets because of lower salaries compared to senior academics. 88

As shown in Table 4.6, all respondents were asked whether they preferred to publish research articles in open access journals rather than subscription-based journals if they had similar reputation or ranking of citation impact. Of the 1,626 academics whose work involved publishing research papers, 642 (39%) stated that they had no preference and it all depended on which journals had higher reputations in their field and 498 (31%) would prefer to publish in open access journals only if they personally did not need to pay the author fees. Another 164 (10%) respondents preferred conventional subscription-based journals and 146 respondents (9%) stated that they preferred OA journals even if they personally had to pay author fees. The Gold OA model, with the high cost of APCs, sets up barriers for researchers with limited resources.

Table 4.6 Preference of Gold OA (n=1,626 as a subgroup of respondents whose work involved publishing research articles)

In general, do you prefer to publish research articles in openaccess journals rather than subscription based journals if they have similar reputation or ranking of citation impact? N % Yes, I prefer OA journals even if I personally have to pay author fee. 146 9% Yes, I prefer OA journals only if I personally don't have to pay author fee. 498 31% No, I prefer conventional subscription-based journals. 164 10% I don't have a preference, it all depends on which journals have higher reputation in my field. 642 39% Don't know enough information about this matter. 176 11% Total 1626 100%

The survey findings suggested that lack of knowledge and awareness about OA publishing together with the high cost of APCs could hinder OA practice. The comments provided by survey respondents explained rationales for supporting OA publishing and also barriers and concerns. This will be discussed next.

4.1.1.5 Understanding attitudes towards OA publishing When asked to rate the important of making research articles freely accessible online to everyone, 594 respondents also left comments in the boxes expanding their thoughts. Many of the respondents from different academic disciplines gave similar reasons as to why they thought open access to research articles were important. As highlighted by two responses, a key aspect was the way in which research was funded:

‘Work is generally publically-funded, so the public should be able to access it! Plus researchers in developing countries may not have the funds to pay subscriptions.’ (Female, Research fellow/postdoc, Public Health, Health Services and Primary Care)

‘Academic journal publishing is an outrageous con, whereby a huge amount of money is siphoned from the taxpayer to private companies in subscription fees, while all the editorial and reviewing work is done free by academics. The general public has paid for 89

these articles, and the general public should have unfettered access to the fruits of their investments’. (Male, Lecturer, Philosophy)

While acknowledging its importance, some respondents were also concerned about potential problems such as copyrights, the issue of quality and not being able to pay author fees. These concerns may explain the differences between attitudes regarding the importance and reported experiences of OA publishing. For example, a senior lecturer stated her concerns about the perceived lower quality of OA journals and not having funds to pay author fees:

‘In principle I am all for open access to other researchers, however in practice it is open to abuse (i.e. copyrights not always adhered to by all readers) and I also do not have the funds to pay the fees that are charged. I have also noted significantly lower quality in online access journals, where authors have to pay to get their articles published.’ (Female, Senior Lecturer, Biological Sciences)

One lecturer in Classics also raised a concern about the potential problems of the author- payment model for OA journals which could discourage and exclude disadvantaged researchers:

‘In principle yes, I prefer open-access journals, but the system of author-payment is very dangerous; it will discourage and possibly exclude young / independent scholars, or those working at impoverished institutions.’ (Female, Lecturer, Classics)

Respondents who indicated that it was ‘not at all important’ or ‘not very important’ to make research articles freely accessible online to everyone were often concerned about the quality control of open access journals or believed that their work would only be understood by other specialists and their research articles could be misinterpreted by non-academic audiences. A senior lecturer in Applied Health stated his distrust of OA journals’ quality:

‘I do not believe that open access journals are of as good quality in terms of peer reviewing and therefore do not rate them highly.’ (Male, Senior lecturer, Applied Health Professions, Dentistry, Nursing and Pharmacy)

The possibility of misinterpretation by non-academic audience was also a concern as raised by a researcher from Business and Management Studies:

‘While in principle open access is a good idea I believe there are some potential issues/hazards in the way that research is interpreted and used by a non-academic audience.’ (Female, Research fellow/post-doc, Business and Management Studies)

Similarly a reader from Biological Sciences commented:

‘Anyone in a position (with the required expertise) to understand my research would have institutional access to it. Lay persons would not understand the work … In biology it is almost never that a “fact” is “proven”. Lay folk do not understand what this means and often imagine the opposite (e.g., since no scientist can be 100% sure MRI vaccine is “completely safe”, some lay folk assume this means that there is a good chance it is unsafe.’ (Male, Reader, Biological Sciences)

This respondent did not feel that it was his responsibility to help the public understand his research. This raises a question of how to communicate research to the general public in a language that they can understand and who should take responsibility for that.

The next sections explore the relationships between OA publishing and various background characteristics to examine whether there are differences between groups defined by gender, discipline, age, grade or research experience.

4.1.2 Differences in OA publishing by academic discipline As explained in Section 3.4.8, descriptive statistics ran by cross tabulation and inferential statistics using Chi-square tests are the main methods used in Chapters Four and Five to explore the relationships between two observed variables. The first focus is academic discipline. In general, respondents in Medical and Life Sciences were most likely to publish in Gold OA journals. Those in Natural Sciences and Engineering were most likely to publish through the Green OA route.

Table 4.7 Experience of Gold OA by academic discipline (n=1,596 as a subgroup of respondents who had published research articles)

Have you published an article in a journal that is open-access? (Gold OA) Yes No, but I plan to No, I have no plan to No, not sure Total p in the future. publish in OA journal. about this. value Medical & Life N 354 144 32 49 579 Sciences % 61% 25% 6% 8% 100% Natural Sciences N 123 117 44 66 350 & Engineering % 35% 33% 13% 19% 100% Social Sciences N 115 152 51 108 426 % 27% 36% 12% 25% 100% Arts & N 56 84 36 65 241 Humanities % 23% 35% 15% 27% 100% Total N 648 497 163 288 1596 0.000 % 41% 31% 10% 18% 100% 91

As shown in Table 4.7, of the 579 respondents in Medical and Life Sciences, 61% (354) had published articles in OA journals, compared to 35% (123 out of 350) in Natural Sciences and Engineering, 27% (115 out of 426) in Social Sciences and 23% (56 out of 241) in Arts and Humanities. Respondents in Medical and Life Sciences (6%), were also least likely to choose ‘no, I have no plan to publish in OA journal’, whilst the proportion of those in the other three discipline areas were between 12-15%. Respondents who stated ‘no, not sure about this’ were more likely to be in Arts and Humanities (27%) and Social Sciences (25%), compared to 19% in Natural Sciences and Engineering and only 8% in Medical and Life Sciences. The Chi-squared test gave an overall p value of <0.001 which indicated a statistically significant association between experience of Gold OA and academic discipline.

Table 4.8 Experience of Green OA by academic discipline (n=1,609 as a subgroup of respondents who had published research articles)

Have you deposited your research articles in an open-access online repository? (Green OA) Yes No, but I plan to No, I have no No, not sure Total p in the future. plan to do so. about this. value Medical & Life N 231 145 45 146 567 Sciences % 41% 26% 8% 26% 100% Natural Sciences N 193 78 29 57 357 & Engineering % 54% 22% 8% 16% 100% Social Sciences N 180 124 30 104 438 % 41% 28% 7% 24% 100% Arts & N 92 64 22 69 247 Humanities % 37% 26% 9% 28% 100% Total N 696 411 126 376 1609 0.001 % 43% 26% 8% 23% 100%

For Green OA publishing, over half of respondents in Natural Sciences and Engineering (54%, 193 out of 357) had deposited their research articles in OA repositories compared to 41% (231 out of 567) in Medical and Life Sciences, 41% (180 out of 438) in Social Sciences and 37% (92 out of 247) in Arts and Humanities. Those in Natural Sciences and Engineering (16%) were least likely to choose ‘no, not sure about this’ for Green OA publishing whilst the proportions of those in the other three discipline areas were between 24-28%. The proportion of respondents who selected ‘no, I have no plan to do so’ was similar for all four discipline areas (between 7-9%).

Respondents in Arts and Humanities seem to be the least experienced with both Gold and Green OA publishing (23% and 37%). The proportion of respondents in Arts and Humanities (15%) who selected ‘no, I have no plan to publish in OA journal’ was more than twice that of respondents in Medical and Life Sciences (6%). 92

Respondents in Natural Sciences and Engineering (82%) were most likely to be aware of the OA repositories for depositing research articles whilst the proportion of respondents who answered ‘yes’ to the awareness of the OA repositories in the other three discipline areas ranged from 71% to 73% (overall p<0.001). Thus there is around 10% difference between those in Natural Sciences and other discipline areas.

The disciplinary difference of OA repository awareness could be due to the long existence and adoption of OA repositories in Natural Sciences and Engineering. A number of respondents in Natural Sciences and Engineering commented that they frequently used a preprint repository called ArXiv to search and self-archive research articles. Using repositories as an information source and dissemination tool was the norm in Natural Sciences, such as Mathematical Sciences and Physics as suggested by two respondents:

‘We have “arxiv", i.e. almost every article in high energy physics is freely available since about 1992.’ (Male, Reader, Physics)

‘All articles in my field are made available for free on the internet at arxiv.org. Publication after that only serves as peer review and not as dissemination.’ (Male, Professor, Mathematical Sciences)

The difference in the availability of tools for various discipline areas could explain why academics in Natural Sciences and Engineering were ahead of peers from other discipline areas in terms of their knowledge and practice of Green OA publishing.

4.1.3 Differences in OA publishing by gender In general, men were more likely to have experience of using both Gold and Green OA publishing compared with women. As shown in Table 4.9, of the 869 men, 43% (370) had published articles in OA journals compared to 37% (276 out of 718) of women (p<0.001), amounting to a 6% difference.

Table 4.9 Experience of Gold OA by gender (n=1,587 as a subgroup of respondents who had published research articles) 93

Have you published an article in a journal that is open-access? (Gold OA) Yes No, but I plan to No, I have no plan to No, not sure Total p in the future. publish in OA journal. about this. value Female N 276 232 57 153 718 % 38% 32% 8% 21% 100% Male N 370 262 105 132 869 % 43% 30% 12% 15% 100% Total N 646 494 162 285 1587 0.001 % 41% 31% 10% 18% 100%

The difference between men and women in terms of their experience of OA publishing was larger for the Green OA route (9%). As shown in Table 4.10, out of 873 men 47% (411) had deposited their research articles in open access repositories compared to 38% (277 out of 727) of women (p<0.001).

Table 4.10 Experience of Green OA by gender (n=1,600 as a subgroup of respondents who had published research articles)

Have you deposited your research articles in an open-access online repository? (Green OA) Yes No, but I plan to No, I have no plan No, not sure Total p in the future. to do so. about this. value Female N 277 194 38 218 727 % 38% 27% 5% 30% 100% Male N 411 217 87 158 873 % 47% 25% 10% 18% 100% Total N 688 411 125 376 1600 0.000 % 43% 26% 8% 24% 100%

Men were more likely to be aware of RCUK’s policy on open access to the outputs of RCUK- funded research with 48% of men (446 out of 925) answering ‘yes’ compared to 35% of women (269 out of 777). In total 24% of men answered ‘no’, indicating they were not aware of the OA policy, compared to 31% of women (p<0.001).

In relation to OA repositories for depositing research articles, overall men (79%, 738 out of 933) were also more likely to know about them compared to women (69%, 540 out of 786) (p<0.001).

4.1.4 Differences in OA publishing by age, job grade and research experience In general, respondents who were older, more experienced and in higher job grades had more experience of both Gold OA and Green OA publishing14. The respondents’ awareness of OA

14 p<0.001 for all six associations between age / job grade/ research experience and Gold/Green OA publishing. 94

repositories and RCUK’s OA policy also increased with age, job grade and research experience. Older and senior academics usually have more experience with publication and thus they learn about funding councils’ OA policy and the existence of OA repositories.

Of the 401 professors/readers, 55% (221) had published articles in OA journals, compared to 41% (110 out of 268) of senior lecturers/senior researchers, 38% (242 out of 634) of lecturers/research fellows/postdocs and 26% (59 out of 229) of researchers in training. In relation to Green OA publishing, 56% (223 out of 399) of professors/readers had self-archived research articles, compared to 46% (123 out of 266) of senior lecturers/senior researchers, 45% (290 out of 644) of lecturers/research fellows/postdocs and 18% (43 out of 238) of researchers in training. As age, research experience and job grade are positively correlated with one another, the patterns for age and research experience were similar to job grade for the awareness and experience of OA publishing.

Those aged under 35, researchers in training and respondents with research experience between 1-5 years were the least likely to have published in OA journals or deposited papers in OA repositories. The findings that younger and junior academics were less likely to be aware of OA repositories and RCUK’s OA policy may explain why they were less likely to have experience of both Gold and Green OA publishing. The age difference will be further explored in Chapter Six using statistical modelling.

The next section explores the second aspect of open science – sharing primary research data online. It provides an overall picture and examines differences between groups by gender, discipline, age, grade and research experience.

4.2 Research question B: To what extent do academics share primary research data online?

4.2.1 Overall attitudes and experience of sharing data online

4.2.1.1 Attitudes towards sharing research data online As discussed in Chapter Three, in this study ‘data’ refer to primary research data and metadata as opposed to research articles. Although many data repositories have restricted access and may require users to register or sign in with their institution user names, sharing data online for reuse 95

still implies an open attitude towards open science. Thus open access to data refers to cost-free access to primary research data that can be reused by other researchers.

All respondents were asked to rate how important they thought it was, in general, to make research data available online for reuse. As shown in Table 4.11, the vast majority of respondents acknowledged the importance with 1,459 out of 1,695 (86%) respondents rating it ‘very important’ or ‘fairly important’ to make research data available online for reuse. This proportion is slightly lower than those who rated making research articles accessible online as important (93%). The proportion of respondents who rated it ‘very important’ (40%) to make research data available online for reuse was much lower than those who rated it ‘very important’ to make research articles freely accessible online (57%). There were 12% of respondents who rated it ‘not very important’ to make research data available online for reuse in comparison to just 6% for making research articles freely accessible online. The proportions of respondents who rated it ‘not at all important’ were similar for both questions at around 1%. Those who rated it ‘not very important’ or ‘not at all important’ tended to be evenly distributed across all discipline areas and in terms of age and gender.

Table 4.11 Attitudes towards the importance of making research data available online for reuse

How important do you think it is, in general, to make research data available online for reuse? N % Very important 673 40% Fairly important 786 46% Not very important 211 12% Not at all important 25 1% Total 1695 100%

4.2.1.2 Experience of using and sharing research data online Around one-fifth of respondents had experience of sharing their own data. As shown in Table 4.12, of 1,724 respondents 360 (21%) reported having deposited their own primary research data in online repositories that could be reused by other researchers.

Table 4.12 Experience of depositing primary data in online repositories

Have you deposited your own primary research data in an online repository that can be re-used by other researchers? N % Yes 360 21% No 1364 79% Total 1724 100% 96

Respondents were also asked whether they had used secondary research data from online repositories that were collected by other researchers. As shown in Table 4.13, a slightly higher percentage of respondents answered ‘yes’ (29%, 499 out of 1,728) to having used secondary research data collected by others in their research work.

Table 4.13 Experience of using secondary data from online repositories

In your research work, have you used secondary research data from an online repository that were collected by other researchers? N % Yes 499 29% No 1229 71% Total 1728 100%

Table 4.14 indicates the characteristics of respondents who had used secondary data by whether they had deposited data. Around 47% of respondents who had used secondary data for their research had also deposited their own primary data online for reuse; while only 10% of those who had no experience using secondary data had deposited their own primary research data in online repositories that can be reused by other researchers (overall p<0.001). Thus respondents who had used secondary research data from online repositories collected by other researchers were more likely to have shared their own primary research data online.

Table 4.14 Use of secondary data by depositing data

Deposited data p value, yes no Total overall Used secondary data yes N 234 263 1223 % 47% 53% 100% no N 125 1098 497 % 10% 90% 100% Total N 359 1361 1720 0.000 % 21% 79% 100%

4.2.1.3 Rationales of sharing data In total 408 respondents left comments about the importance of making research data available online for reuse. This is a very rich stream of data to explore and means it is possible to develop my understanding of the issues surrounding the sharing of data. One of the most common rationales of making data available online for reuse was similar to the rationale of making research articles freely accessible online as commented by a lecturer in Engineering: 97

‘If the research is funded by public fund, then the data should be accessible for public.’ (Female, Lecturer, Civil and Construction Engineering)

Many respondents acknowledged the benefits of having primary research data available online for reuse, such as improving quality and reducing instances of fraud. As one senior lecturer in Medical Sciences stated:

‘It should stop duplication of research and transparency ensures honesty and quality.’ (Male, Senior lecturer, Psychology, Psychiatry and Neuroscience)

Making primary research data available online for reuse can advance science by validating research findings, avoiding duplicate data collection, saving time and cost, maximising the use of resources and encouraging collaboration. These advantages were highlighted by a number of respondents from various discipline areas:

‘Validation of research findings by the community; pump-priming ideas from other scientists and giving value for money from tax-funded research.’ (Male, Senior lecturer, Clinical Medicine)

‘I think that secondary analysis is becoming increasingly important. Also, a lot of primary data is collected, perhaps analysed once or twice for thesis, a report or a paper and then is forgotten about - which is a waste! Finally, it is very useful for research data to be available online for comparative purposes.’ (Female, Lecturer, Education)

‘It reduces the resources required (time and money) to make research discoveries and encourages collaboration.’ (Female, Lecturer, Psychology, Psychiatry and Neuroscience)

The requirements from major UK funding councils and some journals such as PLoSOne should also be taken into account as a reason why some academics share their primary data.

4.2.1.4 Challenges and barriers A number of respondents discussed the challenges of sharing data based on what kind of data they produced. One issue is the time and effort required to produce detailed meta-data alongside the primary data; otherwise, the data could be misinterpreted by others without first-hand insights. This issue could be a barrier in Sciences or Humanities as stated by respondents from various discipline areas:

‘Of course it depends on the kind of data. It could require quite a lot of extra work to make the data clear and easy to access without ambiguity.’ (Male, Research assistant, Aeronautical, Mechanical, Chemical and Manufacturing Engineering) 98

‘It depends on the nature of the data, and outside the framework of the project it was gathered for it may be less useful, or may possibly be misinterpreted.’ (Female, Research fellow/postdoc, History)

‘The use of any data without data collectors’ insight has a great risk for misinterpretation.’ (Female, Senior lecturer, Clinical Medicine)

There were also concerns related to ethical and confidentiality issues, especially when there are human subjects involved. For example, a senior lecturer in Sociology stated the challenge of anonymity and confidentiality of human participants with qualitative data.

‘I work with qualitative data, where issues of confidentiality and anonymity are paramount… Placing data in online repositories open for all would feel too risky in my mind with regards to protecting research participants' anonymity, confidentiality, but could also be a breach of trust if the data were to be used in unethical ways.’ (Female, Senior lecturer, Sociology)

Similarly, a PhD student in History raised an ethical and privacy issue:

‘It depends on the type of data. I can see instances where it would be very important (maybe in science when repeat experiments could be done) but this is not always the case. Some of my interviews I would be very reluctant to put online for ethical/privacy reasons.’ (Female, PhD candidate, History)

Among the 1,303 respondents who had not deposited primary research data in online repositories, 10% (135) selected ‘no, because of ethical issues’ when asked whether they would share primary research data online to be reused by others in the future. Respondents from Medical and Life Sciences, Social Sciences and Humanities were more likely to be concerned about ethical issues compared to respondents from Natural Sciences and Engineering.

Some academics might only want to share with colleagues privately and with whom they could collaborate to produce research papers together. For example, a research fellow/postdoc in Sociology explained her reason of choosing ‘not very important’ for making research data available online for reuse:

‘I think researchers should freely share research data with those colleagues who email and ask personally. So if someone is interested in the primary data that I have collected then they can email me, I will share the data and we will produce a paper together. I resent, however, spending the time and effort gaining ethics approval, securing access to respondents, spending the time collecting the data and then transcribing all of the interviews for someone else to swoop in and just use the end product (i.e. the data that I produced).’ (Female, Research fellow/postdoc, Sociology) 99

Another lecturer from Biological Sciences highlighted concerns about competition:

‘I am not sure that all data needs to be available to everyone immediately because of issues with competition.’ (Female, Lecturer, Biological Sciences)

It seems academic competition can be a major barrier for sharing data as academics often depend on their primary data to publish findings and advance their career. A researcher could spend considerable time preparing the primary data they have collected. Sharing would be potentially benefitting competitors. When asked whether they would share primary research data online in the future, 10% (126 out of 1,303) of respondents who had not deposited data in online repositories selected ‘no, I want to secure publication’. Another 39% (512 out of 1,303) stated ‘not sure’.

A number of respondents commented that they were unsure about the benefits and usefulness of sharing primary research data and the usefulness of sharing may vary for different disciplines. For example, a research fellow/post-doc in Engineering suggested disciplinary differences of primary research data’s usefulness:

‘There are existing repositories that work well for data intensive research (genomics, proteomics), but not all research really benefits from these. Most of my work is deposited in institution data centres and can be accessed upon enquiry, but it is also within the publication (in the shape of statistical summaries) and the usefulness of the raw data is unclear.’ (Male, Research fellow/postdoc, General Engineering)

These disciplinary differences will be examined in detail in the next section.

4.2.2 Differences in using and sharing data by academic discipline Overall respondents in Natural Sciences and Engineering were more likely to use secondary research data from online repositories. Of the 391 respondents in Natural Sciences and Engineering, 34% (134) had used secondary research data from online repositories, compared to 29% (137 out of 465) in Social Sciences, 29% (76 out of 261) in Arts and Humanities and 25% (150 out of 603) in Medical and Life Sciences (overall p<0.05). Respondents in Natural Sciences and Engineering also appeared to be slightly more likely to share primary research data online, but the differences were small and not significant. 100

As a number of respondents commented when considering the possibility of sharing primary research data, anonymity and confidentiality could be more of an issue in some disciplines than others:

‘Not always very ethical - in my discipline but the value and the feasibility may be very different in other disciplines.’ (Female, Lecturer, Sociology)

‘Really depends on the data and how sensitive it is, due to ethical issues. Also you don't know how other people are going to use it - I would hate for my data to be used in a way that I didn't agree with, yet at the same time, it saves money and inconvenience to multiple participants. The issues are endless.’ (Female, Research Fellow/postdoc, Applied Health Professions, Dentistry, Nursing and Pharmacy)

The difference in data sharing issues in various disciplines was addressed by one respondent:

‘It really depends on the nature of the data. It is certainly not applicable to all disciplines; any plans have to take into account the intricacies of data collection and sampling in different disciplines.’ (Female, lecturer, English Language and Literature)

Data sharing is more common in some disciplines than in others as shown by survey results and as highlighted by the comments from the respondents. Some disciplines could have a long history of a data sharing culture and formation of popular peer-reviewed data repositories such as the RCSB Protein Data Bank (Maron and Smith 2008). Some disciplines might not produce any primary data that can be reused. It is clear that concerns about competition and ethical issues are also important factors.

4.2.3 Differences in using and sharing data by gender In general, men were more likely to share primary research data online compared to women. Of the 927 male respondents, 24% (226) reported ‘yes’ to having deposited their own primary research data in online repositories that can be reused by other researchers, compared to 17% (132 out of 783) of women (overall p<0.001). Men were also more likely to use secondary research data from online repositories that were collected by other researchers. Of the 929 men, 32% (298) reported ‘yes’ to having used secondary data compared to 25% of women (196 out of 785) (overall p<0.001).

The possible explanation for gender differences in using and sharing data could be related to women being less aware of OA repositories for depositing research data. Since women were less likely to be aware of RCUK’s OA policy, they could also be less likely to hear about funding councils’ 101

data sharing policies. The significance of gender difference will be further investigated in Chapter Six.

4.2.4 Differences in using and sharing data by age, job grade and research experience In general, respondents who were older, more experienced and in higher job grades were more likely to deposit their own primary research data in online repositories. These patterns were consistent with the experiences of depositing research articles online (Green OA). Of the 404 professors/readers, 35% (140 out of 404) had deposited primary research data in online repositories, compared to 21% (58 out of 271) of senior lecturers/senior researchers, 19% (128 out of 657) of lecturers/research fellows/postdocs and 9% (27 out of 314) of researchers in training (p<0.001). Respondents’ experience of depositing primary research data also increased with age and research experience.

Unlike the reported experiences of depositing data, no such pattern was found for using secondary data by age, grade or research experience. Around 34% (106 out of 313) of researchers in training and 33% (136 out of 407) of professors/readers used secondary data compared to 24% (66 out of 270) of senior lecturers/senior researchers and 25% (168 out of 660) of lecturers/research fellows/postdocs (p<0.01). No significant difference was found between groups defined by age or research experience.

The next section explores the third aspect of open science – publishing ongoing research updates on social media. It provides an overall picture and examines differences between groups by gender, discipline, age, grade and research experience.

4.3 Research question C: To what extent do academics publish ongoing research updates on social media?

4.3.1 Overall attitudes and experience of publishing ongoing research updates on social media The survey asked respondents how often they posted updates of ongoing research on (i) ‘blogs’, (ii) ‘Twitter’ and (iii) ‘social networking sites (e.g. Facebook and ResearchGate)’, with the choices of ‘never’, ‘sometimes’, ‘often’ and ‘always’. The questions gave a list of various social media services. These three social media services are quite different in their services provided. Twitter only allows messages of up to 140 characters. Social networking sites such as 102

Academia.edu and ResearchGate allow users to upload a full paper or add a link to an online paper. Blogs usually have no words limit and academics may write a summary or a short story of their published journal paper. It is notable that a number of participants from the pilot interviews of this study stated that they would link their Twitter and Facebook accounts to their blogs to publicise their blog posts.

4.3.1.1 Twitter The vast majority of respondents had ‘never’ used Twitter to post updates of ongoing research with (84%, 1,401 out of 1,673). As shown in Table 4.15, 1% (25) stated that they ‘always’ posted updates on Twitter, 4% (65) stated ‘often’ and 11% (182) stated ‘sometimes’.

Table 4.15 Experience of publishing ongoing research updates on Twitter

How often do you post updates of ongoing research on Twitter? N % Never 1401 84% Sometimes 182 11% Often 65 4% Always 25 1% Total 1673 100%

4.3.1.2 Blogs The proportion of respondents who ‘never’ used blogs to post updates of ongoing research was similar to the use of Twitter, at around 84% (1,407 out of 1,668). As indicated by Table 4.16, 1% (17) ‘always’, 2% (40) ‘often’ and 12% (204) ‘sometimes’ used blogs to post updates of ongoing research.

Table 4.16 Experience of publishing ongoing research updates on research blogs

How often do you post updates of ongoing research on a research blog? N % Never 1407 84% Sometimes 204 12% Often 40 2% Always 17 1% Total 1668 100%

4.3.1.3 Social networking sites As shown in Table 4.17, out of 1,671 respondents 81% (1,360) stated they ‘never’ used social networking sites to post ongoing research updates, 1% (10) stated ‘always’ and 3% (46) ‘often’. The slightly higher rate of users compared to the use of blogs and Twitter was because of the 103

higher percentage of those who ‘sometimes’ (15%, 255 out of 1,671) posted updates of ongoing research.

Table 4.17 Experience of publishing ongoing research updates on social networking sites (SNs)

How often do you post updates of ongoing research on social networking sites (eg.,Facebook & ResearchGate)? N % Never 1360 81% Sometimes 255 15% Often 46 3% Always 10 1% Total 1671 100%

4.3.1.4 Super users Those who had ‘always’ and ‘often’ posted research updates on social media could be described as ‘super users’. Combining the respondents who selected ‘always’ and ‘often’ together, Twitter appears to be the most popular tool for these types of users to post updates of ongoing research (5%), with 90 out of 1,673 respondents choosing ‘always’ or ‘often’ compared to 57 out of 1,668 (3%) using blogs and 57 out of 1,671 (3%) using social networking sites.

When looking at the use of the three suggested social media services together, the majority of respondents (70%, 1167 out of 1677) had ‘never’ posted updates of ongoing research on any of those three social media services. A small proportion (2%, 32) ‘always’ posted updates of ongoing research on at least one of those social media sites. Another 6% (107) ‘often’ posted updates of ongoing research online, while 22% (371) ‘sometimes’ posted research updates online. As shown in Table 4.18, 8% (139) of the respondents can be seen as ‘super users’ in publishing one or more of the three social media tools. Factors associated with the likelihood of being a super user publishing research updates on social media will be investigated in Chapter Six.

Table 4.18 Experience of publishing ongoing research updates on social media

Experience of publishing ongoing research updates on any of blog/Twitter/SNs N % Never 1167 70% Sometimes 371 22% Often 107 6% Always 32 2% Total 1677 100%

104

Academic social media users seemed to be younger and juniors from Humanities and Social Sciences, as will be discussed further below.

4.3.1.5 Understanding why academics publish research updates When asked what kind of information respondents blogged about and why they wrote research blogs, a small number of respondents explained how they blogged updates of ongoing research and the purposes of doing so. One respondent stated that he would blog about findings to keep a record and seek ongoing feedback:

‘I write updates about the research that I do; detailed discussion about exactly what I have done recently, and why, including any results. It's a handy way of keeping a lab book, gives me something to refer to, and other people read it, giving me rapid feedback on my work.’ (Male, Lecturer, Computer Science and Informatics)

Another stated that he blogged about important breakthroughs to keep stakeholders engaged in the study:

‘My research team blogs updates about the study to keep agencies and potential beneficiaries engaged in the study throughout. I blog on my personal blog about ongoing research for the same purpose. The updates are usually milestones reached or preliminary observations from the data.’ (Male, reader, Social Work and Social Policy)

The function of self-discipline and self-motivation provided by blogging was raised by a professor from Modern Languages and Linguistics:

‘My ongoing work -- the joys and the woes of it! -- as a kind of self-discipline or self- motivating activity.’ (Male, Professor, Modern Language and Linguistics)

One respondent who stated that he ‘sometimes’ read research blogs and gathered research information on Twitter but ‘never’ posted updates of ongoing research on any of those suggested social media sites emphasised the difference between communicating published research and ongoing research and the risk posed:

‘You need to distinguish between communicating published research and communicating research in progress. Huge difference. Communicating completed research is outreach, communicating ongoing research is giving the game away.’ (Male, Senior researcher, Chemistry)

105

This kind of view is related to the tension between communicating research on social media and securing more formal academic recognition. In order to secure academic rewards and not give ideas away to potential competitors, many academics may be reluctant to discuss their ongoing research on the internet which can be leaked to their competitors.

4.3.2 Differences in publishing ongoing research updates on social media by academic discipline Respondents who posted ongoing research updates online were more likely to be in Arts and Humanities and Social Sciences. Respondents in Medical and Life Sciences were most likely to be non-users of social media sites to post ongoing research updates.

When asked whether they used Twitter to post research updates, 22% (57 out of 256) of respondents in Arts and Humanities and 21% (92 out of 445) in Social Sciences indicated ‘always’, ‘often’ or ‘sometimes’ compared to 14% (52 out of 382) in Natural Sciences and Engineering and 12% (71 out of 584) in Medical and Life Sciences (overall p<0.001). The patterns were similar for the use of blogs and social networking sites to post ongoing research updates.

For the combination of using any of the three platforms to post ongoing research updates, 40% (103 out of 256) of respondents in Arts and Humanities and 38% (169 out of 446) in Social Sciences stated that they had ‘always’, ‘often’ or ‘sometimes’ posted research updates online compared to 27% (102 out of 383) of those in Natural Sciences and Engineering and 23% (134 out of 586) in Medical and Life Sciences (overall p<0.001).

4.3.3 Differences in publishing ongoing research updates on social media by gender In general, the survey results indicate only limited difference between men and women in terms of whether and/or how often they posted updates of ongoing research online when combining the use of the three suggested social media services together. Women (19%, 144 out of 756) were slightly more likely to use Twitter to post ongoing research updates than men (14%, 126 out of 903) (p<0.05). Twitter users who ‘always’ or ‘often’ posted ongoing research updates were slightly more likely to be women (7%) than men (4%). 106

Men and women were similar in using any of the three platforms to post ongoing research updates. No significant difference was found between men and women for the use of blogs or social networking sites to post research updates.

4.3.4 Differences in publishing ongoing research updates on social media by age, job grade and research experience In general, respondents who were younger, less experienced and in lower job grades were more likely to publish ongoing research updates on social media.

In terms of age, respondents aged under 45 were most likely to have used Twitter to post ongoing research updates. In total 22% (125 out of 556) of those aged under 35 and 21% of those aged 35-44 (91 out of 438) had posted research updates on Twitter, followed by 11% (42 out of 373) of those aged 45-54 and 4% (13 out of 298) of those aged 55 and over (p<0.001). The experiences of posting ongoing research updates on Twitter, blogs and social networking sites all decreased with age for those aged 35 and over.

In relation to research experience, respondents with over 21 years research experience were least likely to use Twitter, blogs or social networking sites for posting ongoing research updates. For example, only 6% (28 out of 459) of respondents with over 21 years’ research experience had used Twitter to post ongoing research updates, compared to 18% (66 out of 377) of those with 6- 10 years’ research experience, 21% (81 out of 388) of those with 1-5 years’ research experience and 22% (94 out of 434) of those with 11-20 years’ research experience (p<0.001).

Similarly, professors/readers who generally would have had more research experience were least likely to use Twitter and social networking sites to post research updates. Out of 399 professors/readers, 12% (46) had used Twitter to post ongoing research updates, compared to 14% (35 out of 257) of senior lecturers/senior researchers, 19% (121 out of 642) of lecturers/research fellows/postdocs and 20% (60 out of 300) of researchers in training (p<0.01). Lecturers/research fellows/postdocs (24%, 152 out of 641) were most likely to have posted ongoing research updates on social networking sites compared to 17% (51 out of 298) of researchers in training, 15% (40 out of 259) of senior lecturers/senior researchers and 13% (52 out of 397) of professors/readers.

The patterns of age and seniority differences for publishing ongoing research updates on social media are different to those for publishing research articles and data online. While the older and senior academics had more experience of using open access publishing and sharing research data, 107

the younger and junior academics had more experience of publishing research updates on social media.

4.4 Conclusion and discussion

This Chapter has examined survey respondents’ attitudes towards and experiences of open access publishing, using and sharing research data online, and publishing ongoing research updates on social media. It explored whether there were differences between groups defined by gender, discipline, age, job grade or research experience.

The evidence suggests that the vast majority of respondents (93%) agreed with the principle of making knowledge freely available to everyone but a number of respondents raised concerns related to author fees, copy right, quality control of OA journals and the danger of misinterpretation of their work by non-academic audiences. Of the respondents who had publishing experience, 41% had published in open access journals through the Gold OA route and 43% had deposited papers online through Green OA route. Overall, 60% of respondents had experience in using at least one of the open access publishing routes.

The difference between attitudes and practice may be due to respondents’ lack of knowledge about the existence of OA repositories and the requirements for publishing through OA routes by major research funders. Approximately 26% of respondents were not aware of OA repositories and another 58% indicated that they were not aware or were not sure about the details of RCUK’s OA policy which came into effect on 1 April 2013. This is in line with studies by Xia (2007), Xia and Sun (2007) and Kim (2011), which suggest that many academics had little awareness of OA publishing or chose to remain in ignorance of its implications in spite of having heard of the term (Swan 2006). Other reasons for the difference between attitudes and practice may be related to potential problems such as copyrights, quality concerns and not being able to pay author fees. A number of respondents expressed their concerns about the quality of OA journals and not having the funds to pay APCs.

Similar to the findings from a study by Swan and Brown (2004) that 4% of authors of OA journal articles paid the author fee by themselves, this current study found that 4% of respondents in our survey who had published articles in open access journals (Gold OA) paid the author fees out of their own pockets for their last OA publication. When asked whether they prefer to publish in OA journals rather than subscription-based journals if they had similar 108

reputation or ranking of citation impact, 31% of respondents stated that they would prefer to publish in OA journals only if they personally had no responsibility of paying the author fee, while almost 40% of them only cared about journals with higher reputations in their field. The evidence suggests that the reputation and citation impact of the journals remain a key factor for decision making regarding choice of journal, which is in line with findings from other studies (Rowlands and Nicholas 2006; Solomon and Björk 2012).

The costs of author fees for the Gold OA model have become barriers for academics who are not funded by research councils or who have limited resources to publish in Gold OA journals. An alternative model for making research articles open access without paying an author fee is the Green OA which provides opportunities for financially disadvantaged researchers to self-archive their work which has been published in non-OA journals (Chan and Kirsop 2001).

An early study with international senior academics in 2005 found only 16% of respondents had deposited an academic paper in institutional repositories and 33% indicated no intention for depositing articles, with UK authors being more ignorant of repository movement (Rowlands and Nicholas 2006). The findings of this thesis suggest that institutional repositories have gained popularity and obtained more users over the years in the UK since 43% of respondents share research articles in online repositories and only 8% indicated no plan to do so. UK academics appear to be increasingly using Green OA publishing.

Respondents who had a higher awareness and more experience of using OA publishing tended to be male, older, senior and more experienced. Since senior academics would be more likely to write funding applications to research councils, they would be more likely to know about OA repositories and RCUK’s OA policy and thus more likely to have experience of using both Gold and Green OA publishing. For junior academics, if they had worked in a research team on projects that required publishing in OA journals or their senior collaborators preferred to publish through the Gold OA route, these junior members would have had Gold OA experience by default. However, early career researchers who have had no experience of applying for funding with research councils would be less likely to learn about OA publishing and policy. As job grades are positively correlated with age and research experience, this explains why the patterns for these three variables reported in this section were similar for the awareness and experience of using OA publishing.

Descriptive analyses of the survey data suggested that men were more likely to have used both Gold and Green OA publishing compared with women. Gender inequality in the workplace is not a 109

new topic. Women usually earn less than men and this gap remains both persistent and universal (Lips 2013). In academia, gender differences were evident in job status, job mobility and academic achievements (Fox 2001; Børing et al. 2010; Hopkins et al. 2013), although some research has found no evidence to suggest discriminatory treatment of women over men for those in the same circumstances after controlling for structural, family and discipline variables (Ceci and Williams 2011). In the survey conducted as part of this thesis, men were more likely to have senior jobs in both Sciences and Humanities discipline areas. Since men have higher job status and more research experience, they would be more likely to have experience of funding applications and thus have learned about OA policy and experienced OA publishing. The difference by gender in awareness of and experience in using OA publishing may in part be explained by differences in job grade.

Respondents in the Medical and Life Sciences were more likely to have experience of Gold OA publishing. Respondents in Natural Sciences and Engineering were more likely to be aware of the open access repositories and have more experience of Green OA publishing. Academics in Social Sciences and Humanities were less likely to have experience of OA publishing compared to colleagues from Medical and Natural Sciences. Those in Arts and Humanities seemed to be the least experienced in both Gold and Green OA publishing. This indicates that OA publishing is not fully established in Arts and Humanities compared to other academic disciplines. This is backed up by a study by Darley et al. (2014) which found that journals in Humanities, particularly English and Modern languages, had very low levels of OA availability outside the UK. The lack of OA availability explains the low publishing rate in OA journals for those in Humanities disciplines. In Medical and Life Sciences, however, the Gold OA model was already well-developed by 2010 (Björk et al. 2010).

The Green model has been adopted to a greater extent in Natural Sciences, such Physics and Mathematics (ibid.). This was also backed up by a number of comments from respondents who indicated that they frequently used a preprint repository called ArXiv as a resource for searching and disseminating research articles. The difference in the use and history of resources may explain why academics in Natural Sciences are ahead of their colleagues from other discipline areas in terms of their knowledge and experience of Green OA publishing.

The vast majority (86%) of respondents acknowledged the importance for making research data available online for reuse. This proportion is slightly lower than those who rated making 110

research articles accessible online as important (93%). Among all respondents, 21% had deposited their own primary research data in online repositories and 29% had used secondary research data collected by others. Academics who had reused research data collected by other researchers were more likely to have shared their own primary research data online which suggests the effect of reciprocity.

The rationales of making data available online for reuse were similar to the rationales for making research articles freely accessible online, as publicly-funded research should be accessible to the public. Other reasons given by respondents included: improving quality, reducing instances of fraud, validating research findings, avoiding duplicate data collection, saving time and cost, maximising the use of resources and encouraging collaboration. However there were also challenges and barriers. The major barrier of sharing data is related to academic competition. Many respondents pointed out that they only wanted to share with colleagues privately and with whom they could produce research papers together to secure academic rewards. Other challenges included the time and effort required to produce detailed meta-data alongside the primary data, ethical and confidentiality issues, and the uncertainty of benefits and usefulness for sharing primary research data.

Disciplinary culture may influence the practice of using and sharing research data. Some disciplines may have a long history of a data sharing culture such as the genomics community with its data sharing policies which started in 1996 in Bermuda Principles (Rodriguez et al. 2009). In the Biomedical Sciences, journals often have data sharing policies especially those open access journals. Piwowar (2011) found that authors are more likely to share data if related articles were published in an OA journal or a journal with a stronger data sharing policy. Moreover, some disciplines may not produce any primary data, as suggested by a number of survey respondents. There is a lack of evidence about the nature of sharing primary research data in Humanities. Ethical issues for disciplines that involve human subjects may affect academics’ data sharing practice and limit those with human participants sharing research data because of confidentiality or anonymity reasons. Academics in Natural Sciences and Engineering would be less likely to involve human subjects in their studies and thus face fewer ethical issues in sharing data.

Men were more likely to have used secondary data and deposited their own primary data online. The possible explanation of the gender differences in using and sharing data could be related to women having more junior jobs and less experience of funding application and open access publishing. 111

It is notable that Piwowar and Chapman (2010) found that biomedical authors were more likely to practise data sharing when the first or last authors were highly experienced in their careers with high level of professional impact. This is consistent with the finding from this study that senior and more experienced academics were more likely to have shared their primary research data online. Older and senior academics would usually have more experience with funding applications and thus learn about funding councils’ data sharing policies.

These patterns of gender and grade differences in sharing data online are consistent with the experiences of allowing OA publishing of research articles.

The vast majority of respondents had not yet shared their ongoing research work on Twitter (84%), blogs (84%) or social networking sites (81%). The overall proportion of respondents who never used any of those social media tools to publish ongoing research was 70%. A key reason for this might be because the contribution of scholarly work on social media has not been recognised by the academic reward system. The predominant indicator of professional performance for researchers and the institutions that employs them has always been the publication of articles in journals and the relative prestige of the journals (Merton 1957; Schauder 1993; Correia and Teixeira 2005). Under the academic reward system, individual researchers’ career advancement and promotions are based on their professional performance in terms of the quality and quantity of publications (Kim 2011). Thus the majority of academics still view the traditional distribution channels as most important and have yet to adopt social media for sharing research work, which is in line with findings from a study by Procter et al. (2010a).

However, the advantages of sharing research work on social media were also acknowledged. A number of respondents who had experiences of blogging research updates commented that they would write updates about their research including detailed discussion about their work and any results. The suggested benefits included engaging agencies and potential beneficiaries, gaining quick feedback, preserving a record for reference and self-motivation.

Disciplinary differences were evident as respondents in the Social Sciences and Humanities were more likely to share ongoing research information on social media than those in Medical and Natural Sciences. This is in line with findings from a survey study of a range of international scholars by Nicholas and Rowlands (2011). Social Sciences and Humanities are less likely to have definitive findings or ‘results’ like in the medical and natural sciences. Thus academics in Social 112

Sciences and Humanities might be less likely to worry about intellectual property rights and publishing research updates online.

This study found that younger and junior academics had more experience of publishing research updates on social media. The age differences found in this study are in line with previous studies which found age being inversely associated to internet and other new media use (Dutton et al. 2005; Helsper and Eynon 2010; Nicholas and Rowlands 2011). However, other studies found no evidence that job grade or rank disproportionally affected academics’ use of new media tools (Priem et al. 2011; Procter et al. 2010b).

In contrast to the findings from studies by Shema et al. (2012) and Procter et al. (2010b) that the use of new technology is more easily accepted by men, this study found that women appeared to be slightly more likely to use Twitter to share research updates, although the difference between men and women was very small (5%).

In this chapter, the main method used to explore differences between different groups was using descriptive statistics and inferential statistics. In Chapter Six, statistical modelling will be used to explore the factors associated with academics’ practices in supporting open science. Background characteristics will be examined using logistic regression modelling to determine whether there are statistically significant differences between groups by gender, discipline or age, after taking account of other factors. The next chapter explores respondents’ attitudes towards and experiences of using social media services in research work.

113

5. Chapter Five Research Question D: To What Extent do Academics Support the Use of Social Media for Research?

Summary: This chapter presents the survey results of respondents’ experience of using various social media services for research. It then explores respondents’ experiences of using social media to promote publications and their views on the potential benefits and risks related to the use of social media. In addition, it examines whether there are differences between groups defined by gender, discipline, age, grade or research experience.

5.1. To what extent do academics use social media services in their research work?

5.1.1 Overall experience of using social media in research work Survey question 27 asked how often respondents used a series of various online services in their research work. Table 5.1 reports respondents’ use of eight social media services. Facebook and Twitter were the least popular social media services for research purposes. The vast majority of respondents (79%, 1,325 out of 1,677) reported having ‘never’ used Twitter and 84% (1,414 out of 1,675) had ‘never’ used Facebook in their research work. Research blogs seemed to be the most common online services used with 40% (663 out of 1,672) of respondents having used research blogs, although only 1% (11) indicated having ‘always’ and 6% (92) indicated having ‘often’ used research blogs whilst 34% (560) selected ‘sometimes’. Google Drive/Doc seemed to be the second most popular social media service with 35% (578 out of 1,670) having ‘sometimes’, ‘often’ or ‘always’ used them. Google Drive/Doc also had more respondents reporting ‘always’ (3%, 51) and ‘often’ (9%, 149), followed by Twitter with 2% (38) indicating ‘always’ and 6% (93) indicating ‘often’.

Table 5.1 Experience of using eight social media services for research 114

How often do you use the following online services in your RESEARCH work? Never Sometimes Often Always Total Twitter N 1325 221 93 38 1677 % 79% 13% 6% 2% 100% Facebook N 1414 200 46 15 1675 % 84% 12% 3% 1% 100% Research Blogs N 1009 560 92 11 1672 % 60% 34% 6% 1% 100% YouTube or other N 1173 435 57 8 1673 video sharing services % 70% 26% 3% 1% 100% Google Drive/Doc N 1092 378 149 51 1670 % 65% 23% 9% 3% 100% Linked.In N 1115 451 90 17 1673 % 67% 27% 5% 1% 100% Academia.edu N 1217 349 94 8 1668 % 73% 21% 6% 1% 100% Researchgate.net N 1222 351 83 9 1665 % 73% 21% 5% 1% 100%

Approximately 33% of respondents had joined LinkedIn and 30% had used YouTube or other video sharing services in their research work. Academia.edu and ResearchGate are similar social networking sites specifically designed for academic users. They have similar patterns of frequency of use with 73% of respondents indicating ‘never’ and 21% indicating ‘sometimes’ using these two social networking sites in research work.

Twitter Survey question 28 asked how often respondents conducted specific activities. In relation to Twitter, two specific activities were asked: (i) whether the respondents gathered research information on Twitter; and (ii) whether they posted updates of ongoing research on Twitter (as reported in the previous chapter in Section 4.3.1). The vast majority of respondents (85%, 1,415 out of 1,670) had ‘never’ used Twitter to gather research information in their research work. As shown in Table 5.2, the proportion of respondents (15%) who indicated they ‘sometimes’, ‘often’, or ‘always’ used Twitter to gather research information was similar to those who used Twitter to post updates of ongoing research. However, after cross-tabulating the two variables, the results showed that 29% of respondents who posted ongoing research updates on Twitter had ‘never’ used Twitter to gather research information. After combining the two Twitter activities together, in total 80% (1,335 out of 1,667) of respondents had never used Twitter to gather research information or post ongoing research updates. This is consistent with the result from Survey 115

question 27 as shown in Table 5.1, that 79% (1,325) of respondents had ‘never’ used Twitter in their research work.

Table 5.2 Experience of gathering research information and publishing ongoing research updates on Twitter

How often do you gather research How often do you post updates of information on Twitter? ongoing research on Twitter? N % N % Never 1415 85% 1401 84% Sometimes 176 11% 182 11% Often 54 3% 65 4% Always 25 1% 25 1% Total 1670 100% 1673 100%

Research blogs In relation to research blogs, three specific activities were examined with question 28: (i) whether the respondents read research blogs; (ii) whether they commented on other research blogs; and (iii) whether they posted updates of ongoing research on research blogs as reported in Section 4.3.1. As shown in Table 5.3, the majority of respondents (51%, 860 out of 1,677) reported having ‘sometimes’ read research blogs in their research work, while 8% (133) indicated they ‘often’ and 1% (22) indicated they ‘always’ did so. Approximately 39% (662) reported having ‘never’ read research blogs. This is not consistent with the responses to question 27 which asked how often respondents used research blogs in their research work as 60% answered ‘never’. When those respondents were asked in question 27 whether they had used research blogs, they might not have considered reading blogs as ‘using research blog’. However, many research institutions and academics journals have had their own blogs sites and many news channels would direct audience to various blogs posts. Thus the proportion of respondents reading research blogs was higher than that of using research blogs.

Table 5.3 Frequency of reading, commenting and posting updates on research blogs

How often do you How often do you comment How often do you post updates of read research blogs? on other research blogs? ongoing research on research blogs? N % N % N % Never 662 39% 1313 78% 1407 84% Sometimes 860 51% 341 20% 204 12% Often 133 8% 17 1% 40 2% Always 22 1% 3 0% 17 1% Total 1677 100% 1674 100% 1668 100% 116

Approximately 22% (361 out of 1,674) of respondents reported having commented on other research blogs compared to 16% (261 out of 1,668) having posted updates of ongoing research on research blogs. Whilst over half of the respondents (61%) had read research blogs, the vast majority of respondents neither commented on (78%) nor posted updates of ongoing research on research blogs (84%).

5.1.2 Differences in using social media in research work by academic discipline After running cross tabulations of relevant variables, significant associations were found between four discipline areas and the use of all eight social media services in research work (in all cases, p<0.001). In general, respondents in Humanities and Social Sciences were more likely to use Twitter, Facebook, research blogs, Academia.edu and YouTube or other video sharing services, but less likely to use LinkedIn and ResearchGate than those in Medical and Natural Sciences.

Twitter As shown in Figure 5.1, respondents in Social Sciences (28%, 125 out of 446) and Arts and Humanities (25%, 64 out of 258) were more likely to use Twitter in their research work compared to those in Medical and Life Sciences (17%, 98 out of 585) and Natural Sciences and Engineering (17%, 65 out of 383) (p<0.001). This pattern is similar to that for posting ongoing research updates on Twitter by discipline which was reported in Section 4.3.2. Respondents in Social Sciences (21%, 93 out of 445) and Arts and Humanities (20%, 50 out of 256) were also more likely to gather research information on Twitter compared to those in Medical and Life Sciences (12%, 67 out of 582) and Natural Sciences and Engineering (12%, 44 out of 381) (p<0.001).

Figure 5.1 Use of Twitter in research work by academic discipline

Arts & Humanities 75% Never Social Sciences 72% Sometimes Natural Sciences & Engineering 83% Often Medical & Life Sciences 83% Always

0% 20% 40% 60% 80% 100%

Research blogs 117

Over half of the respondents in Arts and Humanities (56%, 145 out of 257) reported having used research blogs in their research work, followed by 45% (199 out of 444) of those in Social Sciences and 38% (146 out of 382) of those in Natural Sciences and Engineering (p<0.001). Respondents in Medical and Life Sciences (29%, 171 out of 584) were the least likely to use research blogs, leaving a large 27 percentage point difference with respondents in Arts and Humanities. This pattern is similar to that found in relation to posting ongoing research updates on research blogs by academic discipline, which were reported in the previous chapter.

Figure 5.2 Use of research blogs in research work by academic discipline

Arts & Humanities 44% Never Social Sciences 55% Sometimes Natural Sciences & Engineering 62% Often Medical & Life Sciences 71% Always

0% 20% 40% 60% 80% 100%

Respondents in Arts and Humanities were also most likely to read research blogs with 73% (187 out of 257) of users doing so, compared to 65% in Social Sciences (290 out of 446), 58% in Natural Sciences and Engineering (220 out of 382) and 54% in Medical and Life Sciences (314 out of 586) (p<0.001). Respondents in Arts and Humanities were also most likely to comment on other research blogs than those in the other three discipline areas.

Facebook, Google Drive/Doc, YouTube, LinkedIn, Academia.edu and ResearchGate The data on use of all social media services in research work were recoded to create a binary variable (0, 1) with 0 representing ‘never’ and 1 representing ‘sometimes’, ‘often’ or ‘always’. After running a cross tabulation of the binary variable of social media use against discipline area, the patterns of using Facebook, YouTube or other video sharing services and Academia.edu in research work were found to be similar to the use of research blogs, with an increase in use from Medical and Natural Sciences to Social Sciences and Humanities. Table 5.4 reports the proportions of use for the six social media services by academic discipline and all associations are significant. Contradicting these patterns, the use of ResearchGate and LinkedIn were higher for those in Medical and Natural Sciences. There was a 15 percentage point difference between respondents from Medical and Life Sciences (37%) and those from Arts and Humanities (22%) for the use of LinkedIn. 118

Table 5.4 Use of six social media services in research work by academic discipline

Facebook Google YouTube/other Academia Research LinkedIn Drive/Doc video sharing .edu gate.net services Medical & Life Sciences 9% 26% 22% 14% 42% 37% Natural Sciences & Engineering 10% 41% 28% 17% 23% 36% Social Sciences 22% 38% 35% 39% 19% 33% Arts & Humanities 29% 38% 41% 50% 9% 22% Academia.edu and ResearchGate are similar academic social networking sites in terms of their features however the findings indicate that there are different levels of use across different disciplines. Half of respondents in Arts and Humanities (50%, 129 out of 257) and 39% (175 out of 444) in Social Sciences reported having used Academia.edu compared to 14% (83 out of 581) in Medical and Life Sciences and 17% (64 out of 381) in Natural Sciences and Engineering. In contrast, 42% (244 out of 583) of respondents in Medical and Life Sciences indicated using ResearchGate compared to only 9% (24 out of 255) in Arts and Humanities. There was a 33 percentage point difference between those from Medical and Life Sciences and those from Arts and Humanities in their use of ResearchGate.

Respondents from Natural Sciences and Engineering (41%) were most likely to use Google Drive/Doc, followed by those in Social Sciences (38%), Arts and Humanities (38%) and Medical and Life Sciences (26%).

5.1.3 Differences in using social media in research work by gender Among the eight social media services, only the use of Twitter and Facebook in research work showed significant associations with gender (p<0.001 for both associations). Women appeared to be more likely to use Twitter and Facebook in their research work. For female respondents, 25% (192 out of 758) used Twitter and 19% (145 out of 758) used Facebook in their research work compared to men, among whom 17% (156 out of 905) used Twitter and 13% used Facebook (115 out of 903).

5.1.4 Differences in using social media in research work by age, job grade and research experience In general, respondents’ experience of using these eight social media services in their research – excluding ResearchGate and LinkedIn – was lower amongst older academics, more senior academics and those with more research experience. 119

In terms of age, respondents under 35 were most likely and those 55 and over were least likely to have used Twitter, research blogs, Facebook, Google Drive/Doc, Academia.edu, or YouTube or other video sharing services in their research work. However, respondents aged 45-54 (32%, 120 out of 370) were most likely to use ResearchGate and those aged under 35 (23%, 128 out of 553) were least likely. Respondents aged 35-44 (38%, 166 out of 438) were most likely to use LinkedIn, followed by those aged 45-54 (35%, 129 out of 370), under 35 (31%, 174 out of 553) and 55 and over (28%, 86 out of 303). The associations between age group and the use of all eight social media services were significant.

In relation to research experience, respondents with 6-10 years’ research experience were most likely to use Twitter, research blogs, Facebook, Google Drive/Doc, Academia.edu, or YouTube or other video sharing services in their research work, whilst those with over 21 years’ research experience were the least likely. However, respondents with 11-20 years’ research experience (32%, 139 out of 436) were most likely to use ResearchGate and those with 1-5 years’ research experience (20%, 79 out of 387) were least likely. The associations between research experience and the use of all eight social media services, except LinkedIn, were significant.

In relation to job grade, lecturers/research fellows/postdocs were most likely to use all eight social media services reported here except YouTube or other video sharing services. Compared to academics in senior job grades, researchers in training were most likely to use YouTube or other video sharing services but the least likely to use ResearchGate and LinkedIn. Senior lecturers/senior researchers were more likely to use ResearchGate and LinkedIn. Professors/readers were less likely to use all social media services reported here except ResearchGate. The associations between job grade and the use of all eight social media services were significant.

5.2 To what extent do academics use social media to promote their publications?

5.2.1 Overall experience of using Twitter, blogs and social networking sites to promote publications Respondents were asked, using a multiple choice question if they did anything to promote their most recent peer-reviewed publication. The question listed Twitter, research blog and 120

Facebook or other social networking sites as answer choices for promoting their publications. Many respondents (262) left comments for this question. Some listed various social media services they used to promote their recent publications instead of choosing an answer to the survey question, some chose answers and also gave comments. The most common comments stated that the respondents posted the articles and links on their own websites or institutional websites. Thirty-four respondents commented about promoting their most recent publications at conferences, 23 respondents noted press release, 15 respondents listed Academia.edu, 13 respondents noted LinkedIn, 12 respondents noted ResearchGate and 9 respondents promoted through workshops.

Some of the comments were coded into the existing categories based on the contents. For example, if the comments mentioned ResearchGate and LinkedIn, they would be coded into the category of ‘Facebook or other social networking sites’. If the comments were academics’ own websites or institutional websites, they would be coded into the categories of blogs/websites. If they answered ‘no’, but commented that they had indeed promoted their publications, their response would be coded into the existing categories. The responses of ‘I haven't published anything yet’ were filtered out.

As shown in Table 5.5, of the 1,545 respondents who had published research articles, 15% (234) posted the article titles and links on Facebook or other social networking sites, 12% (180) posted on Twitter and 15% (225) posted on research blogs or their websites. Over half of respondents (55%, 856) told colleagues in person about their recent publications and 32% (488) promoted their publications through emails either by emailing the papers to colleagues or adding the web link of the articles in their email signatures. Around 29% (450) of respondents answered ‘no’ regarding doing any promotion for their recent peer-reviewed publications.

Table 5.5 Experience of promoting recent peer-reviewed publication (n=1,545 as a subgroup of respondents who had published research articles)

Regarding your most RECENT peer-reviewed publication, did you do anything to promote it? (Tick all that apply) N % No 450 29% Yes, I told colleagues in person 856 55% Yes, I emailed the paper to colleagues/ added the web link in my email signature. 488 32% Yes, I used Twitter to post the web link to the online article. 180 12% Yes, I posted the article title and link in my research blog/website. 225 15% Yes, I posted the article title and link on Facebook or other social networking sites. 234 15% Total 1545 100% 121

After regrouping the six multiple choices categories into four single choice as shown in Table 5.6, 32% (487 out of 1,545) of respondents who had published articles promoted their most recent peer-reviewed publication on social media (including Twitter, blogs/ websites, and Facebook or other social networking sites), 19% promoted through emails but not on the social media services listed here, 20% only told colleagues in person and another 29% did nothing to promote their recent publications.

Table 5.6 Experience of promoting recent peer-reviewed publication regrouped (n=1,545 as a subgroup of respondents who had published research articles)

N % No 450 29% Yes, I promoted on social media 487 32% Not on social media, but I promoted through emails 297 19% I promoted in person only 311 20% 1545 100% Total 5.2.2 Differences in using social media to promote publications by academic discipline Cross tabulations of publication promotion by four discipline areas were run using the original answers. Those with the answer ‘I haven’t published anything yet’ were filtered out as missing. The associations between academic discipline and the experience of promoting recent peer reviewed publications on three social media services were all significant.

As shown in Table 5.7, the results show similar patterns to the use of social media services in research work as discussed in Sections 5.1.2. In general, respondents in Arts and Humanities were more likely to promote their recent publications on social media services whilst those in Medical and Life Sciences were less likely to do so. Respondents in Arts and Humanities were more than twice as likely to post their article title and link on research blogs or social networking sites compared to those in Medical and Life Sciences.

Table 5.7 Experience of promoting recent peer-reviewed publications on social media by academic discipline (n=1,486 as a subgroup of respondents who had published research articles)

Twitter Research blogs Facebook or other SNs Medical & Life Sciences 10% 6% 11% Natural Sciences & Engineering 9% 11% 12% Social Sciences 16% 11% 12% Arts & Humanities 15% 12% 21% 122

When comparing the proportion of those who answered ‘no’ to any sort of promotion of their publications including personal and email promotion, respondents in Medical and Life Sciences (35%, 187 out of 532) were more likely not to do any promotion, compared to 31% (105 out of 336) of those in Natural Sciences and Engineering, 27% (105 out of 394) of those in Social Sciences and 25% (55 out 224) of those in Arts and Humanities.

5.2.3 Differences in using social media to promote publications by gender Amongst a subgroup of respondents who had published research articles, men and women were similar in their use of social media to promote publications. Women (15%) appeared to be slightly more likely to use Facebook or other social networking sites to promote their recent publications than men (11%) (p<0.05). Men (11%) were slightly more likely to use research blogs to post article title and links than women (7%) (p<0.01). The association between gender and the use of Twitter for promoting publications was not significant.

Cross tabulations were also run between gender and the regrouped four categories as shown in Table 5.6. Women appeared to be slightly more likely to promote their recent publications on social media services as a whole (34%, 228 out of 665) than men (29%, 254 out of 865) (p<0.05).

5.2.4 Differences in using social media to promote publications by age, job grade and research experience In general, respondents’ experiences of using Twitter, research blogs or social networking sites to promote publications decreased with age, job grade and research experience. Amongst a subgroup of respondents who had published research articles, those who were younger, less experienced and in junior grades were more likely to promote their most recent peer-reviewed publications on social media. Respondents who were older, more experienced and in senior grades were more likely not to do anything to promote their recent publications15.

In terms of age, those aged 55 and over (2%, 5 out of 298) were much less likely to use Twitter to promote publications compared to those under 35 (17%, 77 out of 448), those aged 35-44 (15%, 59 out of 390) and those aged 45-54 (10%, 36 out of 346). Respondents aged 55 and over (3%) were least likely to adopt research blogs to promote publications compared to the other three age groups (all at 11%). Respondents aged under 35 (18%, 80 out of 448) were more likely to use

15 P<0.05 for all associations reported in this section except the association between grouped grade and promotion on research blogs (p=0.074). 123

Facebook or other social networking sites to promote publications compared to those aged 35-44 (16%, 61 out of 390), 45-54 (8%, 28 out of 346) or 55 and over (6%, 19 out of 298). The likelihood of not doing anything to promote recent publications also increased with age from 23% for those under 35 to 39% for those 55 and over.

In relation to research experience, respondents with over 21 years research experience were much less likely to use Twitter, research blogs or social networking sites to promote publications. Respondents with 6-10 years’ research experience were more likely to use research blogs and Facebook or other social networking sites, while those with 1-5 years’ experience were more likely to use Twitter to promote publications. Those with over 21 years’ research experience appeared to be most likely to not do anything to promote their recent publications.

In relation to job grade, researchers in training and lecturers/research fellows/postdocs were most likely to use Twitter and Facebook or other social networking to promote publications. Professors/readers (36%, 137 out of 379) and senior lecturers/senior researchers (36%, 92 out of 255) were more likely to not do anything to promote their recent publications compared to 31% (57 out of 183) of researchers in training and 25% (153 out of 615) of lecturers/research fellows/postdocs.

5.3 Academics’ views on the benefits and risks of using social media

5.3.1 Overall attitudes towards the benefits and risks related to the use of social media

Attitudes Survey question 42 asked to what extent respondents agreed or disagreed with a number of statements regarding using social media in their research work and also whether research published on social media could be trusted. Seven of the statements were related to the potential positive effects of using social media in research work and four statements were related to the potential negative effects.

As shown in Figure 5.3, in relation to the benefits for the public good, over half of all respondents (54%, 885 out of 1,627) agreed that communicating research on social media would 124

benefit the public and 30% (479 out of 1,626) agreed that communicating research on social media would accelerate scientific discovery.

Regarding personal-level benefits, 45% (723 out of 1,625) of respondents agreed that using social media would promote their professional profiles. Approximately 40% (642 out of 1,627) of respondents agreed that using social media would help them find collaboration opportunities and 30% (491 out of 1,617) agreed that using social media would benefit their careers. Citations are closely related to academic progression and are regarded as very important for developing an academic career. In this study, 31% (499 out of 1,631) of respondents agreed that blogging or tweeting about their publications would increase their citations. There was less agreement with the idea that using social media could increase the chances of getting funding – 14% (234) agreed and 39% (633 out of 1,624) disagreed with this.

Figure 5.3 Attitudes towards the positive effects of using social media in research work

Communicating research on SM benefits the public. strongly agree Using SM promotes my professional profile. Using SM helps me find collaboration agree opportunities. Blogging or tweeting about my publication will increase citations. neither disagree nor agree Using SM benefits my career. disagree Communicating research on SM accelerates scientific discovery. Using SM increases my chances of strongly getting funding. disagree

0% 50% 100%

As shown in Figure 5.4, in relation to the possible negative effects, over half of respondents (58%, 949 out of 1,625) agreed that research published on social media could not be trusted because of concerns about it not being peer reviewed. Approximately 44% (722 out of 1,625) agreed that communicating research on social media might result in plagiarism and 30% (487 out of 1,627) agreed that communicating research on social media might leak results to competitors. Another 30% (492 out of 1,625) agreed that communicating research on social media could result in the risk of good ideas being stolen. 125

Figure 5.4 Attitudes towards the negative effects of using social media in research work

Research published on SM cannot be strongly agree trusted as not being peer-reviewed.

Communicating research on SM may agree result in plagiarism. neither disagree Communicating research on SM risks nor agree my good ideas being stolen. disagree Communicating research on SM may leak results to competitors. strongly disagree 0% 50% 100%

Understanding attitudes towards the effects of using social media Over 40% of respondents chose ‘neither disagree nor agree’ for statements such as ‘using social media increases my chances of getting funding’, ‘blogging or tweeting about my publication will increase citations’ and ‘communicating research on social media may leak results to competitors’. A number of respondents indicated that they did not have enough knowledge of or experience in using social media to have an opinion about these statements. One senior lecturer suggested that the answers to these statements depended on how and what social media was used:

‘Difficult to respond to the statements in Question 39 - it very much depends how social media is used e.g. some social media sites can't be trusted whereas others can so it's a case by case situation for me rather than a clear-cut yes or no.’ (Senior Lecturer, Music, Drama, Dance and Performing Arts)

In relation to the potential risks of sharing research information on social media, some respondents agreed with communicating published research but were sceptical about publishing research on social media that had not been peer reviewed. A senior lecturer in Earth Systems and Environmental Sciences pointed out the difference between communicating published papers and non-peer reviewed contents:

‘Presumably timing is crucial. Social media can be hugely helpful, but cannot replace the scrutiny of the peer review process. Most the questions above would be “agree” or “strongly agree” if the social communication came after the paper had been accepted for publication. They would be the opposite if the social communication replaced peer-review.’ (Male, Senior Lecturer, Earth Systems and Environmental Sciences)

126

There were a number of negative comments in relation to academics using or disseminating non-peer reviewed content on social media. For example, one respondent suggested that using non-peer reviewed content is problematic and another implied that those who disseminated work on social media rather than through the traditional publication route were not trustworthy:

‘Anyone using sources that lack peer reviewed content is a fool.’(Male, Reader, Biological Sciences)

‘I think scientists who use social media a lot are likely to be distrusted. We have been taught to allow our work speak for itself.’ (Male, Professor, Psychology, Psychiatry and Neuroscience)

Several respondents suggested strategies to avoid risks. One academic from Computer Science and Informatics suggested that plagiarism and good ideas being stolen could be avoided by careful exposure of research information, such as giving clues but not exact algorithms, in order to avoid giving the game away:

‘Ideas being stolen depend on what level of detail you describe your research in. I wouldn't post exact algorithms before publication, but might give a clue to the sorts of things that we are up to.’ (Male, Research Fellow/postdoc, Computer Science and Informatics)

Another similar comment from an academic in Business and Management Studies suggested being very selective of what to communicate on social media to minimise risks:

‘I have "agreed" to risks as e.g. plagiarism could occur if social media were used injudiciously but I'm sure you could work round this by judicious usage.’ (Female, Research Fellow/postdoc, Business and Management Studies)

Positive views were also suggested by early adopters who experienced and acknowledged the benefits of communicating published research. One professor emphasised the citation impact of self-promoting publication on social media and pointed out the possibility of colleagues missing out on the opportunity for increasing their citations if they chose not to:

‘…and on social media, most of my colleagues make me laugh because they are so clueless. Clueless first about research dissemination - they think their work is over when they hand the final proofs to the publisher when, in fact, that's when the real work begins: do you think I got those 8,000 citations from being an academic genius? No, from relentless online promotion (though of course you have to have a product a good-enough quality for promotion to work). And second, clueless about social media: they think it's some dumb thing for their kids. Long may they remain clueless - makes it easier for me to 127

raise my profile compared to them!’ (Male, professor, Anthropology and Development Studies)

5.3.2 Differences in attitudes towards using social media by academic discipline The original eleven questions about attitudes towards the benefits and risks related to the use of social media were recoded from five categories (strongly agree, agree, neither disagree nor agree, disagree and strongly disagree) to three categories (agree or strongly agree, neither disagree nor agree, and disagree or strongly disagree) in order to make the cross tabulations between different groups more clear and meaningful.

In general, respondents in the Medical and Life Sciences have more negative views towards using social media for research. The associations were statistically significant between academic discipline and seven statements related to the positive effects of using social media. Those in the Medical and Life Sciences were more likely to disagree and less likely to agree with the statements related to the positive effects of using social media in research work. On the other hand, respondents in Arts and Humanities and Social Sciences appeared to be more likely to agree with the positive effects of using social media for research. For example, 37% (160 out of 437) of respondents in Social Sciences and 36% (89 out of 247) in Arts and Humanities agreed that blogging or tweeting about their publications would increase citations, compared to 29% (108 out of 375) of those in Natural Sciences and Engineering and 25% (140 out of 565) in Medical and Life Sciences.

The only association between academic discipline and the four negative effects of using social media that was significant was for the statement ‘research published on social media cannot be trusted as not being peer-reviewed’. Respondents in Natural Sciences and Engineering (65%, 242 out of 373) and Medical and Life Sciences (61%, 342 out of 564) were more likely to agree with this statement compared to 53% (130 out of 246) in Arts and Humanities and 53% (232 out of 435) in Social Sciences.

5.3.3 Differences in attitudes towards using social media by gender In general, women were more likely to agree with statements related to the positive effects of using social media in their research work. The associations were statistically significant between gender and all seven statements related to the positive effects of using social media. The gaps between the percentage of men and women who agreed with the positive statements were 128

between 5-10%. Men were more likely to disagree with the positive statements. For example, 50% (366 out of 734) of women agreed that ‘using social media promotes my professional profile’ compared to 40% (352 out of 877) of men, whilst 23% (205) of men disagreed with this statement compared to 15% (109) of women.

Overall the survey results indicate no difference between men and women in terms of attitudes towards the potential negative effects of using social media in their research work.

5.3.4 Differences in attitudes towards using social media by age, job grade and research experience In general, respondents who were younger, less experienced and in junior job grades were more likely to have positive attitudes towards using social media in their research work. Respondents who were older, more experienced and in senior job grades appeared to be more likely to disagree with the statements related to the potential positive effects of using social media. The associations were statistically significant between age, job grade and research experience with all seven statements related to the positive effects of using social media.

In relation to age, respondents aged 55 and over were least likely to agree with all seven statements related to the positive effects whilst those aged under 35 were most likely to agree with the positive statements. For example, 15% (43 out of 292) of respondents aged 55 and over agreed that using social media helps them find collaboration opportunities compared to 31% (111 out of 362) of those aged 45-54, 44% (188 out of 425) of those aged 35-44 and 55% (296 out of 539) of those under 35. In relation to research experience and job grade, respondents with over 21 years’ experience and professors/readers were much less likely to agree with all seven statements related to the positive effects and most likely to disagree with the statements related to the positive effects of using social media.

Regarding the questions related to the potential negative effects of using social media, respondents who were younger, less experienced and in junior grades were more likely to agree that ‘communicating research on social media may leak results to competitors’ and ‘communicating research on social media risks my good ideas being stolen’. The associations were statistically significant between age, job grade and research experience with these two statements. For example, 39% (114 out of 289) researchers in training agreed that communicating research on social media may leak results to competitors compared to 31% (192 out of 628) of lecturers/research fellows/postdocs, 27% (66 out of 246) of senior lecturers/senior researchers 129

and 23% (90 out of 391) of professors/readers. This suggests that junior researchers are more concerned with protecting their research outputs and securing academic rewards.

On the other hands, respondents who were older, more experienced and in senior grades were more likely to agree that ‘research published on social media cannot be trusted as not being peer-reviewed’ and ‘communicating research on social media may result in plagiarism’. The associations were statistically significant between age and research experience with these two statements. Respondents with over 21 years’ research experience (69%, 307 out of 446) were more likely to agree that research published on social media could not be trusted as not being peer-reviewed compared to those with 1-5 years’ research experience (51%, 192 out of 379), those with 6-10 years’ research experience (57%, 206 out of 363) and those with 11-20 years’ research experience (57%, 239 out of 422). This suggests that senior researchers are more concerned with the credibility of content on social media.

5.4 Conclusion and discussion

This chapter examined survey respondents’ experiences of using various social media services in research work. It also discussed respondents’ experiences of using social media to promote publications and their views on the potential benefits and risks related to the use of social media. It explored whether there were differences between groups defined by gender, discipline, age, job grade and research experience.

The evidence suggests that the majority of respondents had never used Twitter (79%), Facebook (84%), research blogs (60%), YouTube or other video sharing services (70%), LinkedIn (67%) or Academia.edu (73%) in their research work. However, compared to the findings from a 2009 study by Procter et al. (2010a; 2010b), academic users of Twitter, LinkedIn and Academia.edu have grown while YouTube and research blogs have similar percentages of academic users. The evidence also suggests that 32% of respondents who had published articles had promoted their recent peer reviewed publications on social media sites including Twitter, research blogs and social networking sites.

The use of Facebook in research decreased from 24% in 2009 to 16% in 2013. The use of Twitter in research increased from 10% to 21% and LinkedIn increased from 18% to 33% while Academia.edu increased from 9% to 27%. LinkedIn was founded in 2003 to connect professionals but was not adopted by academics for research purposes until recent years. LinkedIn had 130

75,000,000 members while Facebook already had 400,000,000 members by 2010 (Crush et al. 2012). Since Facebook was already very popular in 2009, it was possible that some academics left Facebook after 2009 or only used Facebook for private functions as suggested by participants from pilot studies. Academia.edu, launched in 2008, was originally chiefly concerned with setting up a global directory of universities and research institutes but later on added the social networking function (Nentwich 2010). It was reasonable that Procter et al.’s study found 9% of respondents using Academia.edu in 2009 and this thesis finds 27% in 2013. Twitter was first launched in October 2006 (Honeycutt and Herring 2009) and did not gain popularity in academia until 2009 as most research articles of Twitter’s academic use were from 2010 onwards. Twitter was first used experimentally in academic conferences to support communication (Ebner et al. 2010). Over the past four years, Twitter has been often used in various academic conferences with specific hashtags to facilitate communication between participants and distribute web links to presentation slides and published papers. This explains why the proportion of academic Twitter users doubled from 2009 to 2013 as reflected by the two survey studies.

When asked to what extent respondents agreed or disagreed with the potential positive and negative effects of using social media, over half of all respondents (54%) agreed that communicating research on social media would benefit the public. However, 58% agreed that research published on social media could not be trusted as it would not have been being peer reviewed. A large proportion of respondents (between 30-50%) felt that they lacked enough knowledge or experience to have a clear opinion in relation to the benefits and risks of using social media for research. The lack of knowledge over the precise benefits of using social media remains an important barrier for social media adoption (Nicholas and Rowlands 2011). Comments from respondents suggest that communicating early-stage ideas might bring certain risks but communicating published work can promote researchers’ profiles and may increase citations. If done carefully, many believed that social media could bring about great benefits. As previous studies found that disseminating information about publications on social media might increases citations (Priem and Costello 2010; Eyenbach 2011), academics who have yet to adopt social media to promote published papers face the possibility of missing out the opportunities for increasing their citations.

Procter et al.’s (2010b) findings suggested disciplinary differences in Web 2.0 adoption, as Computer Science and Mathematics had a higher percentage of frequent users than did the Medical and Life Sciences. In this study, there were also clear differences between different academic disciplines. In general, respondents in Arts and Humanities and Social Sciences 131

disciplines were more likely to use social media services including Twitter, Facebook, research blogs, Academia.edu and YouTube or other video sharing services and have positive views towards the effects of using social media. Those in Medical and Life Sciences were least likely to use social media in research, but were more likely to have negative views on the effects of using social media. However, respondents in Medical and Natural Sciences disciplines were more likely to use ResearchGate and LinkedIn compared to those from Humanities and Social Sciences. Academia.edu and ResearchGate are similar academic social networking sites in terms of their features and the findings indicated different disciplinary focuses. In line with findings from Gross and Suttor (2013), while Academia.edu users were across all disciplines, ResearchGate had strengths in Medical and Natural Sciences. This is probably because ResearchGate have targeted communities of users from specific Sciences (Gewin 2010) and Academia.edu tends to be more heavily used by academics in the Social Sciences and Humanities since it was first founded by a philosopher from Oxford University (Thelwall and Kousha 2014a).

Respondents in Arts and Humanities were most likely to promote their recent publications on social media services whilst those in Medical and Life Sciences were least likely to do so. Respondents in Arts and Humanities were more than twice as likely to post their article title and link on social networking sites compared to those in Medical and Life Sciences. Respondents in Arts and Humanities were much more likely to rate it ‘very important’ (81%, 211 out of 262) for academic book or book chapter to be the dissemination means for communicating research compared to 16% (98 out of 601) of those in Medical and Life Sciences and 20% (77 out of 380) in Natural Sciences and Engineering. Since Facebook and other social networking sites can be used to market and sell books (Shih 2010), promoting academic books on Facebook may bring benefits beyond just citations. Another cultural difference between the Sciences and Humanities is the expectations of publications. In general, academics in Medical and Natural Sciences publish more papers than those in Social Sciences and Humanities. This study found that the mean publication numbers of those in Medical and Life Sciences and Natural Sciences and Engineering (16 and 19) were almost twice as the means for those in Social Sciences and Humanities (10 and 8). It would have required more effort for someone to publish four articles a year than for someone to publish one book a year to promote each published output on social media.

There were also gender differences in experiences and attitudes towards social media use. Women appeared to be more likely to use Twitter and Facebook in their research work. Since Twitter and Facebook require users to interact with others in order to gain followers/friends and exchange information, this finding suggests women a higher capability for building social 132

networks and connections (Li et al. 2005). Women were also more likely to agree with the potential positive effects of communicating research on social media, although this could be because women tend to rate more positively (Attwood 2012).

In general, younger, less experienced and junior academics were more likely to use social media services in research work and more likely to promote publications. Since research experience and job grade positively correlate with age, the use of social networking sites such as Facebook may be because of age differences rather than seniority differences. Early stage researchers in their 20s and 30s would have been the Facebook generation who embraced the site since its launch in 2004. Age differences found in this study are consistent with previous studies of academics’ social media use (Nicholas and Rowlands 2011).

Older and senior respondents were less likely to agree with the positive effects of using social media in research work. According to the pattern of answers to the negative effects of using social media, junior researchers appeared to be more concerned about protecting their research outputs and securing academic rewards. Senior academics might not be so concerned with securing academic rewards since they are more established in their profession. However, senior researchers seemed to be more sceptical about the trustworthiness of contents on the social web, which was also suggested by a study by Maron and Smith (2008).

The next chapter will use statistical modelling to explore the factors associated with academics’ use of social media. Logistic regression modelling will investigate whether there are still differences in terms of academic discipline, gender and age after taking account of the attitudes towards the potential effects of using social media and other factors. Chapter Six will also explore the factors associated with the three aspects of practising of open science.

133

6. Chapter Six Research Question E: Understanding Factors Associated with Social Media Use for Research and the Support for Open Science

Summary: This chapter explores the factors associated with the experience of using social media for research and practising open science using statistical modelling. It first explores what factors are associated with the likelihood of using social media services in research. Next the chapter examines what factors are associated with the likelihood of using each of the three aspects of open science: open access publishing, sharing primary research data and publishing ongoing research updates online. The background characteristics discussed in Chapters Four and Five will be examined in logistic regression models to determine whether there are statistically significant differences between groups by gender, discipline or age after taking account of other factors.

6.1 Factors associated with the likelihood of using social media for research

6.1.1 Introduction and hypotheses Based on the findings from the pilot studies, the previous analysis of the survey data and also existing literature, personal, social and technical factors were considered for exploratory analysis of using social media in research work.

Logistic regression models were fitted to explore what factors are associated with the likelihood of academics using social media in research work. The findings in Chapter Five suggested that there were differences in using social media for research in terms of gender, discipline, age, job grade and research experience. Thus these background characteristics were considered as independent variables to explore whether there were statistically significant differences after taking account of personal, social and technical factors.

However, research experience, age and job grade are highly correlated with one another. For example, the Spearman’s correlation coefficient between research experience and the original variables for age group is 0.8 and it is 0.7 between research experience and job grade. When two predictors are correlated at more than 0.7, the effects of the two predictors cannot be separated and this condition is called ‘multicollinearity’ (Weisberg and Bowen 1977: 214). To avoid 134

multicollinearity, only one of these three variables – namely age – was included in as an independent variable in the logistic regression analysis16.

Personal, social and technical factors were considered in relation to the likelihood of using social media in research work and four hypotheses are proposed. Hypothesis 1 considers technical factors and Hypothesis 2 includes social factors. Hypotheses 3 and 4 take account of personal views on dissemination audience and the effects of using social media. Logistic regression models were fitted to specify whether the factors are associated with social media use in research work after controlling for key background characteristics. These hypotheses are discussed in more detail below.

Hypothesis 1: The use of digital device, including personal laptops, smartphones and tablet computers, is positively associated with the likelihood of using social media in research work.

New technology has changed the way people access the Internet. Laptops, smartphones and tablet computers liberated the physical restriction of using a PC in a specific location to access the Internet. Social media sites can be accessed using smartphones and tablet computers. Studies have found that in the United States, smartphone ownership increased from 35% in 2011 to 56% in 2013 (Smith 2013). As of January 2014, 42% of American adults owned a tablet computer (PewResearch 2014). In the UK, the tablet user penetration rate was reported to have increased from 22.3% in 2012 to 34.5% in 2013 (Statista 2014). According to Deloitte, 7 in 10 UK adults aged 16-64 owned a smartphone by 2013 (Styles 2013). A study with academic social media users of the #phdchat network on Twitter found that over one quarter of participants were using mobile devices to publish tweets (Ford et al. 2014). Nicholas and Rowlands (2011) found that professional users of social media are 1.68 times more likely to use a smartphone or other mobile device than non-users; and 2.11 times more likely to use an iPad. This is consistent with the previous key findings. Hence, personal laptops, smartphones and tablet computers may have influenced academics’ use of social media. For example, an academic with a 3G smartphone is able to access Twitter feeds and send messages from anywhere and at any time, including using timed releases. Social media enthusiasts may be inclined to purchase relevant mobile devices to facilitate ease of accesses to new media applications.

16 The author ran logistic regression models with research experience, age and job grade as independent variables and the same outcome variable which will be explained in 6.1.2. When research experience, age and job grade were used separately as an independent variable in the model, each showed a significant association with the outcome variable. However, when all three were included in the model as independent variables, only age showed a significant association with the outcome variable. 135

Hypothesis 2: Encouragement from one’s institution and colleagues is positively associated with the likelihood of using social media in research work.

During the scoping study interviews, a number of interviewees suggested that they started to use social media because their supervisors or colleagues recommended them to. As reported in Section 3.3.4, non-users suggested that peers’ recommendations and their positive experiences would be great encouragement for them to use social media in the future. Of course, social network effects are a key part of behaviour change. A study found that encouragement from friends or acquaintances has a strong positive impact on senior citizen’s Internet use in Switzerland (Friemel 2014). As Friedman et al. (2000) noted, people trust people, not technology. E-commerce companies such as Amazon and EBay have long adopted the reputation system to help online consumers build trust in products by viewing other consumers’ likes and feedback (Schafer et al. 1999). To make one trust social media enough to use it for research, recommendations from one’s institution and colleagues can be very influential. Universities may promote social media use though newsletters, workshops, providing resources and funding for setting up research blogs, or providing social media training to research staff and research students.

Hypothesis 3: Consideration for dissemination of research findings to the public is positively associated with the likelihood of using social media in research work.

Open science suggests that publicly-funded research literature should be available online without charge. This has been supported by many academics as well as the UK research councils and major funding bodies. In order to disseminate research outputs to the general public with free access, academics can publish in one of the open access channels. Moreover, the new media tools are able to assist academics to post links to their published articles or summaries of findings on, for example, Twitter, Facebook or LinkedIn. Academics are also able to upload working papers to repositories such as Academia.edu or discuss their research in progress on social media sites. A number of interviewees and survey respondents who were not using social media for their research indicated that their research was very specific and only a small number of academics in their fields would understand their research findings. This view fits into the dominant discourse of scientific popularisation which regards the public as ignorant and unfit (Hilgartner 1990). Thus academics that have reservations about communicating their research to the general public may be less likely to use social media to communicate research. Besley (2014) found that scientists’ 136

belief in the need for them to contribute to public debates predict their actual online engagement of science communication in the past two years. Hence, those who consider that the public should know about their research findings may be more likely to use social media for research.

Hypothesis 4: Positive views on social media are positively associated with the likelihood of using social media in research work.

Previous studies suggested various benefits of and barriers to using social media for research (e.g. Gu and Widén-Wulff 2011; Powell et al. 2011). Interviewees from the scoping studies discussed their personal views towards using social media from positive and/or negative perspectives. Thus in the survey, respondents were asked to what extent they agreed or disagreed with eleven statements related to the potential effects of using social media including seven benefits and four risks as discussed in Chapter Five. Chapter Five found that younger academics and those in the Humanities and Social Sciences disciplines were more likely to agree with the benefits and more likely to use social media in research. Thus it is highly likely that positive attitudes would be associated with greater adoption. It is important to control attitudes variables because if positive attitudes towards social media ‘perfectly’ define academic social media use, then other factors will not have any effects on social media use. On the other hand, academics might have a positive view on new media but lack the required skills or might not have time to develop skills to be able to use social media tools or have other concerns such as plagiarism or intellectual property rights. Thus it is useful to include the variables of personal views on the benefits and risks as independent variables to test their significance.

6.1.2 Dependent variable Use of social media tools in research work (0,1). Twitter, Facebook and research blogs are the three key social media tools which have been discussed in previous chapters and which will be combined here to examine the likelihood of a respondent using one or more of the three tools for research. The use of these three tools was reported in Section 5.1.1. The outcome variable was coded as either 1 (having ‘always’, ‘often’ or ‘sometimes’ used any one or more of the three tools in research work) or 0 (‘never’ using any of the three tools). In the sample, 51% never used any of the three tools and 49% used one or more of the three tools.

6.1.3 Independent variables 137

Digital device ownership of a laptop, smartphone or tablet computer. Respondents were asked whether they use a (i) personal laptop (95%), (ii) smartphone (73%) or (iii) iPad or other Tablet (46%). The three variables were coded as 0 (no) and 1 (yes).

Encouragement from one’s institution or colleagues. There are three questions related to whether respondents received any encouragement from their institutions or colleagues for using social media in research work. The first question asked whether respondents had attended any social media training courses organised by their institutions and the answers were coded to 0 (no, 89%) and 1 (yes, 11%). The second question asked whether respondents were aware of any encouragement or promotion by their institution in terms of using social media to communicate and help their research. Approximately 60% answered no (coded 0) and 40% said yes (coded 1). The third question asked whether respondents had received any recommendation from colleagues for using social media to communicate and help their research. The answers had two categories of 0 (no, 68%) and 1 (yes, 32%).

Consideration for dissemination of research findings to the public. Respondents were asked who should know about their research findings by a ‘tick all that apply’ question. The choices given included other researchers (98%), policy makers (65%), the general public (66%), service users (40%) and other (11%). Those who ticked ‘general public’ were coded as 1 (66%) to represent those who consider the public as a dissemination audience and 0 (34%) for those who did not select the general public as an audience who should know about their research findings.

Personal views on social media. This relates to personal attitudes towards social media. Factor scores of three factors were generated through factor analysis to represent personal views on the effect of using social media in research work. These three factors represent three independent variables which will be discussed in more detail.

As discussed in Section 5.2.1, respondents were asked to what extent they agreed or disagreed with eleven statements related to seven positive effects and four negative effects of using social media in research work (from ‘strongly agree’ to ‘strongly disagree’ on a five-point scale). The seven positive effects items were recoded from ‘strongly disagree’ to ‘strongly agree’ so that all eleven items were measured in the same direction representing from negative to positive attitudes towards the effects of using social media. Factor analysis was conducted to reduce them to identify the key components of the attitudes and these were used as independent variables in the logistic regression models. 138

The reliability of the eleven items was tested and the Cronbach’s coefficient alpha was 0.73 which would be classified as ‘respectable’ by DeVellis (2003). The Kaiser-Meyer-Olkin measure of sampling adequacy for the eleven items is 0.845 which means the variables are good for factor analysis. These eleven items were generally well correlated with each other with the highest correlation at 0.812 and each item had at least one correlation above 0.3. Bartlett’s test of Sphericity is highly significant (p<0.001), indicating that correlations between items were sufficiently large for factor analysis. Principle component analysis was conducted on the eleven items to reduce the data to a set of factor scores17. An initial analysis was run to obtain eigenvalues for each component in the data in order to verify the number of factors. The initial variable communalities range from 0.37 to 0.82. Two components had eigenvalues greater than 1 with eigenvalues of 4.3 and 2.3. A third component had an eigenvalue of 0.83 and a fourth component had an eigenvalue of 0.74. Kaiser (1960) suggest that useful factors should account for at least as much variance as a single variable and must have an eigenvalue greater than 1. Jolliffe regarded Kaiser’s criterion as too strict and recommended a cutoff of 0.7 (Joliffe and Morgan 1992). After the author manually set the factor numbers as 2, 3 and 4 and compared the results of total variance explained and factor loadings using orthogonal rotation (varimax), the author decided to extract three factors18.

The meaning of the factors can be understood by looking at their relationship with the observed variables. The factor loadings reflect the strength of the relationship between the factors and the indicators. The varimax rotation method aims to minimise the number of variables with a high loading on each factor which makes it easier to identify the factors. Social scientists typically consider the absolute value of loading over 0.3 to be important (Field 2009). Table 6.1 shows the rotated factor loadings on the three factors. All eleven items had the value of loading over 0.3.

Overall 68% of the variance in these eleven items was attributable to three latent variables. The factor loading for the first factor, which explained 39% of the total variance, highlighted benefits for the individuals in terms of using social media in research. Thus, the first factor would be termed as ‘social media benefits for individuals’. Five items indicated benefits which were all for helping individuals advance their academic careers, for example, to help them find

17 Principle axis factoring model was also conducted and the solutions were very similar as using principle component analysis. 18 The total variance explained by four factors was 75%. However, there is only one item in the fourth component (‘research published on social media cannot be trusted as not being peer-reviewed’). If three factors were extracted, the total variance explained was 68% and this item had a rotated factor loading of 0.474 on the second component. Thus the author decided to extract three factors. 139

collaboration and funding or promote their professional profiles. One of the items, ‘blogging or tweeting about my publication will increase citations’ had a middling loading 0.559 on Factor 1 and 0.428 on Factor 3. Citation increase is an individual benefit for academics but is also a benefit for the public good as it means the research outputs have been publicised and potentially built up on by other researchers.

Table 6.1 Factor loadings of eleven attitudes items on three rotated factors.

Rotated Component Matrixa Component 1 2 3 Blogging or tweeting about my publication will increase citations. 0.559 0.060 0.428 Using social media helps me find collaboration opportunities. 0.756 -0.031 0.304 Using social media increases my chances of getting funding. 0.809 0.003 0.134 Using social media promotes my professional profile. 0.848 0.051 0.203 Using social media benefits my career. 0.875 0.073 0.221 Communicating research on social media may leak results to competitors. -0.050 0.852 -0.061 Research published on social media cannot be trusted as not being peer-reviewed. 0.270 0.474 0.262 Communicating research on social media risks my good ideas being stolen. -0.022 0.882 0.003 Communicating research on social media may result in plagiarism. 0.024 0.796 0.102 Communicating research on social media benefits the public. 0.251 0.060 0.854 Communicating research on social media accelerates scientific discovery. 0.385 0.079 0.776 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 3 iterations.

The second factor, which could be termed as ‘social media risks’, accounted for 21% of the variance in the items after rotation and loaded highest on issues pertaining to risks related to using social media. Four items indicated the potential risks relating to the consequence of disseminating scientific content online or gathering information to be used for research from social media sites. Three of the four items indicated the risks of disseminating scientific contents online which could lead to individual loss of scientific priority or academic rewards. Another item, ‘research published on social media cannot be trusted as not being peer-reviewed’ had a weaker loading of 0.474 than the other items representing Factor 2, but its rotated loadings for Factor 1 and 3 were smaller than 0.3. Thus it is clearly an indication of the risks related to using social media in research work and it implies the risk with using information from social media sites for research is that information cannot be trusted. Trust of social media content may not be associated strongly with other risks such as plagiarism or ideas being stolen, but still show as being important.

Factor 3, which could be termed ‘social media benefits for the public good’, accounted for 8.6% of the variance after rotation, and had high loadings on benefits for the public good through communicating research on social media. Two items indicated benefits for all – to accelerate 140

scientific discovery and benefit the public. This factor indicated how much academics consider the benefit of using social media for the general public (instead of individuals as in Factor 1). This variable and the previous independent variable, ‘consideration for dissemination of research findings to the public’, both showed levels of concern for the public good.

Factor scores based upon these three factors were computed using the regression method and saved as variables for using in logistic regression models. Factor scores weighted item responses by the factor loadings. A higher score of Factors 1 and 3 indicated the respondent’s stronger agreement with benefits of using social media. A higher score of Factor 2 indicated the respondent’s stronger disagreement with risks related to using social media. These three factor scores were used as independent variables to measure respondents’ personal views towards the effects of using social media.

Control Variables

Gender. 0=female (46%); 1=male (54%)

Academic discipline. The results in Chapter Five suggested that respondents from Humanities and Social Sciences were more likely than those from Medical and Natural Sciences to use social media for research. The patterns for those from Arts and Humanities and Social Sciences were quite similar, whilst the patterns for those from Medical and Life Sciences and Natural Sciences and Engineering were also quite similar. Thus the previous four categories of academic discipline were grouped into two: 0=Medical and Life Sciences/Natural Sciences and Engineering (58%); 1=Social Sciences/Arts and Humanities (42%).

Age. The four age groups which were discussed in Chapters Four and Five are ‘under 35’ (35%), ‘35-44’ (26%), ‘45-54’ (21%) and ‘55 and over’ (18%). The youngest group ‘under 35’ was used as the reference group.

6.1.4 Logistic regression results Logistic regression analysis was run in SPSS using a stepwise procedure which only included gender in the first block. The second block added discipline and age. The third block added technical factors, social factors and personal views on the preferred dissemination audience. Finally personal views on social media were added in the fourth block. The stepwise method enables the modelling results to compare significance levels of independent variables after including other independent variables. Table 6.2 reports the coefficient B values and their odds ratios (ORs). 141

Table 6.2 Logistic regression analysis on the likelihood of using social media for research

Model 1 Model 2 Model 3 Model 4 B OR B OR B OR B OR Background Gender (male) -0.23 0.80* -0.04 0.96 0.03 1.03 0.14 1.15 Characteristics Discipline (Sciences (0); Humanities (1)) 0.72 2.06*** 0.65 1.91*** 0.52 1.69*** Age (reference - under 35) 35-44 -0.13 0.88 -0.16 0.85 -0.06 0.94 45-54 -0.61 0.54*** -0.57 0.56*** -0.31 0.74 55 and over -0.97 0.38*** -0.81 0.44*** -0.30 0.74 Technical Factors Use of laptop -0.03 0.97 -0.02 0.98 Use of smartphone 0.56 1.75*** 0.37 1.45* Use of tablet 0.37 1.45** 0.39 1.47** Social Factors Having received SM training 0.40 1.50 0.21 1.23 Institutional encouragement 0.38 1.46** 0.29 1.33* Peer recommendation 1.06 2.88*** 0.82 2.27*** Personal Views on Consideration for dissemination to public 0.34 1.41** 0.28 1.33* Dissemination Attitude towards SM benefits for public good 0.48 1.61*** Audience & Social Attitude towards SM benefits for individuals 0.70 2.02*** Media Attitude towards SM risks 0.23 1.26*** Constant 0.15 1.17 0.08 1.09 -1.23 0.29*** -1.11 0.33*** Nagelkerke R Square 0.00 0.08 0.22 0.33 N=1358 Significance level of OR *p<.05 **p<.01 ***p<.001. B: Coefficient OR: odds ratio SM: Social Media

As outlined in Chapter Five, descriptive analysis found that women were slightly more likely to use Twitter and Facebook in research work as reported in Section 5.1.3. Model 1 only included gender which showed a significant positive association (p<0.05) with the likelihood of using social media tools. The odds ratio for male was 0.8 which meant that women were 1.25 (1/0.8=1.25) times more likely than men to use one or more of the three social media tools in research work. In Model 2 being female no longer had an association on the likelihood of using social media in research after adding discipline and age as independent variables. This suggests that gender might be correlated with one or more of the other added variables. Being in the Humanities disciplines had a significant positive association (p<0.001) with using social media. An academic in Arts and Humanities or Social Sciences was 2.06 times as likely as someone in Medical and Life Sciences or Natural Sciences and Engineering to use one or more of the three social media tools for research having allowed for gender and age in the model. Model 2 results suggest that older age groups were less likely to use social media for research after controlling for gender and discipline. Those aged under 35 were 2.63 times more likely than those aged 55 and over, and 1.85 times more likely than those aged 45-54 to use social media for research. There was no significant difference between those aged under 35 and 35-44 in terms of the likelihood of using social media for research.

Model 3 added the three variables of technical factors, three variables of social factors and personal views on dissemination audience. The three technical factors included the use of three types of digital devices. The use of smartphones and tablet computers were positively associated 142

with the likelihood of an academic using social media for their research. Smartphone users were 1.75 times more likely than non-users and tablet computer users were 1.5 times more likely than non-users to use social media in research work. The inclusion of personal laptops showed no significant association with the likelihood of social media use which might be explained as most respondents owned personal laptops (95%). The three social factors include training and encouragement by one’s institution and recommendations from colleagues. No significant association was found for having received social media training. Peer recommendation had a strong association with an academic’s social media use in their research work. Persons who had received a recommendation from colleagues were 2.88 times more likely to use social media in research work than those who had not. Moreover, being aware of encouragement or promotion by one’s institution for using social media to communicate research was also a significant factor in explaining a person’s social media use in research work. People that were aware of institutional encouragement for using social media were 1.46 times more likely to conduct such practice than those who were not aware. Consideration for dissemination of research findings to the public was also significantly associated with using social media in research work. Those who considered that the general public should know about their research findings were 1.41 times more likely to use social media in research work than those who exclude the public for disseminating research outputs.

Model 4 added the three factors generated from the eleven items indicating attitudes towards social media’s benefits for public good and for individuals, and attitudes towards social media’s risks.

In general, positive attitudes towards social media in general were positively associated with academic social media use. Respondents who agreed that communicating research on social media would benefit the public good and those in agreement with benefits for individuals were more likely to use social media in research work. Those who were more concerned about the potential risks of using social media for research were less likely to use it for research. In the final model, all other previous significant explanatory variables remained significant except age. No age group having a significant association in Model 4 suggests that age had high correlations with attitudes towards social media benefits and risks. As discussed in Chapter Five, older academics had more negative views towards social media.

The Nagelkerke R Square indicated how well the models explained whether an academic used one or more of the three social media tools for research. Gender alone explained less than 1% of the variance compared to model 2, which added discipline and age, and explained 8%. The 143

inclusion of technical factors, social factors and personal view on dissemination audience increased the proportion of explained variance to 22%. Attitudes towards social media’s benefits and risks substantially increased the explained variance to 33% for the final model. Thus 33% of variance in the dependent variable was explained by these explanatory variables with 1,358 cases in the models.

Having controlled for key demographic characteristics, access to smartphones and tablet computers, institutional encouragement, peer recommendation, consideration for dissemination to the public and positive views on social media were positively associated with academic use of social media as suggested in the final model. Being in the Humanities and Social Sciences rather than Medical and Natural Sciences had a consistent positive association with high significance (p<0.001) on the likelihood of academic social media use from Model 2 to 4. Age remained significant from Model 2 to 3 but was highly correlated with attitudes towards social media. There was no significant difference between men and women in using one or any of the three social media tools for research after taking account of other factors. Thus there was no real gender difference in using the three social media tools for research.

Hypothesis 1 was supported except for laptops. After controlling for social factors and personal attitudes, the uses of smartphones and tablets still showed significant positive associations with academic social media use. Hence, the use of the digital device of smartphones and tablet computers were significantly associated with academics’ social media use. Hypothesis 2 was supported for institutional encouragement and peer recommendation, but not for having received social media training. Compared with other binary independent variables, peer recommendation had the biggest beta size which suggests that it was a strong predictor. This suggests a strong social network influence over one’s academic social media use. Compared to peer recommendation, institutional encouragement is a weaker predictor for social media research use but still significant after taking account of personal attitudes towards social media. Whether a person had received social media training seemed to be irrelevant to the likelihood of social media research use.

Hypothesis 3 was supported as academics who considered that the general public should know about their research findings were more likely to use social media in research work. Hypothesis 4 was supported as positive personal views towards social media are positively associated with the likelihood of academic social media use.

144

6.2 Factors associated with the likelihood of open access publishing

6.2.1 Introduction and hypotheses Based on the findings from the analysis of the survey data reported in Chapter Four and existing literature, a number of factors were important to consider for inclusion in the logistic regression analysis of the likelihood of academics supporting and using open access publishing. Gold OA publishing and Green OA publishing were examined individually as the two outcome variables in the logistic regression models. They are referred to as the ‘Gold OA model’ and the ‘Green OA model’ in short in this section.

The findings reported in Chapter Four suggest that there were differences in terms of gender, discipline, age, job grade and research experience in both Gold OA and Green OA publishing. Similarly as in Section 6.1, gender, discipline and age were used as control variables in the models. Four hypotheses are proposed for the likelihood of publishing in Gold OA journals and five hypothesizes are proposed for the likelihood of publishing through Green OA channels.

Regarding the likelihood of using Gold OA publishing, Hypothesis 1 accounts for awareness of OA policy. Hypothesis 2 accounts for attitudes towards making research articles freely accessible online to everyone. Hypothesis 3 considers attitudes towards the citation impact of open access articles. Hypothesis 4 accounts for social media experience. In relation to the likelihood of Green OA publishing, the first four hypotheses account for the same attitudes and experiences as for Gold OA publishing. In addition, Hypothesis 5 considers the awareness of OA repositories. These hypotheses are discussed in more detail below.

Hypothesis 1a (Gold OA): Awareness of RCUK’s OA policy is positively associated with the likelihood of using Gold OA publishing.

Hypothesis 1b (Green OA): Awareness of RCUK’s OA policy is positively associated with the likelihood of using Green OA publishing.

RCUK’s OA policy came into effect on 1 April 2013. However, as discussed in Chapter Two, OA policy has been discussed in academia across various disciplines and the major funding bodies in the UK have had policies on Open Access since 2005. Thus those who had full knowledge about 145

OA policy might be more likely to have used both Gold and Green OA publishing. A key issue here is whether awareness is associated with practice among academics.

Hypothesis 2a (Gold OA): Positive attitudes towards the importance of making research articles freely accessible to everyone are positively associated with the likelihood of using Gold OA publishing.

Hypothesis 2b (Green OA): Positive attitudes towards the importance of making research articles freely accessible to everyone are positively associated with the likelihood of using Green OA publishing.

As found in Section 6.1, academics that had used social media in research work were more likely to consider for the general public to know about their research findings and for the public to benefit from communicating research on social media. Academics that had stronger positive views towards the importance of making research articles freely accessible to everyone indicated more concern for the public good, and might be more likely to practise OA publishing.

Hypothesis 3a (Gold OA): Agreeing that OA articles receive more citations is positively associated with the likelihood of using Gold OA publishing.

Hypothesis 3b (Green OA): Agreeing that OA articles receive more citations is positively associated with the likelihood of using Green OA publishing.

Citation impact is highly associated with academic rewards in academia as discussed in Chapter Two. Citation increase is a potential positive effect of publishing in OA journals (Eysenbach 2006) and self-archiving (Gargouri et al. 2010). As reported in Section 5.3.1, 31% of survey respondents in this study agreed that blogging or tweeting about their publications would lead to an increase of citations. Thus it is likely that academics who believe that OA articles will receive more citations are more likely to publish through OA channels.

Hypothesis 4a (Gold OA): Social media experience is positively associated with the likelihood of using Gold OA publishing. 146

Hypothesis 4b (Green OA): Social media experience is positively associated with the likelihood of using Green OA publishing.

Since OA publishing and using social media in research are the two main focuses of this study, it is important to explore whether there is an association between the two behaviours. Academics that had used social media in their research work might be more likely to learn about OA policy and find out about self-archiving articles while browsing tweets or following colleagues on Academia.edu. On the other hand, academics who had published articles in OA channels might use social media because that they were more eager to promote their research to wider audience and also blog or tweet about their papers.

Hypothesis 5 (Green OA): Awareness of OA repositories is positively associated with the likelihood of using Green OA publishing.

In relation to the likelihood of Green OA publishing, the awareness of OA repositories and the use of OA repositories are likely to be associated with each other. 95% of survey respondents from this study who used Green OA publishing were aware of OA repositories. It is important to control for awareness of an OA repository because if awareness of an OA repository ‘perfectly’ explains self-archiving behaviour, then other factors will not have any effects on the likelihood of experiencing Green OA publishing.

6.2.2 Dependent variables Two outcome variables will be tested in logistic regression models for the likelihood of experiencing Gold OA and Green OA publishing.

Outcome Variable 1: Gold OA publishing experience (0,1). The variable was coded from the original question where respondents were asked whether they had published articles in a journal that was open access (OA). This variable only includes the subgroup of respondents who had published research articles. The answers were coded to either 1 (40%) for ‘yes’ or 0 (60%) representing no such experience.

Outcome Variable 2: Green OA publishing experience (0,1). The variable was coded from the original question where respondents were asked whether they had deposited their research articles in an open access online repository. This variable also only includes the subgroup of 147

respondents who had published research articles. This outcome variable was coded to either 1 (43%) for ‘yes’ or 0 (57%) indicating no such experience.

6.2.3 Independent variables All the independent variables except awareness of OA repositories reported here are used in both the models examining the use of Gold OA and Green OA publishing.

Awareness of RCUK’s OA policy. The answers were ‘yes’ (42%), ‘heard of, but not sure about the detail’ (30%) and ‘no’ (28%). ‘Yes’ was used as the reference group.

Attitudes towards the importance of making research articles OA to everyone. Respondents were asked how important they thought it was, in general, to make research articles freely accessible online to everyone. The answer ‘not at all important’ was selected by only twelve cases and was thus grouped together with ‘not very important’ to represent ‘not important’. The final three groups that were analysed in the models were ‘very important’ (57%), ‘fairly important’ (36%) and ‘not important’ (7%). ‘Not important’ was used as the reference group.

Attitudes towards citation impact of OA articles. Respondents were asked to what extent they agreed or disagreed that ‘articles that are made open access will receive more citations’. The answer ‘strongly disagree’ was selected by only nine cases and was grouped together with ‘disagree’ to represent ‘disagree’. The final four groups analysed in the models were strongly agree (16%), agree (39%), neither disagree nor agree (38%) and disagree (8%). ‘Disagree’ was used as the reference group.

Use of social media tools in research work. This was the same variable used as the dependent variable in Section 6.1.2.

Awareness of OA repositories. The answers were coded as 1=yes (74%) and 0=no (26%). This variable was only used in the Green OA model.

Control Variables

Academic discipline. The results in Chapter Four suggest that there were big differences in the use of Gold OA between respondents from Medical and Life Sciences (61%) and Natural Sciences and Engineering (35%). Thus the four discipline areas will be tested individually. The four discipline areas are Medical and Life Sciences (35%), Natural Sciences and Engineering (23%), Social Sciences (27%) and Arts and Humanities (15%). Natural Sciences and Engineering was used 148

as the reference group because their respondents were more likely to use Green OA publishing, as reported in Section 4.1.2.

Gender and Age. As detailed in Section 6.1.3.

6.2.4 Logistic regression results Logistic regression analysis was conducted to test the hypothesised association of various factors with the likelihood of using Gold OA and Green OA publishing. Table 6.3 summarises the results of the analyses19.

Table 6.3 Logistic regression analysis on the likelihood of publishing in Gold OA and Green OA

Gold OA model Green OA model B OR B OR Background Characteristics Gender (male) 0.31 1.37* 0.05 1.05 Discipline (reference-Natural Sciences & Engineering) Medical & Life Sciences 1.22 3.37*** -0.53 0.59** Social Sciences -0.36 0.70* -0.39 0.68* Arts & Humanities -0.50 0.61* -0.63 0.53** Age (reference - under 35) 35-44 0.44 1.56** 0.47 1.60** 45-54 0.38 1.47* 0.55 1.73** 55 and over 0.28 1.32 0.36 1.44 Awareness of RCUK OA policy Awareness of RCUK policy (reference - yes) No -0.75 0.47*** -0.76 0.47*** Heard of, but not sure about the detail -0.85 0.43*** -0.64 0.52*** Attitudes towards the OA importance (reference - not important) Importance of OA Publishing Very important 0.94 2.57*** 0.91 2.48*** Fairly important 0.66 1.93* 0.71 2.04** Attitudes towards Citation OA increases citation (reference - disagree) Impact of OA Articles Strongly agree 0.75 2.12** -0.24 0.78 Agree 0.58 1.79* -0.17 0.84 Neither disagree nor agree 0.42 1.53 -0.39 0.67 Social Media Experience Use of social media in research work 0.06 1.06 0.31 1.36* Awareness of OA Repositories Awareness of OA repositories 2.00 7.40*** Constant -2.03 0.13*** -2.22 0.11*** Nagelkerke R Square 0.23 0.26 N=1469 N=1484 Significance level of OR *p<.05 **p<.01 ***p<.001. B: Coefficient OR: odds ratio

The explanatory variables explain 23% of the variance in the Gold OA model. With an additional explanatory variable of OA repositories awareness, 26% of variance in the Green OA model was explained. As Table 6.3 shows, discipline, age, awareness of RCUK OA policy and attitudes towards the importance of OA publishing were significantly associated with the likelihood of both Gold OA and Green OA publishing. Gender and attitudes towards the citation impact of OA articles were significantly associated with Gold OA publishing, but not with Green

19 Both Gold OA and Green OA models were run using a stepwise inclusion of independent variables, similarly as in 6.1.4 and the significance level of each individual variable was the same in the final model and earlier models. Thus only the two final models are reported here. 149

OA publishing. Experience of using social media had a significant association with Green OA publishing, but not with Gold OA publishing.

Of those academics who had published research articles, men were 1.37 times more likely than women to use Gold OA publishing having allowed for all other factors in the model (p<0.05). However, there was no significant difference between men and women in using Green OA publishing which contradicts the findings from Chapter Four. Those in Medical and Life Sciences an academic were 3.37 times more likely than those in Natural Sciences and Engineering (p<0.001) to publish in OA journals. Respondents in Natural Sciences and Engineering were 1.43 times (1/0.70) more likely than those in Social Sciences (p<0.05) and 1.64 times (1/0.61) more likely than those in Arts and Humanities (p<0.05) to use Gold OA publishing. In the Green OA model, respondents in Natural Sciences and Engineering were significantly more likely to self-archive research articles than those in any other disciplines. These patterns are in line with findings from Chapter Four as reported in Section 4.1.2. In both the Gold OA model and the Green OA model, the age group 55 and over showed no significant difference from the reference group under 35. Respondents aged 35-44 were 1.56 times more likely to use Gold OA publishing and 1.6 times more likely to use Green OA publishing than those aged under 35. Those aged 45-54 were 1.47 times more likely to use Gold OA and 1.73 times more likely to use Green OA than those aged under 35.

Hypothesis 1a and 1b were supported. Awareness of RCUK’s OA policy had a positive association with the likelihood of using both Gold and Green OA publishing. Those who were aware of RCUK’s OA policy were 2.12 times more likely than those that said ‘no’ and 2.32 more likely than those that said ‘heard of, but not sure about the detail’ to publish in OA journals (both p<0.001). Those aware of OA policy were 2.12 times more likely than those that said ‘no’ and 1.92 times more likely than those that said ‘heard of, but not sure about the detail’ to self-archive research articles (both p<0.001).

Hypothesis 2a and 2b were supported. Positive attitudes towards the importance of OA publishing had significant positive associations with the likelihood of using both Gold and Green OA publishing. Respondents who rated it ‘very important’ for OA publishing were 2.57 times more likely (p<0.001) and those that rated it ‘fairly important’ were 1.93 times more likely (p<0.05) to use Gold OA publishing than those who rated it ‘not important’. The patterns are similar for the likelihood of using Green OA publishing. 150

Hypothesis 3a was supported. Agreement with the belief that OA articles receive more citations was positively associated with the likelihood of using Gold OA publishing. Respondents who strongly agreed and agreed that OA articles would receive more citations were 2.12 times (p<0.01) and 1.79 times (p<0.05) more likely than those that disagreed with OA citation impact to publish in OA journals. There was no significant difference between the reference group who disagreed with the statement and those who selected ‘neither disagree nor agree’. Those who selected ‘neither disagree nor agree’ were perhaps likely to not have known enough information about open access publishing to have a strong opinion and likely to have not published in OA journals. However, attitudes towards OA citation impact had no significant association with using Green OA publishing. Hence hypothesis 3b was not supported.

Hypothesis 4b was supported. Having used social media in research work was positively associated with the likelihood of academics using Green OA publishing. Respondents who used one or more of the three social media tools in their research work were 1.36 times more likely to self-archive research articles (p<0.05). Having controlled for key background characteristics, awareness and attitudes towards OA publishing, experience of using social media was not significantly associated with the likelihood of Gold OA publishing. Thus Hypothesis 4a was not supported.

Hypothesis 5 was supported as the awareness of OA repositories was positively associated with the likelihood of using Green OA publishing. Those who were aware of OA repositories for depositing research articles were 7.4 times more likely to self-archive research articles (p<0.001).

6.3 Factors associated with the likelihood of sharing primary research data

6.3.1 Introduction and hypotheses Based on the findings from the previous analysis of the survey data reported in Chapter Four and existing literature, a number of factors were considered for the logistic regression analysis of the likelihood of sharing primary research data in online repositories. The findings reported in Chapter Four suggested that those who were male, older and in Natural Sciences and Engineering academics were more likely to share primary research data. Similar to Section 6.1 and 6.2, gender, discipline and age are used as control variables in the logistic regression models. Four hypotheses are proposed for the likelihood of sharing primary research data in online repositories. Hypothesis 151

1 accounts for social media experience. Hypothesis 2 accounts for the awareness of RCUK’s OA policy and Hypothesis 3 accounts for the attitudes towards the importance of data sharing. Hypotheses 4 and 5 consider for Gold and Green OA publishing. Hypothesis 6 controls for the experience of using secondary data.

Hypothesis 1: Social media experience is positively associated with the likelihood of sharing primary research data in online repositories.

As found in Section 6.2, having used social media in their research work was positively associated with the likelihood of depositing research articles in online repositories. One possible explanation is that academics that use social media in their research work might be more likely to discover the existence of online repositories through academic blogs or tweets. It is important to examine whether the use of social media would also be associated with depositing research data in online repositories.

Hypothesis 2: Awareness of RCUK’s OA policy is positively associated with the likelihood of sharing primary research data in online repositories.

The Research Council UK’s OA Policy requires research articles published from RCUK funded projects, if applicable, to include a statement of how the underlying research materials such as data, sampling details or models can be accessed (RCUK 2013). As discussed in Chapter Two, similar data policies have been established by the majority of UK research funders since 2008 (RIN 2008; 2011). Thus academics that have learnt about RCUK’s OA policy might be more likely to have practised data sharing.

Hypothesis 3: Positive attitudes towards the importance of making research data available online for reuse are positively associated with the likelihood of sharing primary research data in online repositories.

As found in Section 6.2, positive attitudes towards the importance of making research articles freely accessible to everyone had positive associations with the likelihood of OA publishing. Similarly, academics who have positive views towards the importance of making research data available online for reuse might be more likely to share research data. 152

Hypothesis 4: Experience of Gold OA publishing is positively associated with the likelihood of sharing primary research data in online repositories.

As research by Piwowar (2011) found, academic authors are more likely to share data if related articles were published in an open access journal or a journal with a stronger data-sharing policy. Many OA journals require authors to share data in order to publish the articles in those journals (e.g. the PLoSOne). Studies published in journals with stronger data-sharing policies seemed to be associated with increased data-sharing behaviour (Piwowar and Chapman 2010).

Hypothesis 5: Experience of Green OA publishing is positively associated with the likelihood of sharing primary research data in online repositories.

Academics who deposit research articles in online repositories might be more likely to be open to new technologies and more willing to share research outputs. Thus they might also be more likely to share research data.

Hypothesis 6: Experience of using secondary data is positively associated with the likelihood of sharing primary research data in online repositories.

As reported in Chapter Four, Section 4.2.1, respondents who had used secondary research data from online repositories collected by other researchers were more likely to have shared their own primary research data online. Almost half of the respondents who had used secondary data for their research had also deposited their own primary data online for reuse whilst only 10% of those who had no experience using secondary data had shared their own. Piwowar (2011) also found that academic authors were more likely to share data if they had prior experience reusing data. Use of secondary data can also be used as a control variable to test the association with other factors having allowed for the previous experience of reusing data.

6.3.2 Dependent variable Experience of sharing primary research data in online repositories (0,1). Respondents were asked whether they had deposited their own primary research data in online repositories that can 153

be reused by other researchers. This outcome variable was coded to either 1 (21%) for ‘yes’ or 0 (79%) for ‘no’.

6.3.3 Independent variables Attitude towards the importance of making research data available online for reuse. Respondents were asked how important they thought it was, in general, to make research data available online for reuse. The answer ‘not at all important’ was selected by only 26 cases and was grouped together with ‘not very important’ to represent ‘not important’. The final three groups analysed in the regression models were ‘very important’ (40%), ‘fairly important’ (46%) and ‘not important’ (14%). ‘Not important’ was used as the reference group.

Experience of using secondary research data from online repositories (0,1). Respondents were asked whether they had used secondary research data from online repositories that were collected by other researchers. As reported in Chapter Four (4.2.1), 29% reported ‘yes’ and 71% reported ‘no’.

Gold OA publishing experience (0,1) and Green OA publishing experience (0,1). These were the same variables used as the dependent variables in Section 6.2.2.

Use of social media tools in research work and Awareness of RCUK’s OA policy. As detailed in Section 6.2.3.

Control Variables

Gender, Academic discipline and Age. As detailed in Section 6.2.3.

6.3.4 Logistic regression results Logistic regression analysis was conducted using a stepwise procedure similar to that used in Section 6.1.4. Table 6.4 summarises the results of the analysis. Model 1 included gender, discipline and age. Having allowed for discipline and age in the model, men were 1.37 times more likely than women to deposit their own primary research data in online repositories (p<0.05). Those in Natural Sciences and Engineering were 1.51 times more likely than those in Medical and Life Sciences (p<0.05) and 1.67 times more likely than those in Arts and Humanities (p<0.05) to share data. However, no significant difference was found for academics in Social Sciences compared to being in Natural Sciences and Engineering. The Model 1 results indicate a steady increase in practising data sharing with increases in age (p<0.001). Those 55 and over were 2.57 times more likely than those under 35 to share primary research data in online repositories. 154

Table 6.4 Logistic regression analysis on the likelihood of sharing primary research data

Model 1 Model 2 Model 3 Model 4 Model 5 B OR B OR B OR B OR B OR Background Gender (male) 0.32 1.37* 0.28 1.32* 0.23 1.25 0.24 1.27 0.14 1.15 Characteristics Discipline (reference-Natural Sciences & Engineering) Medical & Life Sciences -0.41 0.66* -0.47 0.62** -0.43 0.65* -0.28 0.76 -0.22 0.80 Social Sciences -0.35 0.71 -0.25 0.78 -0.13 0.87 -0.03 0.97 -0.10 0.90 Arts & Humanities -0.55 0.60* -0.47 0.62* -0.38 0.68 -0.28 0.76 -0.27 0.76 Age (reference - under 35) 35-44 0.81 2.26*** 0.67 1.94*** 0.65 1.92*** 0.57 1.77** 0.60 1.82** 45-54 0.89 2.43*** 0.74 2.09*** 0.80 2.22*** 0.67 1.96** 0.71 2.04** 55 and over 0.94 2.57*** 0.83 2.30*** 0.84 2.31*** 0.78 2.19*** 0.91 2.49*** Social Media Experience Use of social media in research work 0.35 1.42** 0.17 1.19 0.09 1.10 -0.09 0.92 Awareness of RCUK OA Awareness of RCUK policy (reference - yes) Policy No -0.83 0.44*** -0.77 0.46*** -0.53 0.59** -0.39 0.67 Heard of, but not sure about the detail -0.43 0.65** -0.45 0.64** -0.31 0.73 -0.16 0.85 Gold OA Experience Gold OA publishing experience 0.45 1.57** 0.42 1.53** 0.25 1.28 0.30 1.35 Attitudes towards the OA importance (reference - not important) Importance of Sharing Very important 2.41 11.15*** 2.36 10.57*** 1.61 4.98*** Data Fairly important 1.14 3.12*** 1.12 3.06*** 0.73 2.07* Green OA Experience Green OA publishing experience 1.02 2.78*** 1.02 2.78*** Use of Secondary Data use of secondary data 1.83 6.25*** constant -1.80 0.17*** -1.76 0.17*** -3.37 0.03*** -3.88 0.02*** -4.08 0.02*** Nagelkerke R Square 0.05 0.10 0.22 0.27 0.38 N=1471 Significance level of OR *p<.05 **p<.01 ***p<.001. B: Coefficient OR: odds ratio

In Model 2 social media experience, awareness of RCUK’s OA policy and Gold OA publishing experience were added. Those who used one or more of the three social media tools were 1.42 times more likely to share primary research data in online repositories having controlled for gender, discipline and age (p<0.01). Those who were aware of RCUK’s OA policy were 2.27 times (1/0.44) more likely than those who said ‘no’ (p<0.001) and 1.54 times (1/0.65) than those who said ‘heard of, but not sure about the detail’ (p<0.01) to share data. Those that had published in Gold OA journals were 1.57 times more likely to share data than those who had not used Gold OA publishing (p<0.01). Gender, discipline and age were still significant in Model 2.

Model 3 added the academics’ attitudes towards the importance of making research data available online for reuse. Perhaps not surprisingly respondents who rated it ‘very important’ were 11.15 times more likely (p<0.001) and those rated it ‘fairly important’ were 3.12 times more likely (p<0.001) than those who rated it ‘not important’ to share data. Having allowed for the attitudes towards the importance of sharing data, gender and social media experience were no long significantly associated with the likelihood of sharing data. This indicates that after taking account of academics’ attitudes towards the importance of data sharing, there was no significant difference between men and women in terms of sharing data. Having allowed for attitudes towards the importance of data sharing, there was no significant difference between those who had used social media or not in terms of sharing primary research data. Moreover, being in Arts 155

and Humanities had no significant difference from being in Natural Sciences and Engineering after taking account of attitudes towards the importance of data sharing.

Model 4 added Green OA publishing experience. The result suggested that respondents who had deposited their research articles in online repositories were 2.78 times more likely to deposit research data having taking into account their awareness of OA policy and attitudes towards the importance of sharing data. Gold OA publishing experience was no longer significant after taking into account Green OA experience. Answering ‘heard of, but not sure about the detail’ for the awareness of RCUK’s OA policy and being in Medical and Life Sciences were no longer significant in Model 4.

Model 5 controlled for the experience of using secondary data from online repositories. The experience of using secondary data had the strongest association with the outcome variable compared to the other binary independent variables in terms of the beta size. Respondents who used secondary data from online repositories were 6.25 times more likely to share data (p<0.001). Age, attitude towards the importance of sharing data and Green OA experience were still significant after controlling for experience of using secondary data. The odds ratios for those who rated it ‘very important’ for data sharing reduced from 10.57 in Model 4 to 4.98 in Model 5 (p<0.001) and for those who rated it ‘fairly important’ reduced from 3.06 to 2.07 (p<0.05). Those aged 35-44 were 1.82 times more likely (p<0.01), those aged 45-54 were 2.04 times more likely (p<0.01) and those aged 55 and over were 2.49 times more likely (p<0.001) than those under 35 to share data. Answering ‘no’ for the awareness of RCUK’s OA policy was no longer significant. This was perhaps because awareness of RCUK’s OA policy was associated with Green OA experience as found by the logistic regression analysis in Section 6.2.4.

The background characteristics of gender, discipline and age explained 5% of the variance in the dependent variable. The inclusion of social media experience, awareness of RCUK’s OA policy and Gold OA experience increased the explained variance to 10%. Attitudes towards the importance of sharing data substantially increased the explained variance to 22% and Green OA experience further increased the explained variance to 27%. The final model explained 38% of the variance after controlling for experience of using secondary data.

In conclusion, the use of secondary data, Green OA publishing experience, positive attitudes towards the importance of data sharing and older age were positively associated with the likelihood of sharing primary research data. Age had a significant association in Models 1 to 5 having controlled for other factors. Discipline showed some significance in Models 1 to 3 but was 156

highly correlated with Green OA experience. There was no significant difference between men and women in sharing data after taking account of other factors. Social media experience, awareness of RCUK’s OA policy and Gold OA publishing experience had some associations with the experience of data sharing, but was correlated with other independent variables.

Hypotheses 3, 5 and 6 were supported. Positive attitudes towards the importance of making research data available online for reuse were positively associated with the likelihood of sharing primary research data in online repositories. Experiences of Green OA publishing and using secondary data were positively associated with the likelihood of sharing data. Hypotheses 1, 2 and 4 were not fully supported according to the final model. Experience of using social media had some association with data-sharing experience. Model 3 suggests that for those who have the same attitudes towards the importance of data sharing, there is no real difference between those who used social media or not in terms of having shared primary research data. Awareness of RCUK’s OA policy and Gold OA experience had some associations with data-sharing experience. Academics who stated that they had shared data online were more likely to have used secondary data and deposited research articles in online repositories, thus more likely perhaps to have learnt RCUK OA policy. For academics who had no experience of using secondary data, being aware of the RCUK OA policy had no association with sharing primary research data. Gold OA publishing experience was no longer significant having controlled for Green OA experience. This suggests that for those who had no experience of depositing research articles online, there is no significant difference from those who published in OA journals or not in terms of data sharing.

6.4 Factors associated with the likelihood of publishing ongoing research updates on social media

6.4.1 Introduction This section aims to examine what factors are associated with the likelihood of publishing ongoing research updates on social media which is the third aspect of open science considered in this thesis. Moreover, those academics who stated that they ‘always’ or ‘often’ posted ongoing research could be described as ‘super users’ (as discussed in Section 4.3.1.4) and seen as strong supporters of open science. Thus it is worth investigating what factors are associated with being a super user of publishing research updates on social media. 157

As previously found in Section 6.1, academic discipline, the use of smartphones and tablet computers, institutional encouragement, peer recommendation, consideration for dissemination to the public and positive views on social media and all had significant associations with academic use of social media. Section 6.4 explores whether the same factors associated with social media use in general can predict the likelihood of publishing ongoing research updates on social media.

The outcome variable used in Section 6.1 (social media use in general) included academics that stated that used one or more of the three social media services (Twitter, Facebook and research blogs) in their research work in general. In this section, the focus is on the academic experiences of publishing ongoing research updates on Twitter, research blogs and social networking sites. Logistic regression models with the same explanatory variables used in Section 6.1 compare and contrast with the results shown in Table 6.2 to explore whether the same factors are associated with the likelihood of publishing research updates on social media and the likelihood of being a super user.

6.4.2 Factors associated with the likelihood of publishing research updates on social media Using the same explanatory variables as in previous models20 in Section 6.1, logistic regression analysis explores what factors are associated with the likelihood of publishing research updates on social media.

6.4.2.1 Dependent variable Publishing updates of ongoing research on Twitter, research blogs and social networking sites (0,1). The outcome variable was coded as either 1 (having ‘always’, ‘often’ or ‘sometimes’ posted research updates on one or more of the three social media tools) or 0 (having ‘never’ done so). In the sample, 30% of respondents had posted research updates on one or more of the three tools and 70% had never had such experience.

6.4.2.2 Independent variables The same variables are used as detailed in Section 6.1.3.

6.4.2.3 Logistic regression results Table 6.5 reports the coefficient B values and corresponding odds ratios (ORs) from the logistic regression models on the likelihood of publishing ongoing research updates on social media. The

20 ‘Previous models’ in Section 6.4 refers to the four models on the likelihood of academic social media use reported in Section 6.1.4 and shown in Table 6.2. 158

Nagelkerke R Square indicated that all explanatory variables explained 32% of variance in the final model. There were less significant associations in the final model compared to the results from the models on the likelihood of academic social media use in general.

Table 6.5 Logistic regression analysis on the likelihood of publishing ongoing research updates on social media

Model 1 Model 2 Model 3 Model 4 B OR B OR B OR B OR Background Gender (male) -0.07 0.93 0.12 1.12 0.18 1.20 0.31 1.36* Characteristics Discipline (Sciences (0); Humanities (1)) 0.78 2.19*** 0.73 2.07*** 0.64 1.90*** Age (reference - under 35) 35-44 0.14 1.15 0.16 1.17 0.30 1.35 45-54 -0.38 0.68* -0.32 0.72 0.03 1.04 55 and over -0.98 0.38*** -0.80 0.45*** -0.25 0.78 Technical Factors Use of laptop -0.24 0.79 -0.21 0.81 Use of smartphone 0.48 1.62** 0.24 1.27 Use of tablet 0.27 1.31* 0.29 1.33* Social Factors Having received SM training 0.28 1.32 0.03 1.03 Institutional encouragement 0.34 1.41* 0.23 1.25 Peer recommendation 0.89 2.44*** 0.62 1.85*** Personal Views on Consideration for dissemination to public 0.56 1.75*** 0.50 1.65*** Dissemination Attitude towards SM benefits for public good 0.43 1.54*** Audience & Social Attitude towards SM benefits for individuals 0.80 2.24*** Media Attitude towards SM risks 0.40 1.50*** Constant -0.73 0.48*** -1.00 0.37*** -2.20 0.11*** -2.23 0.11*** Nagelkerke R Square 0.00 0.08 0.19 0.32 N=1366 Significance level of OR *p<.05 **p<.01 ***p<.001. B: Coefficient OR: odds ratio SM: Social Media

Having controlled for key demographic characteristics, the use of tablet computers, peer recommendation and personal views towards research impact and social media were significantly associated with the likelihood of publishing research updates on social media. Gender was not associated with the outcome variable in Models 1, 2 and 3, but was significant in Model 4. This might be because men were more likely to publish research updates on blogs21. Discipline was significantly associated with the outcome variable in Models 2, 3 and 4, which is consistent with the findings from previous models. The odds ratios were slightly higher in all three models for academic discipline indicating that there was a larger difference between academics in Humanities and Sciences disciplines for publishing ongoing research updates than there was for general social media use in research work. Differences in terms of age were not significant in Model 4 which is similar with the findings from the models on the likelihood of social media use in general.

The use of smartphones and institutional encouragement were significant in Model 3. However, taking account of academics’ attitudes towards the effects of using social media, the

21 To investigate the gender significance, logistic regression models were run with the same explanatory variables with three outcome variables of posting research updates on blogs/Twitter/SNS individually. Gender shows significant effect in the models on blogs, but not significant with the models on Twitter and SNS. 159

use of smartphones and institutional encouragement were no longer significant. The use of tablet computers and peer recommendation were significantly associated with publishing ongoing research after controlling for academics’ attitudes towards social media. No significant association was found for the use of laptops and having received social media training which is consistent with previous models.

Consideration for the dissemination of research findings to the public was significantly associated with publishing ongoing research updates on social media, which is consistent with previous models. Academics who considered that the general public should know about their research findings were more likely to use social media to publish ongoing research. The odds ratios are higher in both Models 3 and 4 compared to previous models indicating consideration of the public as a dissemination audience might have a stronger association with disseminating research updates than in the general use of social media. The academics’ personal views on the positive and negative effects of using social media in research work showed similar patterns to previous models. Those with positive views towards the effects of social media were more likely to disseminate ongoing research updates on social media.

6.4.3 Factors associated with the likelihood of being a super user publishing ongoing research updates on social media. Using the same explanatory variables, logistic regression analysis is now used to explore what factors are associated with the likelihood of being a super user in terms of publishing ongoing research updates on social media.

6.4.3.1 Dependent variable Being a super user vs an occasional user publishing updates of ongoing research on Twitter, research blogs and social networking sites (0,1). The outcome variable was coded to either 1 (having ‘always’ or ‘often’ posted research updates on one or more of the three social media tools) or 0 (having ‘sometimes’ done so). In the sample, 27% (371) are super users and 73% (139) are occasional users.

6.4.3.2 Independent variables The same variables are used as detailed in Section 6.1.3.

6.4.3.3 Logistic regression results Table 6.6 highlights the associations of being a super user compared to being an occasional user. Being in Humanities, the use of tablet computers and positive attitudes towards the effects 160

of using social media had positive associations with the likelihood of being a super user in the final model. The odds ratios were higher in Models 3 and 4 for the use of tablet computers compared to the results from the likelihood of publishing ongoing research updates indicating that the use of tablet computers had a stronger association with being a super user. Respondents who had used tablet computers were 2.09 times more likely to be super users instead of occasional users. Having received social media training was significant in Model 3 and lost its significance in Model 4. Without controlling for attitudes towards the effects of using social media, respondents who had received social media training were 2.26 times more likely to be super users than those who had not received training. The use of smartphones, institutional encouragement, peer recommendation and consideration for dissemination of research findings to the public were not significant.

Table 6.6 Logistic regression analysis on the likelihood of being a super user compared to an occasional user publishing ongoing research updates on social media

Model 1 Model 2 Model 3 Model 4 B OR B OR B OR B OR Background Gender (male) 0.02 1.02 0.20 1.22 0.23 1.26 0.31 1.36 Characteristics Discipline (Sciences (0); Humanities (1)) 0.55 1.73* 0.59 1.81* 0.60 1.82* Age (reference - under 35) 35-44 0.06 1.06 0.04 1.04 0.17 1.18 45-54 -0.69 0.50* -0.74 0.48* -0.32 0.72 55 and over -1.24 0.29* -1.31 0.27* -0.74 0.48 Technical Factors Use of laptop 0.34 1.40 0.38 1.46 Use of smartphone 0.53 1.69 0.31 1.36 Use of tablet 0.66 1.94** 0.74 2.09** Social Factors Having received SM training 0.81 2.26** 0.49 1.63 Institutional encouragement -0.11 0.90 -0.07 0.93 Peer recommendation 0.30 1.35 0.05 1.06 Personal Views on Consideration for dissemination to public 0.18 1.19 0.07 1.07 Dissemination Attitude towards SM benefits for public good 0.46 1.59*** Audience & Social Attitude towards SM benefits for individuals 0.68 1.98*** Media Attitude towards SM risks 0.25 1.28* Constant -1.00 0.37*** -1.22 0.29*** -2.74 0.06*** -3.27 0.04*** Nagelkerke R Square 0.00 0.06 0.14 0.24 N=432 Significance level of OR *p<.05 **p<.01 ***p<.001. B: Coefficient OR: odds ratio SM: Social Media

6.5 Conclusion and discussion

This chapter has explored the factors associated with academics using social media in research work and contributing to open science. Logistic regression analysis investigated factors associated with the likelihood of using social media in research work, publishing in Gold and Green OA and sharing primary research data. It also explored factors associated with the likelihood of publishing ongoing research on social media and being a super user. Background characteristics were 161

included in the logistic regression models to explore whether there were significant differences in terms of gender, discipline or age after taking account of other factors.

6.5.1 Gender differences Gender differences were found for the likelihood of using Gold OA publishing. Men were more likely to publish in OA journals having controlled for discipline, age, social media experience, awareness of and attitudes towards OA publishing. However, there was no significant difference between men and women in using Green OA publishing. No significant gender difference was found for the likelihood of sharing data after taking account of academics’ attitudes towards the importance of data sharing. Gender differences were not evident in using social media for research or being a super user in publishing updates on social media having controlled for other factors.

6.5.2 Disciplinary differences Discipline was significantly associated with the likelihood of using both Gold and Green OA publishing after controlling for other factors. Academics in Medical and Life Sciences were much more likely than those in Natural Sciences, Humanities or Social Sciences to publish in OA journals. Those in Natural Sciences and Engineering were more likely to self-archive research articles than those in any other discipline. However, there was no significant difference between the four discipline areas in the likelihood of data sharing after taking account of attitudes towards data sharing and experience of self-archiving, possibly because the disciplinary differences were explained by attitudes towards sharing data and experience of Green OA. Academics in Natural Sciences and Engineering were more likely to share their research data because they had more experience of Green OA publishing and were more likely to regard data sharing as very important.

Academics in Humanities and Social Sciences were less likely to use OA publishing, but more likely to use social media in research work in general and to publish ongoing research updates as well as being super users. One possible explanation is that academics in Medical and Natural Sciences are likely to work in teams and collaborators might not be comfortable with one team member share updates on social media. Academics in Humanities and Social Sciences are more likely to work in smaller groups or alone and thus have more freedom to choose the way to communicate their research. One respondent from pilot studies also suggested unlike Hard Sciences, there were no ‘results’ in his subject area. Some Arts and Humanities disciplines produce art work or monographs as research outputs rather than research papers. Thus 162

academics in Humanities and Social Sciences might be less concerned about revealing results to competitors.

6.5.3 Age differences Age had significant associations with the likelihood of Gold and Green OA publishing and sharing data after controlling for other factors. After taking account of academics’ awareness of RCUK’s OA policy and attitudes towards the importance of OA publishing, older academics were more likely to use both Gold and Green OA publishing. Older academics were also more likely to share primary research data in online repositories having controlled for awareness of OA policy, attitudes towards the importance of sharing data and the use of Green OA and secondary data. Older and senior academics are also more likely to have generated more data to share from a longer experience in research than junior academics. Junior academics may have more concerns of securing academics rewards, thus their chances of publishing before competitors could be jeopardised if they share primary research data. Age had no significant association with the likelihood of social media use after controlling for attitudes towards social media. This may be due to the high correlations between age and attitudes towards social media benefits and risks. As discussed in Chapter Five, older academics had more negative views towards social media.

6.5.4 OA publishing Academics that were aware of RCUK’s OA policy were more likely to use both Gold and Green OA publishing. Those who regarded making research articles open access as ‘very important’ or ‘fairly important’ were two times more likely than those who regarded OA publishing as not important to publish in OA journals and self-archive research articles. Academics who felt that OA articles would receive more citations were more likely to use Gold OA publishing but not Green OA publishing. This might be because academics who published in OA journals had noticed an increase in citations and readership. However, it also suggests that those who had self-archived their research articles were not necessarily convinced that doing so would increase the citation of their articles.

Academics that had used social media in their research work were more likely to use Green OA publishing but not Gold OA publishing. One of the possible explanations is the characteristics of an academic being more open to adopt new technologies and new media tools. Another possible explanation is that academics that had used social media in research work were more likely to use social network sites such as Academia.edu and ResearchGate to self-archive research articles. The awareness of OA repositories is strongly associated with the likelihood of self-archiving research 163

articles. Academics that were aware of OA repositories were much more likely to use Green OA publishing.

6.5.5 Sharing primary research data Having used secondary data, experience of Green OA publishing and positive attitudes towards the importance of sharing data were positively associated with the likelihood of sharing primary research data in online repositories. Prior experience of using secondary data is strongly associated with sharing data which suggests a form of reciprocity. Those who had used data collected by others might view sharing as a way of giving back to the scientific community. Green OA publishing experience was positively associated with data-sharing experience. This suggests that those who had used repositories to deposit articles might be more likely to deposit data alongside their articles as part of funding requirements and have become more skilled as regard how to use repositories to deposit data. Agreeing with the importance of data sharing was also positively associated with sharing data.

6.5.6 Using social media for research The use of tablet computers, peer recommendation, concerns for public good and positive attitudes towards the effects of social media were significantly associated with the likelihood of academic use of social media in general and the likelihood of publishing ongoing research updates on social media. The use of tablet computers was positively associated with the likelihood of being a super user. The use of smartphones was associated with the likelihood of academic social media use in general after controlling for other factors. The use of laptops was very common among academics but not associated with social media use. This suggests that smartphones can be useful for reading information from social media but not useful for publishing research updates or writing blogs possibly because of its small screen and difficulty of typing long sentences. Tablet computers are easier to connect to the internet than laptop computers using similar networks as smartphones. Thus if an academic social media user is travelling, he or she may use tablets to connect to the internet to browse and post messages. Tablet computers can provide social media users a mobile platform from which to disseminate research updates and those who like to post ongoing updates might be more inclined to purchase a tablet computer to support this.

Peer recommendation was associated with the likelihood of using social media tools in general and for publishing research updates. This suggests that encouragement by one’s social network has a strong positive impact on academics’ use of social media which is in line with findings from Friemel (2014). However, peer recommendation had no association with the likelihood of being a 164

super user. This suggests that peer recommendation may influence academics in gathering and disseminating research on social media, but not the frequency of using social media for publishing research updates. Institutional encouragement had a positive association with academic social media use in general, but no association with being a super user. This suggests that institutional encouragement may also influence academics in trying out social media. Whether a person had received social media training seemed to be unrelated to the likelihood of social media research use, but it could be important for super users. Having received social media training was not associated with academic social media use in general or for publishing research updates, but it had some positive association with being a super user. Without controlling for attitudes towards the effects of using social media, an academic who had received social media training was more likely to be a super user. This suggests that among those who are already using social media to disseminate their research updates, those who have received social media training might be more likely to have developed their skills and thus become more inclined to post frequently. Of course it can also be the other way around that those who are interested in publishing research updates would do training to improve their skills.

Two explanatory variables measure the concerns for the public good – ‘consideration for dissemination of research findings to the public’ and ‘attitudes towards the social media benefits for public good’. Both variables had positive associations with academic social media use and for publishing research updates. Academics who considered that the general public should know about their research findings were more likely to communicate research on social media. Attitudes towards the social media benefits for the public good indicate how much academics consider the benefit of using social media for the general public instead of individuals. Those with positive views towards the social media benefits for the public good were more likely to use social media in general and for publishing research updates as well as being super users.

Attitudes towards the social media benefits for individuals had the strongest association out of the three variables representing attitudes towards social media. This suggests that beliefs regarding individual benefits of using social media in research are very much related to its use. Perhaps unsurprisingly academics who believed that social media could bring them individual benefits were much more likely to use social media in general and to publish ongoing research updates. They were also more likely to become a super user instead of an occasional user. Those who agreed that social media could bring risks, such as leaking results to competitors or plagiarism, were less likely to use social media in research work or to publish research updates. On the other hand, academics that used social media in research work were more likely to have 165

positive attitudes towards the effects of using social media. Attitudes and practice are likely to influence each other. More participation in using social media in research may enhance positive views, foster trust and reduce negative views.

6.5.7 Limitations In this chapter, the variable of social media experience only took into account three key social media tools and excluded the use of other social media devices. Although Twitter, Facebook and blogs were popular social media services for academic use, it is possible that some academics might use other social media services instead. Another limitation is that the binary outcome variable coded all those who ‘sometimes’, ‘often’ or ‘always’ used social media tools as 1; an academic, therefore who uses Twitter everyday to disseminate research updates is regarded as the same as someone who occasionally reads blogs. This method of analysis cannot explore the intensity of using social media.

The associations between explanatory variables and the outcome variables do not necessarily imply causation. In some cases, the outcome variables might have been influenced by the explanatory variables and in other cases, it could be the other way around or the two variables influence each other. For example, having used secondary data collected by others might make someone more likely to reciprocate and share their own research data. On the other hand, academics might have deposited primary research data when they were required to by certain journals and in doing so may have realised that there were other valuable datasets that they could use from that repository. Other possible causes such as demands from colleagues or readers to share their data were not investigated in this study.

There was the possibility of endogeneity in the models, for example, between the awareness of RCUK’s OA policy and awareness of OA repositories in the Gold and Green models. Those who were aware of OA policy were also likely to have learnt about the existence of OA repositories as an important part of the OA policy. Another limitation is that respondents who had not nor did not produce research articles were filtered out for the models exploring Gold and Green OA publishing and sharing data. Thus many PhD students and early career academics would have been excluded from the analysis.

The models of being a super user compared to an occasion user only included 432 cases and thus would have an increased sampling error.

166

7. Chapter Seven: Conclusion and Discussion

7.1 Key findings

This thesis has explored new forms of scholarly communication and the practice of open science among UK-based academics. Open science broadly refers to practices that allow cost-free open access to academic research. Three aspects of open science were examined in this study: open access to research articles, sharing primary research data and publishing ongoing research updates on social media. The thesis has explored academics’ experiences and attitudes towards these three aspects of open science and the use of social media in research work. Overall this thesis has shown that whilst there is support for open science, the use of open science by academics was limited. Many academics were not aware of RCUK’s open access policy and had limited experience of making their research articles freely accessible online. Most academics did not share their primary research data online. Although some academics had used a range of social media tools to communicate their research, the majority had not used social media in their research work.

Overall, male, older and senior academics were more likely to use open access publishing and share primary research data, but were less likely to use social media for research. Academics based in Medical and Natural Sciences were more likely to use open access publishing and share research data, but less likely to use social media for their research compared to academics from Humanities and Social Sciences. Academics who were aware of RCUK’s open access policy and who recognised the citation advantages of open access were more likely to publish in open access journals. Academics that were aware of RCUK’s open access policy and had used social media for research were more likely to self-archive research articles. Academics that had used secondary data collected by others and self-archived research papers were more likely to share their own primary research data. Academics seemed to be strongly influenced by their colleagues’ recommendation for the adoption of social media in research. Those who considered that the general public should know about their research findings were more likely to share their research on social media. A group of academics were identified and described as super users who frequently communicated ongoing research on social media. These super users were more likely to use tablet computers and have received social media training organised by their institutions.

7.1.1 OA publishing 167

OA publishing may improve accessibility to publicly funded research, avoid delay of research dissemination, and disseminate research outputs to a wider audience. For individual academics, publishing through OA channel may improve visibilities and recognition for their work. Overall most academics agreed with the principle of making knowledge freely available to everyone, but less of them had OA publishing experience. Many academics were critical of publishers making profit of publicly funded research which were not available to the public. A good number of respondents in this study commented that publically funded research outputs should be freely accessible to the public. However, many academics had little awareness of the existence of OA repositories and OA policies. Other barriers for OA publishing are perhaps related to the high expense of article processing charges (APC), quality concerns about OA journals, copyrights related to self-archiving, and concerns about the misinterpretation of work by non-academic audiences.

Many academics, especially early career researchers and PhD students, are not necessarily funded by the RCUK and thus have no access to the RCUK’s block grants to pay for APCs. Most academics seemed to prioritise journals with a high reputation in their field rather than open access journals. The evidence suggests that the reputation and citation impact of the journals remain key factors for decision making regarding choice of journal. The goal of OA publishing in relation to the improvement of visibility and readership has been recognised by academics. The majority of respondents in this study agreed that open access articles would receive more citations and those academics who believed in this potential citation increases were more likely to publish in OA journals. However, since neither OA publishing models have been established for a long time, it is not easy to confirm whether there is a true citation advantage and this remains a debate in the literature.

The Green OA model provides opportunities for financially disadvantaged researchers to self- archive their work if they cannot afford APCs for the Gold model. UK academics have grown less resistant towards Green OA publishing in the last eight years. However, many academics still have little awareness of the existence of OA repositories. As the research assessments after the 2014 REF requires research outputs accepted for publication after 1 April 2016 to comply with HEFCE’s open access policy, academic institutions need to promote OA policy and its implications to their academics.

Positive attitudes towards OA publishing and awareness of RCUK’s OA policy are both positively associated with the likelihood of Gold and Green OA publishing. Academics that use social media in their research work were more likely to use Green OA publishing but not Gold OA 168

publishing. Some possible reasons were academics’ interests in new technologies and social media as an information source to the existence of online repositories. However, more studies are needed to explain the association between them.

Age differences in both Gold and Green OA publishing were evident after controlling for awareness of RCUK’s OA policy and other factors. This is probably because older and senior academics have more experience of publishing in general, more experience of funded projects with OA requirements, and are more likely to have access to funds to pay for author fees. There were gender differences in terms of publishing in Gold OA but not Green OA. Men were more likely to have senior jobs and thus would be more likely to work on funded projects with OA requirements and have access to funds to pay for author fees for Gold OA. With regard to self- archiving research articles, the gender difference may be explained by age (seniority) differences.

7.1.2 Sharing primary research data Sharing primary research data may improve accessibility to publicly funded research data, advance scientific progress and encourage collaboration. Most academics recognised the importance of sharing research data but most of them had never shared or reused research data. Many academics acknowledged the rationale of sharing research data and recognised the goals such as providing accessibility of public funded data to the public, validating research findings and avoiding duplicate data collection. However, many academics seemed to prioritise securing academic rewards and sharing data might work against that priority. Other barriers identified included the time and effort required to produce detailed meta-data alongside the primary data, ethical and confidentiality issues and uncertainty of the benefits and usefulness of sharing primary data.

The analysis has shown that positive attitudes towards the importance of data sharing are associated with such activities. In relation to the activities of using secondary data, re-using data collected by others might influence academics to be reciprocal and share their own research data. Although it was also possible that these academics generally liked to share and they might come across others’ data after they shared their own on repositories. Academics with experience of depositing research articles were more likely to deposit data, which might be part of funding requirements or being more skilled in how to use repositories to deposit research data. The differences in data-sharing experiences between those who published in OA journals and those who did not, as suggested by Piwowar (2011), could be explained by experience with self- archiving. 169

Older and senior academics have more experience with research and publishing in general and thus may have accumulated more research data to share. On the other hand, younger and junior academics were in greater need of securing publication and funding to advance their career whilst sharing primary research data might jeopardise their chances of publishing before competitors.

7.1.3 Using social media for research Using social media to publish ongoing research may increase public accessibility of research process and improve the efficiency of scientific knowledge production. Academic bloggers in this study recognised the goals of blogging about ongoing research such as keeping a record of detailed research progress and getting timely feedback. Academics who aim to engage stakeholders are able to keep agencies and potential beneficiaries engaged in the study throughout. This offers other academics, policy makers, service users and members of the general public to potentially participate in research process and discussion. Most academics in this study also agreed that communicating research on social media accelerated scientific discovery.

In general, the majority of academics did not use Twitter, Facebook, research blogs, YouTube, LinkedIn or Academia.edu in their research work. However, academic users of Twitter, LinkedIn and Academia.edu have grown from 2009 to 2013. Since the contribution of scholarly work in social media has not been recognised by the academic reward system, the majority of academics still view the traditional distribution channels as most important and many hold sceptical attitudes towards publishing research findings on social media. However, as outlined in Chapter Two, the potential impact of using social media on the citation of research can be substantial. The goal of disseminating publication information on social media in relation to readership and citation increases has been recognised by many academics. Many academics in this study agreed that blogging or tweeting about their publication would increase citations and over 30% of those who had published articles posted the article title and link in research blogs, Twitter or social networking sites.

Most academics recognised the rationale of communicating research on social media in relation to public engagement. The majority of academics in this study agreed that communicating research on social media benefits the public. This can be related to the debates that lay public audience are not able to understand scientific journal articles as we discussed in Chapter Two. Academics can present ideas in a far less formal setting on blogs and social networking sites in a less technical language which can be accessed freely by members of the public. Other rationale and purposes for using social media including finding useful research 170

information, networking with colleagues and potential collaborators and building personal reputation were also recognised by academics in this study. Many academics acknowledged the benefits of finding funding and collaboration, promoting professional profile and benefiting their career through using social media.

However, there were a large number of respondents who were unsure of the benefits and risks related to using social media in research work. The lack of knowledge over the precise benefits and risks of using social media in research remained an important barrier for those activities. Moreover, many academics agreed that research published on social media was not as trustworthy and communicating early-stage ideas might bring certain risks.

Smartphones were useful for reading information from social media but not convenient for posting long messages or writing blogs. As outlined, the research has identified a small group of super users and they were more likely to use tablet computers as they had bigger screens and were easier to type words for posting messages. As such, social media enthusiasts may be more inclined to purchase tablet computers to support their habits. Academics were strongly influenced by their colleagues’ recommendations to adopt new technology including posting research updates on social media. Academics could also be influenced by promotion from their institutions to try out social media. Many super users had received social media trainings which might have improved their skills and thus made them more devoted to frequently posting research updates. All in all, word of mouth by one’s trusted social network is very important to promote the open science movement. Institutional promotion of social media including providing training to academics was associated with greater adoption.

The behaviour of sharing research on social media reflects individual characteristics of being concerned for the public good, although academics were most concerned about how social media could benefit them as individuals. Attitudes and practice are likely to influence each other. More participation in using social media in research would have helped academics gain more understanding and perhaps build greater trust in the longer term.

Gender differences suggest that women preferred to build connections on Facebook and Twitter while men were fond of expressing opinions on blogs. Older and senior academics had more negative views towards social media and used it less.

7.1.4 Disciplinary differences 171

Discipline differences in publication forms, authorship, productivity, target audience and technical difficulties of research outputs as discussed in Chapter Two could influence academics’ attitudes and behaviours of supporting open science.

Discipline norms and culture would have influenced academics’ opportunities and preference of OA publishing. The Gold OA model seemed to be well-established amongst academics in the Medical and Life Sciences and Green OA repositories more common in Natural Sciences and Engineering. As scholars in Humanities and Social Sciences have fewer publications than colleagues in Medical and Natural Sciences, the chance of publishing through OA channels would also be lower. As academics in Medical and Natural Sciences are more likely to collaborate in teams, one or more of their team members might have the habit of depositing papers in repositories or have the funding to publish in OA journals. In terms of publication forms, academics in Humanities and Social Sciences are more likely to publish monographs and there is a lack of availability of OA journals and repositories in Humanities compared to Hard Sciences disciplines. Publishing an open access monograph is very costly and will require a sustainable business model.

Academics in Natural Sciences and Engineering are more likely to share their research data as they have more experience of self-archiving research articles in online repositories and are more likely to regard data sharing as very important. Discipline difference in sharing data can be related to established data policies and availability of data repositories in Science disciplines. Compared to the standardization of research publication as in article format, the format of research data vary largely between different disciplines across Sciences and Humanities. Those disciplines with a longer history of data sharing culture have established standardisation of datasets. As discussed in Chapter Two, Biomedical Sciences, Environmental Sciences, Physical Sciences and Social Sciences, researchers may have more resources and skill sets to store their primary research data. Many academics in Humanities are unable to share data because they have no primary data or their data are not reusable. Academics in disciplines that involve human subjects may be reluctant to share research data because of ethical issues.

Academics in Humanities and Social Sciences are more likely to use social media in research work in general and to publish ongoing research updates as well as being super users compare to colleagues in Medical and Natural Sciences. Since academics in Humanities and Social Sciences are more likely to work alone or in smaller groups, they would have more freedom to choose the way to communicate their ongoing research updates. As academics in Medical and Natural Sciences are likely to work in bigger teams, an individual might not have the freedom or feel 172

uncomfortable to share ongoing research updates in respect to collaborators. Academics in Medical and Natural Sciences might also be more concerned about revealing results to competitors. If too much information is provided, experiments can be reproduced and priority might be claimed by competitors. Unlike Hard Sciences, some Arts and Humanities disciplines produce art work or monographs as research outputs and their ongoing research updates cannot be used easily by others to produce ‘results’.

The differences in target audience and technical difficulties of research outputs are also likely to influence academics in different disciplines in terms of communicating research on social media. The research impact of Humanities and Social Sciences often requires direct engagement of wider publics to enhance their learning and well-being or to influence policy making which might have motivated individual academic and research centres to start use social media for public engagement. There are less technical difficulties of understanding research outputs for Humanities and Social Sciences compared to Hard Sciences. The dominant view that majority of lay public members are not able to understand scientific research outputs still influence academics’ science communication practice. It also requires strong communication skills to write about complex scientific knowledge in a language that is easily understandable by the lay public members.

7.2 Limitations

This study included extensive scoping studies and a large scale survey, which was one of the first of its kind to look at academic attitudes towards open science. However the study does have limitations.

The survey only sampled half of the Russell Group universities (12) and emails were only harvested from those who had a profile on their institution’s website. Those who did not have emails on the university websites would have been missed out because of the sampling strategy. The response rate was only 4.4%, although that is common for this kind of Internet survey. The results cannot represent academics from all institutions in the UK. However, the sample is large enough to provide some insight into the attitudes and behaviour of academics.

The statistical modelling analysis takes into account only three key social media tools and excludes the use of other social media devices. This may have excluded those who specifically used other social media tools instead of the three key ones investigated in this study. There is also 173

a limitation of those social media tools provided in questionnaire, some of which have quite different functions from others. For example, users of Google Drive/Doc might only use it for accessing shared documents and thus it has higher percentage of users than Twitter which requires more effort of networking with other Twitter users. Another limitation is that the binary outcome variable coded ‘sometimes’, ‘often’ and ‘always’ using a social media tool as 1. Therefore this method of analysis cannot explore the intensity of using social media. Moreover the analysis cannot establish causal pathways. In some cases, the outcome variables might have been influenced by the explanatory variables and in other cases, it could be the other way around or the two variables influence each other. For example, positive personal views towards citation impact of OA articles might not be the cause that someone had published in OA journals. Positive personal views could be caused by positive experiences of Gold OA publishing where the authors may have noticed increases in readership and citations. Some findings need to be further investigated by possible qualitative interviews with early adopters who already have experience with social media and open access publishing.

7.3 Implications and ways forward

Overall it seems a range of factors are associated with the likelihood of academics supporting and using open science. This includes background characteristics of their gender, discipline, age and research experience. It also includes the access to a tablet computer, the recommendation from one’s trusted networks and training offered by their institution. Many academics seem concerned about the risks of publishing their research openly including as rolling updates. However in the longer term it is clear that open science is going to be a major factor in academic work and in relation to building an academic career. The number of highly engaged super users is likely to grow in the future.

The awareness of OA policy and OA repositories is highly important for enhancing the use of Gold and Green OA. The survey results suggested many academics had little awareness of the existence of OA repositories or RCUK’s OA policy. This suggests that the promotion for this policy in these Russell Groups institutions had not yet been enforced, three months after the policy came into effect. Many survey respondents were not aware of the existence of OA repositories. Institutions with their own OA repositories need to promote their service to be known and understood by more academics if they want their service to be fully used. Institutions without OA repositories need to find alternative ways to help their academic staff comply with RCUK and 174

HEFCE’s OA policies in the future. As HEFCE’s OA policy is related to the research assessments post 2014 REF, UK higher education institutions need to promote OA policy to ensure it is known and understood by academics including those who are not funded by RCUK or HEFCE but who might be in the future. Institutions should reinforce these OA polices to academics through various channels including training especially for those at the start of their careers such as PhD students and research assistants who are more likely to be unaware of those policies as suggested by the findings of this study. Similarly, data-sharing policies need to be promoted especially to early career researchers who were less likely to be aware of these policies.

Since some academics may have concerns about the quality of open access journals, research funding bodies could have guidelines for what kind of journals academics should consider and not consider. It is clear open access journals need to be quality assured in the same way that closed access journals are, including peer review. Funds need to build up on reliable, high-quality OA journals for various disciplines.

As discussed in Chapter Two, studies of OA articles suggested contradictory findings with some evidence supporting the hypothesis of open access increasing citations and other evidence refuting this. The survey findings also suggest that those who had self-archived their research articles were not necessarily convinced that doing so would increase the citation of their articles. Incentives seem to be very important to academics’ publishing practice. Citation impact is highly relevant to academics’ career advancement and it is clear more robust analysis is needed for investigating the citation effects of OA articles. Other impacts of OA publishing need to be studied. For example, has the increase of OA availability helped the public learn about science and improved the public’s satisfaction with science communication? Once positive and reliable impacts of OA publishing are found, funders and academic institutions need to inform academics along with mandatory OA policies.

The academic community could benefit from open access to research data to validate findings and accelerate scientific progress. However barriers such as lack of incentives and standards could prevent academics from sharing. An incentive system and approach to the citing of data and databases are needed to promote data sharing in the future. Developing and adopting widely- recognised, usable technological and descriptive standards, can avoid the need for individual database-to-database negotiation, and can lower the barriers of technical difficulties such as format differences, while diverse models of data sharing and collaborations should be encouraged and supported by stakeholders (Gardner et al. 2003). Standards should be developed for sub- disciplines so that scientists have common and uniform data-sharing practices in their own 175

research community (Cragin et al. 2010). Stakeholders should also develop approaches to disseminate and promote data-sharing policies and standards, as many scientists are unaware of existing policies and standards. Knowing the related policies and standards early in the research cycle can help researchers identify sharable information and facilitate datasets in an appropriate format for deposition (ibid). Stakeholders should also fund and maintain infrastructure for data sharing, including providing training and support for researchers who intend to share data (Piwowar et al. 2008). Academics would also be more willing to share primary research data if requested by a funding agency or journal policy, or if made aware of the potential benefits such as increased citation impact.

Data policies are more established in some subject areas than the others and studies on developing strategies to encourage data sharing mainly focused on biomedical areas. As the format and volume of research data vary largely between and within disciplines, further studies of data sharing need to focus on individual discipline areas and especially those that have not been studied systematically.

For those who communicate research processes on blogs, in order to make this approach work effectively, more direct engagement in blogging from the scholarly communities will be needed to process feedback and evaluation. While smartphones are useful for academics to read useful information for their research, tablet computers can provide academic social media users a convenient mobile platform to disseminate research updates frequently. If an academic institution would like to encourage open science and public engagement on social media, they should provide more funding for academics to access tablet computers.

Peer recommendation, institutional encouragement and social media training are all important for promoting academic use of social media. It seems academics are mostly likely to start using a new media tool because their colleagues tell them it is worth the time and effort. Promotion from the institutions is also important to encourage academics to try out social media tools. Social media training is helpful for those who want to improve their skills for communicating research. Universities should provide resources and funding for academics that wants to set up research blogs or uses other social media tools to communicate their research. Universities will need to find strategies to promote social media use and training, as many academics reported not knowing about any promotion or training opportunities. For those who are resistant to social media only close colleagues’ recommendations and mandate by their institutions may help change their minds. 176

With the right guidance and reinforcement of relevant policies, the new forms of scholarly communication can provide a pathway to open science which would serve to benefit individual academics, research communities and the public good.

7.4 Future work

This study has found that those who have used social media in research work are more likely to have used Green OA publishing but not Gold OA publishing. One of the possible explanations is the characteristics of an academic being more open to adopt new technologies and new media tools. It is also possible that academics who use social media would be more likely to learn about the repositories function of social network sites such as Academia.edu and ResearchGate. More studies are needed to explain the association between them. Further statistical analysis on each of the individual social media tools can be carried out. Moreover, follow-up qualitative interviews may provide a more detailed explanation as to associations found between the use of open science and other factors in this study. Given the changes in publishing and open access in the planned forthcoming research assessment in the UK and the evolution of new tools for communication and information sharing there are clearly challenges and opportunities ahead for academics and universities and the pathways to open science.

177

References

Alexandrov, G. A. (2011). The meaning of the 'impact factor' in the case of an open-access journal. Carbon Balance Manag, 6(1), 1. Almousa, O. (2011). Users' classification and usage-pattern identification in academic social networks. Applied Electrical Engineering and Computing Technologies (AEECT), 2011 IEEE Jordan Conference on. IEEE. Anderson, N. & Edwards, K. (2010). Building a chain of trust: using policy and practice to enhance trustworthy clinical data discovery and sharing. Proceeding of the 2010 Workshop on Governance of Technology, Information and Policies, 15-20. Antelman, K. (2004). Do open-access articles have a greater research impact? College & Research Libraries, 65(5), 372-382. Antelman, K. (2006). Self-archiving practice and the influence of publisher policies in the social sciences. Learned Publishing, 19(2), 85-95. Antin, J. & Cheshire, C. (2010). Readers are not free-riders: reading as a form of participation on wikipedia. Proceedings of the 2010 ACM conference on Computer supported cooperative work. ACM. Attwood, R. (2012). Satisfaction and its discontents [Online]. Times Higher Education, 8 March 2012. Available: http://www.timeshighereducation.co.uk/419238.article [Accessed 24 March 2014]. Babbie, E. (2004). The Practice of Social Research, 10th edition. Belmont: Wadsworth. Bar-Ilan, J., Haustein, S., Peters, I., Priem, J., Shema, H. & Terliesner, J. (2012). Beyond citations: Scholars' visibility on the social Web. Arxiv preprint: 1205.5611. Barnes, S. & Lewin, C. (2005). An introduction to inferential statistics: Testing for differences and relationships. In: Somekh, B. & Lewin, C. (eds.) Research Methods in the Social Sciences. London: Sage. Belter, C. (2014). Global-level data sets may be more highly cited than most journal articles [Online]. London School of Economics and Political Science. Available: http://blogs.lse.ac.uk/impactofsocialsciences/2014/05/15/global-level-data-sets-highly-cited/ [Accessed 1 May 2015].

Besley, J. C. (2014). What do scientists think about the public and does it matter to their online engagement? Science and Public Policy. doi:10.1093/scipol/scu042.

Björk, B. C. (2004). Open access to scientific publications–an analysis of the barriers to change? Information Research, 9(2), 170. Björk, B. C., Welling, P., Laakso, M., Majlender, P., Hedlund, T. & Guonason, G. (2010). Open access to the scientific journal literature: Situation 2009. PLoS One, 5(6). Björk, B. C. (2012). The hybrid model for open access publication of scholarly articles: A failed experiment? Journal of the American Society for Information Science and Technology, 63(8), 1496-1504.

Bohannon, J. (2013). Who's afraid of peer review? Science, 342(6154), 60-65. Bonetta, L. (2009). Should you be tweeting? Cell, 139(3), 452-453. Borgman, C. L. (2007). Scholarship in the digital age: Information, infrastructure, and the internet: The MIT Press. Børing, P., Flanagan, K., Kaloudis, A. & Karakasidou, A. (2010). Mobility survey of the higher education sector. Report. [Online]. Available: https://www.escholar.manchester.ac.uk/uk-ac-man-scw:158492/ [Accessed 10 June 2015]. 178

Bornmann, L., Neuhaus, C. & Daniel, H. D. (2011). The effect of a two-stage publication process on the Journal Impact Factor: a case study on the interactive open access journal Atmospheric Chemistry and Physics. Scientometrics, 86(1), 93-97. Borrego, Á. & Fry, J. (2012). Measuring researchers’ use of scholarly information through social bookmarking data: A case study of BibSonomy. Journal of Information Science, 38(3), 297-308. Boulton, G., Rawlins, M., Vallance, P. & Walport, M. (2011). Science as a public enterprise: the case for open data. The Lancet, 377(9778), 1633-1635. Braun, V. & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77-101. Brown, E. (2009). A season of change: How science librarians can remain relevant with open access and scholarly communications Initiatives. Issues in Science and Technology Librarianship, 5. Bryman, A. (2008). Why do researchers integrate/combine/mesh/blend/mix/merge/fuse quantitative and qualitative research. Advances in Mixed Methods Research, 87-100. Bryman, A. (2012). Social Research Methods. Oxford: Oxford University Press. Bucchi, M. (2004). Science in Society: An Introduction to Social Studies of Science. London and New York: Routledge.

Bukvova, H. (2011). Taking new routes: Blogs, web sites, and scientific publishing. ScieCom Info, 7(2). Calver, M. C. & Bradley, J. S. (2010). Patterns of citations of open access and non-open access conservation Biology journal papers and book chapters. Conservation Biology, 24(3), 872-880. Campanelli, P. (1997). Testing survey questions: New directions in cognitive interviewing. Bulletin of Sociological Methodology/Bulletin de Méthodologie Sociologique, 55, 5-17. Ceci, S. J. & Williams, W. M. (2011). Understanding current causes of women's underrepresentation in science. Proceedings of the National Academy of Sciences, 108(8), 3157-3162. Chan, L. & Costa, S. (2005). Participation in the global knowledge commons: Challenges and opportunities for research dissemination in developing countries. New Library World, 106(3/4), 141-163. Chan, L. & Kirsop, B. (2001). Open archiving opportunities for developing countries: Towards equitable distribution of global knowledge. Ariadne. Chevalier, A. & Conlon, G. (2003). Does it pay to attend a prestigious university? London: Centre for the Economics of Education, London School of Economics and Political Science. Cloitre, M. & Shinn, T. (1985). Expository practice: Social, cognitive and epistemological linkage. In: Shinn, T. & Whitley, R. P. (eds.) Expository Science: Forms and Functions of Popularisation. Springer Science & Business Media.

Cobbledick, S. (1996). The information-seeking behavior of artists: Exploratory interviews. The Library Quarterly, 343-372. Cole, S. & Cole, J. R. (1967). Scientific output and recognition: A study in the operation of the reward system in science. American Sociological Review, 32(3), 377-390. Correia, A. M. R. & Teixeira, J. C. (2005). Reforming scholarly publishing and knowledge communication: From the advent of the scholarly journal to the challenges of open access. Online Information Review, 29(4), 349-364. Coverdale, A. (2011). Negotiating doctoral practices and academic identities through the adoption and use of social and participative media. Proceedings of the10th European Conference on e-Learning. 10- 11 Nov 2011. Brighton, UK. Cox, B. & Forshaw, J. (2011). Brian Cox and Jeff Forshaw explain the big bang [Online]. Guardian. Available: http://www.guardian.co.uk/science/2011/oct/23/brian-cox-jeff-forshaw-answers [Accessed 6 Feb 2012]. 179

Cragin, M. H., Palmer, C. L., Carlson, J. R. & Witt, M. (2010). Data sharing, small science and institutional repositories. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 368(1926), 4023-4038. Craig, I. D., Plume, A. M., McVeigh, M. E., Pringle, J. & Amin, M. (2007). Do open access articles have greater citation impact? A critical review of the literature. Journal of Informetrics, 1(3), 239-248. Cressey, D. (2009). Open access: Are publishers ‘double dipping’? [Online]. Nature News Blog. Available: http://blogs.nature.com/news/2009/10/story_landis_resigns_from_auti.html [Accessed 18 May 2015].

Creswell, J. W., Plano Clark, V. & Garrett, A. L. (2008). Methodological issues in conducting mixed methods research designs. Advances in Mixed Methods Research, 66-83. Crotty, D. (2011). Researchers and social media: Uptake increases when obvious benefits result [Online]. Available: http://scholarlykitchen.sspnet.org/2011/03/01/researchers-and-social-media-uptake- increases-when-obvious-benefits-result/ [Accessed 22 Oct 2012]. Crow, J. (2008). Open access and scholarly communication. Open and Libraries Class Journal, 1(1). Crush, J., Eberhardt, C., Caesar, M., Chikanda, A., Pendleton, W. & Hill, A. (2012). 16 Diasporas on the web: New networks, new methodologies. Handbook of Research Methods in Migration, 345. Darley, R., Reynolds, D. & Wickham, C. (2014). Open access journals in humanities and social science: A British Academy research project [Online]. Available: http://www.britac.ac.uk/openaccess/index.cfm [Accessed 24/04/2014]. David, P. A. (1998). Common agency contracting and the emergence of 'open science' institutions. American Economic Review, 88(2), 15-21. David, P. A. (2008). The historical origins of 'open science': an essay on patronage, reputation and common agency contracting in the scientific revolution. Capitalism and Society, 3(2). Davies, S. R. (2008). Constructing communication: Talking to scientists about talking to the public. Science Communication, 29(4), 413-434. Davis, P. M. (2009). Author-choice open-access publishing in the biological and medical literature: A citation analysis. Journal of the American Society for Information Science and Technology, 60(1), 3-8. Davis, P. M. (2010). Does open access lead to increased readership and citation. The Physiologist, 53, 197- 201. Davis, P. M. (2011). Open access, readership, citations: a randomized controlled trial of scientific journal publishing. The FASEB Journal, 25(7), 2129-2134. Davis, P. M. & Fromerth, M. J. (2007). Does the arXiv lead to higher citations and reduced publisher downloads for mathematics articles? Scientometrics, 71(2), 203-215. Davis, P. M., Lewenstein, B. V., Simon, D. H., Booth, J. G. & Connolly, M. J. L. (2008). Open access publishing, article downloads, and citations: randomised controlled trial. British Medical Journal, 337. Davis, P. M. & Walters, W. H. (2011). The impact of free access to the scientific literature: a review of recent research. Journal of the Medical Library Association, 99(3), 208-17. De Roure, D., Goble, C., Aleksejevs, S., Bechhofer, S., Bhagat, J., Cruickshank, D., Fisher, P., Hull, D., Michaelides, D., Newman, D., Procter, R., Lin, Y. W. & Poschen, M. (2010). Towards open science: The myExperiment approach. Concurrency and Computation-Practice & Experience, 22(17), 2335- 2353. Deer, B. (2010). Wakefield’s ‘autistic enterocolitis’ under the microscope. BMJ: British Medical Journal, 340, c1127. Dehdarirad, T., Villarroya, A. & Barrios, M. (2014). Research trends in gender differences in higher education and science: A co-word analysis. Scientometrics, 101(1), 273-290. DeVellis, R. F. (2003). Scale development: Theory and applications, 2nd edition. Thousand Oaks, CA: Sage. 180

Dillon, I. F. & Hahn, K. L. (2002). Are researchers ready for the electronic-only journal collection? Results of a survey at the University of Maryland. Portal-Libraries and the Academy, 2(3), 375-390. Ding, W. W. (2011). The Impact of founders' professional-dducation background on the adoption of open science by for-profit biotechnology firms. Management Science, 57(2), 257-273. Dutton, W., Di Gennaro, C. & Millwood Hargrave, A. (2005). Oxford internet survey 2005 report: The internet in Britain. Available at SSRN 1327035. eBiZMBA. (2014). Top 15 most popular social networking sites May 2014 [Online]. Available: http://www.ebizmba.com/articles/social-networking-websites [Accessed 6 Oct 2014]. Ebner, M., Mühlburger, H., Schaffert, S., Schiefner, M., Reinhardt, W. & Wheeler, S. (2010). Getting granular on Twitter: Tweets from a conference and their limited usefulness for non-participants. Key Competencies in the Knowledge Society, 102-113. Edgar, B. D. & Willinsky, J. (2010). A survey of scholarly journals using Open Journal Systems. Scholarly and Research Communication, 1(2). European Commission (2013). The prospects of Science 2.0: Policy Implications. Manchester Business School. Evans, J. R. & Mathur, A. (2005). The value of online surveys. Internet Research, 15(2), 195-219. Eysenbach, G. (2006). Citation advantage of open access articles. PLoS Biology, 4(5), e157. Eysenbach, G. (2011). Can tweets predict citations? Metrics of social impact based on twitter and correlation with traditional metrics of scientific impact. Journal of Medical Internet Research, 13(4). Eysenbach, G. & Sa, E. R. (2001). Letter: Code of conduct is needed for publishing raw data. BMJ, 323(7305), 166. Ferguson, C., Marcus, A. & Oransky, I. (2014). Publishing: The peer-review scam. Nature, 515(7528), 480- 482 Field, A. (2009). Discovering Statistics Using SPSS, third edition. London: Sage Fontana, A. & Frey, J. (1994). The art of science. In: Denzin, N. K. & Lincoln, Y. S. (eds.) The Handbook of Qualitative Research. London: Sage. Ford, K. C., Veletsianos, G. & Resta, P. (2014). The Structure and Characteristics of #PhDChat, an Emergent Online Social Network. Journal of Interactive Media in Education. 2014(1), 8. Fox, M. F. (2001). Women, science, and academia graduate education and careers. Gender & Society, 15(5), 654-666. Fox, M. F. (2010). Women and men faculty in academic science and engineering: Social-organizational indicators and implications. American Behavioral Scientist, 53(7), 997-1012. Friedman, B., Khan Jr, P. H. & Howe, D. C. (2000). Trust online. Communications of the ACM, 43(12), 34-40. Friemel, T. N. (2014). The digital divide has grown old: Determinants of a digital divide among seniors. New Media & Society, 12 June 2014, 1-19. Galesic, M. (2006). Dropouts on the web: Effects of interest and burden experienced during an online survey. Journal of Official Statistics, 22(2), 313-328. Gardner, D., Toga, A., Ascoli, G., Beatty, J., Brinkley, J., Dale, A., Fox, P., Gardner, E., George, J., Goddard, N., Harris, K., Herskovits, E., Hines, M., Jacobs, G., Jacobs, R., Jones, E., Kennedy, D., Kimberg, D., Mazziotta, J., Miller, P., Mori, S., Mountain, D., Reiss, A., Rosen, G., Rottenberg, D., Shepherd, G., Smalheiser, N., Smith, K., Strachan, T., Van Essen, D., Williams, R. & Wong, S. (2003). Towards effective and rewarding data sharing. Neuroinformatics, 1(3), 289-295. Garfield, E. (1976). Significant journals of science. Nature, 264(5587), 609-615. Garfield, E. (2006). Citation indexes for science. A new dimension in documentation through association of ideas. International Journal of Epidemiology, 35(5), 1123-1127. 181

Gargouri, Y., Hajjem, C., Lariviere, V., Gingras, Y., Carr, L., Brody, T. & Harnad, S. (2010). Self-selected or mandated, open access increases citation impact for higher quality research. PLoS One, 5(10), e13636. Gaule, P. & Maystre, N. (2011). Getting cited: does open access help? Research Policy, 40(10), 1332–1338. Gewin, V. (2010). Collaboration: Social networking seeks critical mass. Nature, 468(7326), 993-994. Gezelter, D. (2009). What, exactly, is open science? [Online]. The OpenScience Project. Available: http://www.openscience.org/blog/?p=269 [Accessed 9 February 2015]. Gill, J. (2013). Equity is the best policy [Online]. Times Higher Education. Available: http://www.timeshighereducation.co.uk/comment/leader/equity-is-the-best- policy/2003100.article [Accessed 11 April 2013]. Gould, T. (2009). The future of academic publishing: Application of the long-tail theory. Publishing Research Quarterly, 25(4), 232-245. Grand, A., Bultitude, K., Wilkinson, C. & Winfield, A. (2010). Muddying the waters or clearing the stream? Open Science as a communication medium. Science Communication Without Frontiers: 11th International Conference on Public Communication of Science and Technology. New Delhi: INPCST.

Grand, A., Wilkinson, C., Bultitude, K. & Winfield, A. F. (2014). Mapping the hinterland: Data issues in open science. Public Understanding of Science, 0963662514530374.

Gray, J. (2009). Jim Gray on eScience: A transformed scientific method. In: Hey, T., Tansley, S. & Tolle, K. (eds.) The Fourth Paradigm: Data-Intensive Scientific Discovery. Redmond: Microsoft Research. Greene, J. C., Caracelli, V. J. & Graham, W. F. (1989). Toward a conceptual framework for mixed-method evaluation designs. Educational Evaluation and Policy Analysis, 11(3), 255-274. Greene, J. C., Kreider, H. & Mayer, E. (2005). Combining qualitative and quantitative methods in social inquiry. In: Somekh, B. & Lewin, C. (eds.) Research Methods in the Social Sciences. London: Sage. Greenleaf, C., McGreer, R. & Parham, H. (2006). Physique attitudes and self-presentational concerns: Exploratory interviews with female group aerobic exercisers and instructors. Sex Roles, 54(3-4), 189-199. Gross, J. & Suttor, N. (2013). Getting found-using social media to build your research profile [Online]. Edith Cowan University. Available: http://ro.ecu.edu.au/creswk/60/ [Accessed 13 May 2014]. Groth, P. & Gurney, T. (2010). Studying scientific discourse on the Web using bibliometrics: A chemistry blogging case study. Web Science Conference. Raleigh, NC, USA. Grosseck, G., Bran, R. & Tiru, L. (2011). Dear teacher, what should I write on my wall? A case study on academic uses of Facebook. Procedia-Social and Behavioral Sciences, 15, 1425-1430. Gu, F. & Widén-Wulff, G. (2011). Scholarly communication and possible changes in the context of social media: A Finnish case study. The Electronic Library, 29(6), 762-776. Guédon, J. C. (2001). In Oldenburg's Long Shadow: Librarians, Research Scientists, Publishers, and the Control of Scientific Publishing. Washington, D.C.: Association of Research Libraries. Hagstrom, W. O. (1974). Competition in science. American Sociological Review, 1-18. Hahn, T., Burright, M. & Duggan, H. (2011). Has the revolution in scholarly communication lived up to its promise? American Society for Information Science and Technology, 37(5), 5. Hajjem, C., Harnad, S. & Gingras, Y. (2005). Ten-year cross-disciplinary comparison of the growth of open access and how it increases research citation impact. Bulletin of the IEEE Data Engineering Bulletin 28(4), 39-47. Hall, N. (2014). The Kardashian index: a measure of discrepant social media profile for scientists. Genome biology, 15(7), 424. 182

Hannay, T. (2009). From web 2.0 to the global database. In: T. Hey, Tansley, S. & Redmond, K. T. (eds.) The Fourth Paradigm: Data-Intensive Scientific Discovery. Microsoft Research. Hardigan, P. C., Succar, C. T. & Fleisher, J. M. (2012). An analysis of response rate and economic costs between mail and web-based surveys among practicing dentists: a randomized trial. Journal of Community Health, 37(2), 383-394. Harnad, S. & Brody, T. (2004). Comparing the impact of open access (OA) vs. non-OA articles in the same journals. D-lib Magazine, 10(6). Harnad, S., Brody, T., Vallières, F., Carr, L., Hitchcock, S., Gingras, Y., Oppenheim, C., Stamerjohanns, H. & Hilf, E. R. (2004). The access/impact problem and the green and gold roads to open access. Serials Review, 30(4), 310-314. Helsper, E. J. & Eynon, R. (2010). Digital natives: Where is the evidence? British Educational Research Journal, 36(3), 503-520. Henderson, M. (2010). End of the peer review show? BMJ: British Medical Journal, 340, 738-740. Henneken, E. A., Kurtz, M. J., Eichhorn, G., Accomazzi, A., Grant, C., Thompson, D. & Murray, S. S. (2006). Effect of e-printing on citation rates in astronomy and physics. Journal of Electronic Publishing, 9(2). Hermanowicz, J. C. (2012). The sociology of academic careers: Problems and prospects. Higher education: Handbook of theory and research. Springer. HESA. (2013). Higher education statistics for the United Kingdom 2011/12 [Online]. Available: http://www.hesa.ac.uk/ [Accessed 21 September 2013]. Hewitt, P. (1998). Technology and democracy. In: Franklin, J. (ed.) The Politics of Risk Society. Cambridge: Polity Press. Hewson, C. & Laurent, D. (2008). Research design and tools for internet research. In: Fielding, N., Lee, R. & Blank, G. (eds.) The SAGE Handbook of Online Research Methods. London: Sage. Hilgartner, S. (1990). The dominant view of popularization: conceptual problems, political uses. Social Studies of Science, 20(3), 519-539. Hodge, M. (2006). The fourth amendment and privacy issues on the new internet: Facebook.com and Myspace.com. Southern Illinois University Law Journal, 31, 95-123. Hogg, D. (2012). Hogg's research blog: Galaxies, stellar dynamics, exoplanets, and fundamental astronomy [Online]. Available: http://hoggresearch.blogspot.com/ [Accessed 6 Feb 2012]. Hole, J. D. (2008). Email Overload in Academia. Ann Arbor: ProQuest. Honeycutt, C. & Herring, S. C. (2009). Beyond microblogging: Conversation and collaboration via Twitter. Proceedings of the 42 Hawaii International Conference on System Sciences. Los Alamitos, CA: IEEE Computer Society. Hopkins, A. L., Jawitz, J. W., McCarty, C., Goldman, A. & Basu, N. B. (2013). Disparities in publication patterns by gender, race and ethnicity based on a survey of a random sample of authors. Scientometrics, 96(2), 515-534. Hosker, I. (2008). Statistics for Social Sciences, 2nd edition: Studymates. Howard, J. (2013). Posting your latest article? You might have to take it down [Online]. The Chronicle of Higher Education. Available: http://chronicle.com/blogs/wiredcampus/posting-your-latest-article- you-might-have-to-take-it-down/48865 [Accessed 13 December 2014]. IpsosMORI. (2011). Public attitudes to science 2011 [Online]. Ipsos. Available: http://www.ipsosmori.com/researchpublications/researcharchive/2764/Public-attitudes-to-science-2011.aspx [Accessed 3 Feb 2012]. Jarreau, P. B. (2014). In response to the top 50 science list [Online]. Astronomy Communications. Available: http://www.scilogs.com/from_the_lab_bench/in-response-to-the-top-50-science-list/ [Accessed 10 Oct 2014]. 183

Java, A., Song, X., Finin, T. & Tseng, B. (2007). Why we twitter: Understanding microblogging usage and communities. Proceedings of the 9th WebKDD and 1st SNA-KDD 2007 Workshop on Web Mining and Social Network Analysis. San Jose, CA. Jerejian, A., Reid, C. & Rees, C. S. (2013). The contribution of email volume, email management strategies and propensity to worry in predicting email stress among academics. Computers in Human Behavior, 29(3), 991-996. Jobe, J. B. & Mingay, D. J. (1989). Cognitive research improves questionnaires. American Journal of Public Health, 79(8), 1053-1055. Johnson, C. (2015). In science, irreproducible research is a quiet crisis [Online]. Boston Globe. Available: http://www.bostonglobe.com/ideas/2015/03/19/science-irreproducible-research-quiet- crisis/xunxnfuzwdwYSpVjkx2iQN/story.html [Accessed 28 April 2015]. Joliffe, I. & Morgan, B. (1992). Principal component analysis and exploratory factor analysis. Statistical Methods in Medical Research, 1(1), 69-95. Kaiser, H. F. (1960). The application of electronic computers to factor analysis. Educational and Psychological Measurement, 20, 141-151. Kaplan, A. M. & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of social media. Business Horizons, 53(1), 59-68. Kaye, J., Heeney, C., Hawkins, N., De Vries, J. & Boddington, P. (2009). Data sharing in genomics—re-shaping scientific practice. Nature Reviews Genetics, 10(5), 331-335. Kelly, B. (2011). Social analytics for Russell Group university Twitter accounts, 28 June 2011 [Online]. Available: http://ukwebfocus.wordpress.com/2011/06/28/social-analytics-for-russell-group- university-twitter-accounts/ [Accessed 18 April 2013]. Kelly, B. (2012a). Further evidence of use of social networks in the UK higher education sector, 6 June 2012 [Online]. Available: http://ukwebfocus.wordpress.com/2012/06/06/guest-post-further-evidence- of-use-of-social-networks-in-the-uk-higher-education-sector/ [Accessed 23 October 2012]. Kelly, B. (2012b). Links to social media sites on Russell Group university home pages, 18 January 2012, [Online]. Available: http://ukwebfocus.wordpress.com/2012/01/18/social-media-links-on-russellgroup-university-home-pages/ [Accessed 18 April 2013]. Kelly, B. & Delasalle, J. (2012). Can LinkedIn and Academia.edu enhance access to open repositories? OR2012: the 7th International Conference on Open Repositories, 9-13 July 2012. Edinburgh, Scotland. Khan, A. (2012). Web 2.0 in academia. International Journal of Applied Science and Technology, 2(6), 1-6. Kim, J. (2010). Faculty self-archiving: Motivations and barriers. Journal of the American Society for Information Science and Technology, 61(9), 1909-1922. Kim, J. (2011). Motivations of faculty self-archiving in institutional repositories. Journal of Academic Librarianship, 37(3), 246-254. Kirchner, A. (2013). What Are the Costs of an Open Access Monograph? [Online]. Hybrid Publishing Lab. Available: https://hybridpublishing.org/2013/12/what-are-the-costs-of-an-open-access- monograph/ [Accessed 4 May 2015]. Kjellberg, S. (2010). I am a blogging researcher: Motivations for blogging in a scholarly context. First Monday, 15(8). Kosmopolit. (2009). Blogging: A short introduction for academics [Online]. Available: http://kosmopolito.ideasoneurope.eu/2009/08/02/blogging-a-short-introduction-for-academics/ [Accessed 8 Feb 2012]. Kronick, D. A. (1990). Peer review in 18th-century scientific journalism. The Journal of the American Medical Association, 263(10), 1321. 184

Kurtz, M. J., Eichhorn, G., Accomazzi, A., Grant, C., Demleitner, M., Henneken, E. & Murray, S. S. (2005). The effect of use and access on citations. Information Processing & Management, 41(6), 1395-1402. Kyvik, S. (2003). Changing trends in publishing behaviour among university faculty, 1980-2000. Scientometrics, 58(1), 35-48. Laakso, M., Welling, P., Bukvova, H., Nyman, L., Björk, B. C. & Hedlund, T. (2011). The development of open access journal publishing from 1993 to 2009. PLoS One, 6(6). LaFollette, M. C. (1992). Stealing into print: Fraud, plagiarism, and misconduct in scientific publishing. Berkeley and Los Angeles: University of California Press. Laniado, D. & Mika, P. (2010). Making sense of Twitter. The Semantic Web–ISWC 2010, 470-485. Larivière, V., Archambault, É., Gingras, Y. & Vignola-Gagné, É. (2006). The place of serials in referencing practices: Comparing natural sciences and engineering with social sciences and humanities. Journal of the American Society for Information Science and Technology, 57(8), 997-1004. Lawrence, S. (2001). Free online availability substantially increases a paper's impact. Nature, 411(6837), 521-521. Letierce, J., Passant, A., Breslin, J. & Decker, S. (2010a). Understanding how Twitter is used to spread scientific messages. Web Science Conference. Raleigh, NC, USA. Letierce, J., Passant, A., Breslin, J. G. & Decker, S. (2010b). Using Twitter during an academic conference: The iswc2009 use-case. Proceedings of the Fourth International AAAI Conference on Weblogs and Social Media. Lewin, C. (2005). Elementary quantitative methods. In: Somekh, B. & Lewin, C. (eds.) Research Methods in the Social Sciences. London: Sage. Li, Y., Pickles, A. & Savage, M. (2005). Social capital and social trust in Britain. European Sociological Review, 21(2), 109-123. Lincoln, Y. S., Lynham, S. A. & Guba, E. G. (2011). Paradigmatic controversies, contradictions, and emerging confluences, revisited. In: Denzin, N. K. & Lincoln, Y. S. (eds.) The Sage Handbook of Qualitative Research. London: Sage. Lips, H. M. (2013). The gender pay gap: Challenging the rationalizations. Perceived equity, discrimination, and the limits of human capital models. Sex Roles, 68(3-4), 169-185. Look, H. & Pinter, F. (2010). Open access and humanities and social science monograph publishing. New Review of Academic Librarianship, 16(S1), 90-97. Louise Barriball, K. & While, A. (1994). Collecting Data using a semi-structured interview: a discussion paper. Journal of Advanced Nursing, 19(2), 328-335. Lucas, R. & Willinsky, J. (2009). Open access to e-research. In: Jankowski, N. (ed.) e-Research: Transformation in Scholarly Practice. New York: Routledge. Lupton, D. (2013). Opening up your research: Self-archiving for sociologists [Online]. Available: http://simplysociology.wordpress.com/2013/06/06/opening-up-your-research-self-archiving-for- sociologists/ [Accessed 29 August 2014]. Lyon, L. (2009). Open science at web-scale: Optimising praticipation and predictive potential consultative report [Online]. Available: http://www.jisc.ac.uk/media/documents/publications/research/2009/open-science-report- 6nov09-final-sentojisc.pdf [Accessed 18 July 2011]. Mann, C. & Stewart, F. (2000). Internet Communication and Qualitative Research: A Handbook for Researching Online. London: Sage. Maron, N. L. & Smith, K. K. (2008). Current Models of Digital Scholarly Communication: Results of an Investigation Conducted by Ithaka for the Association of Research Libraries. Washington, D.C.: Association of Research Libraries. 185

Marshall, M. N. (1996). Sampling for qualitative research. Family Practice, 13(6), 522-526. Martinson, B. C., Anderson, M. S. & De Vries, R. (2005). Scientists behaving badly. Nature, 435(7043), 737- 738. Mas-Bleda, A., Thelwall, M., Kousha, K. & Aguillo, I. (2014). Do highly cited researchers successfully use the social web? Scientometrics, 101(1), 337-356. Masnick, M. (2015). Don't Think Open Access Is Important? It Might Have Prevented Much Of The Ebola Outbreak [Online]. Techdirt. Available: https://www.techdirt.com/articles/20150409/17514230608/dont-think-open-access-is-important- it-might-have-prevented-much-ebola-outbreak.shtml [Accessed 11 May 2015] Mason, J. (2002). Qualitative researching, 2nd edition. London: Sage. May, T. (1993). Social research: issues, methods and process, 2nd edition. Buckingham: Open University Press. McDaniel, W. C. (2013). Open Notebook History [Online]. Available: http://wcm1.web.rice.edu/open- notebook-history.html [Accessed 21 May 2015]. McGrath, J. F. (2010). The blogging revolution: New technologies and their impact on how we do scholarship. Blogger and Online Publication Session at the Society of Biblical Literature Annual Meeting. Atkabta, 22 November 2010. McGuigan, G. & Russel, R. (2008). The business of academic publishing: A strategic analysis of the academic journal publishing industry and its impact on the future of scholarly publishing. The Electronic Journal of Academic and Special Librarianship, 9(3). McGuire, A. L., Hamilton, J. A., Lunstroth, R., McCullough, L. B. & Goldman, A. (2008). DNA data sharing: Research participants’ perspectives. Genetics in Medicine: Official Journal of the American College of Medical Genetics, 10(1), 46. Merton, R. K. (1957). Priorities in scientific discovery: a chapter in the sociology of science. American Sociological Review, 22(6), 635-659. Merton, R. K. (1968). The Matthew effect in science. Science, 159(3810), 56-63. Metcalfe, T. S. (2005). The rise and citation impact of astro-ph in major journals. Bull.Am.Astron.Soc. , 37, 555-557. Metcalfe, T. S. (2006). The citation impact of digital preprint archives for solar physics papers. Solar Physics, 239(1), 549-553. Meyer, E., Bulger, M., Kyriakidou-Zacharoudiou, A., Power, L., Williams, P., Venters, W., Terras, M. & Wyatt, S. (2011). Collaborative yet independent: Information practices in the physical sciences [Online]. Research Information Network & IOP Publishing. Available: www.rin.ac.uk/phys-sci-case [Accessed 6 February 2012]. Moed, H. F. (2007). The effect of "open access" on citation impact: An analysis of ArXiv's condensed matter section. Journal of the American Society for Information Science and Technology, 58(13), 2047- 2054. Moran, M., Seaman, J. & Tinti-Kane, H. (2011). Teaching, learning, and sharing: How today's higher education faculty use social media. [Online]. Babson Survey Research Group. Available: http://www.eric.ed.gov/ERICWebPortal/search/detailmini.jsp?_nfpb=true&_&ERICExtSearch_Sear chValue_0=ED535130&ERICExtSearch_SearchType_0=no&accno=ED535130 [Accessed 24 October 2012]. Moser, C. A. & Kalton, G. (2004). Survey Methods in Social Investigation, 2nd edition. Aldershot: Ashgate. Mukherjee, A. & Stern, S. (2009). Disclosure or secrecy? The dynamics of open science. International Journal of Industrial Organization, 27(3), 449-462. 186

Muller, M., Geyer, W., Soule, T. & Wafer, J. (2014). Geographical and organizational distances in enterprise crowdfunding. Proceedings of the 17th ACM Conference on Computer Supported Cooperative Work & Social Computing. Murthy, D. (2011). Twitter: Microphone for the masses? Media, Culture & Society, 33(5), 779. Myers, G. (2003). Discourse studies of scientific popularization: questioning the boundaries. Discourse Studies, 5(2), 265-279. Nature (2014). Why high-profile journals have more retractions [Online]. Nature News 17 September 2014 Available: http://www.nature.com/news/why-high-profile-journals-have-more-retractions- 1.15951 [Accessed 9 February 2015]. Nelson, B. (2009). Data sharing: Empty archives [Online]. Nature News 9 September 2009 Available: http://www.nature.com/news/2009/090909/full/461160a.html [Accessed 8 June 2011]. Nentwich, M. (2010). Web 2.0 and academia. In: Proceedings of the 9th annual ias-sts conference ‘Critical issues in science and technology studies’, 3-4 May 2010, Graz, Austria, 66-78. Neylon, C. & Wu, S. (2009a). Article-level metrics and the evolution of scientific impact. PLoS Biology, 7(11), e1000242. Neylon, C. & Wu, S. (2009b). Open science: tools, approaches, and implications. Pacific Symposium on Biocomputing, 14, 540-544. Nicholas, D., Clark, D., Rowlands, I. & Jamali, H. R. (2009). Online use and information seeking behaviour: Institutional and subject comparisons of UK researchers. Journal of Information Science, 35(6), 660- 676. Nicholas, D. & Rowlands, I. (2011). Social media use in the research workflow. Information Services and Use, 31(1), 61-83. Nicholas, D., Watkinson, A., Rowlands, I. & Jubb, M. (2011). Social media, academic research and the role of university libraries. Journal of Academic Librarianship, 37(5), 373-375. Nicholas, D., Williams, P., Rowlands, I. & Jamali, H. R. (2010). Researchers' e-journal use and information seeking behaviour. Journal of Information Science, 36(4), 494-516. Nicholson, S. W. & Bennett, T. B. (2011). Data sharing: academic libraries and the scholarly enterprise. Portal: Libraries and the Academy, 11, 505-516. Nielsen, M. (2011). An informal definition of OpenScience, [Online]. Available: http://www.openscience.org/blog/?p=454 [Accessed 29 December 2014]. Norris, M., Oppenheim, C. & Rowland, F. (2008). The citation advantage of open-access articles. Journal of the American Society for Information Science and Technology, 59(12), 1963-1972. Opdenakker, R. (2006). Advantages and disadvantages of four interview techniques in qualitative research. Forum: Qualitative Social Research, 7(4). Oppenheim, C. & Summers, M. A. (2008). Citation counts and the Research Assessment Exercise, part VI: Unit of assessment 67 (music). Information Research: An International Electronic Journal, 13(2). Ovadia, S. (2014). ResearchGate and Academia.edu: Academic social networks. Behavioral & Social Sciences Librarian, 33(3), 165-169. Park, J. H. & Shim, J. (2011). Exploring how library publishing services facilitate scholarly communication. Journal of Scholarly Publishing, 43(1), 76-89. Pearson, G. (2001). The participation of scientists in public understanding of science activities: The policy and practice of the UK Research Councils. Public Understanding of Science, 10(1), 121-137. PewResearch. (2014). Mobile technology fact sheet [Online]. PewResearch Internet Project. Available: http://www.pewinternet.org/fact-sheets/mobile-technology-fact-sheet/ [Accessed 17 June 2014]. Peytchev, A. (2009). Survey breakoff. Public Opinion Quarterly, 73(1), 74-97. 187

Pham, N. & Bogdan, M. (2010). Automatic data mining on internet by using PERL. Industrial Communication Systems, 65, 1-9. Pinfield, S. (2003). Open archives and UK institutions: an overview. D-Lib Magazine, 9(3). Pinfield, S. (2005). A mandate to self archive? The role of open access institutional repositories. Serials: The Journal for the Serials Community, 18(1), 30-34. Piwowar, H. A. (2010). Foundational Studies for Measuring the Impact, Prevalence, and Patterns of Publicly Sharing Biomedical Research Data. PhD Thesis, University of Pittsburgh. Piwowar, H. A. (2011). Who shares? Who doesn't? Factors associated with openly archiving raw research data. PLoS One, 6(7), e18657. Piwowar, H. A., Becich, M. J., Bilofsky, H. & Crowley, R. S. (2008). Towards a data sharing culture: Recommendations for leadership from academic health centers. PLoS Medicine, 5(9), 1315-1319. Piwowar, H. A. & Chapman, W. W. (2008). A review of journal policies for sharing research data. International Conference on Electronic Publishing. Piwowar, H. A. & Chapman, W. W. (2010). Public sharing of research datasets: A pilot study of associations. Journal of Informetrics, 4(2), 148-156. Piwowar, H. A., Day, R. S. & Fridsma, D. B. (2007). Sharing Detailed Research Data Is Associated with Increased Citation Rate. PLoS One, 2(3), e308. Piwowar, H. A., Vision, T. J. & Whitlock, M. C. (2011). Data archiving is a good investment. Nature, 473(7347), 285-285. Poliakoff, E. & Webb, T. L. (2007). What factors predict scientists' intentions to participate in public engagement of science activities? Science communication, 29(2), 242-263. Popescu, N. (2007). Scientific evaluation by peer review. The Amfiteatru Economic Journal, 9(21), 191-195. Powell, D. A., Jacob, C. J. & Chapman, B. J. (2011). Using blogs and new media in academic practice: Potential roles in research, teaching, learning, and extension. Innovative Higher Education, 37(4), 271-282. Priem, J., Costello, K. & Dzuba, T. (2011). First-year graduate students just wasting time? Prevalence and use of Twitter among scholars. Metrics 2011 Symposium on Informetric and Scientometric Research New Orleans, LA, USA. Priem, J. & Costello, K. L. (2010). How and why scholars cite on Twitter. Proceedings of the American Society for Information Science and Technology, 47(1), 1-4. Priem, J. & Hemminger, B. H. (2010). Scientometrics 2.0: New metrics of scholarly impact on the social Web. First Monday, 15(7). Procter, R., Williams, R. & Stewart, J. (2010a). If you build It, will they come? How researchers perceive and use web 2.0. Research Information Network. Available at www. rin. ac. uk/system/files/attachments/web_2. 0_screen. pdf [accessed 21 Dec 2011]. Procter, R., Williams, R., Stewart, J., Poschen, M., Snee, H., Voss, A. & Asgari-Targhi, M. (2010b). Adoption and use of Web 2.0 in scholarly communications. Philosophical Transactions of the Royal Society A, 368(1926). Rainie, L., Smith, A., Schlozman, K. L., Brady, H. & Verba, S. (2012). Social media and political engagement. Pew Internet & American Life Project. RCUK. (2013). RCUK policy on open access [Online]. Research Councils UK. Available: http://www.rcuk.ac.uk/research/openaccess/ [Accessed 21 November 2014]. Redfield, R. (2012). RRResearch blog [Online]. Available: http://rrresearch.fieldofscience.com/ [Accessed 6 Feb 2012]. Regis, A. K. (2012). Early career victorianists and social media: Impact, audience and online identities. Journal of Victorian Culture, 1-8. 188

Riazat, N. (2011). A doctoral networking forum: The origins of #phdchat. [Online]. Available: http://phdchat- nasimariazat.blogspot.ca/2011/08/nasima-riazat-doctoral-networking-forum.html?spref=tw [Accessed 16 July 2012]. Rieger, O. Y. (2010). Framing digital humanities: The role of new media in humanities scholarship. First Monday, 15(10). Riesch, H. & Mendel, J. (2014). Science blogging: Networks, boundaries and limitations. Science as Culture, 23(1), 51-72. RIN. (2008). To share or not to share: Publication and quality assurance of research data outputs [Online]. Research Information Network. Available: http://www.rin.ac.uk/our-work/data-management-and- curation/share-or-not-share-research-data-outputs [Accessed 22 Dec 2011]. RIN. (2009). Patterns of information use and exchange: case studies of researchers in the life sciences [Online]. Research Information Network. Available: http://www.rin.ac.uk/our-work/using-and- accessing-information-resources/patterns-information-use-and-exchange-case-studie [Accessed 5 May 2015]. RIN. (2010). Open to all: Case studies of openness in research [Online]. Research Information Network. Available: http://www.rin.ac.uk/our-work/data-management-and-curation/open-science-case- studies [Accessed 25 June 2011]. RIN. (2011a). Data centres: their use, value and impact [Online]. Research Information Network. Available: http://www.rin.ac.uk/our-work/data-management-and-curation/benefits-research-data-centres [Accessed 21 Dec 2011]. RIN. (2011b). Reinventing research? Information practices in the humanities [Online]. Research Information Network. Available: http://www.rin.ac.uk/our-work/using-and-accessing-information- resources/information-use-case-studies-humanities [Accessed 5 May 2015]. RIN. (2014). Nature communications: Citation analysis [Online]. Nature. Available: http://www.nature.com/press_releases/ncomms-report2014.pdf [Accessed 19 August 2014]. Roblye, M. D., McDaniel, M., Webb, M., Herman, J. & Witty, J. V. (2010). Findings on Facebook in higher education: A comparison of college faculty and student uses and perceptions of social networking sites. The Internet and Higher Education, 13(3), 134-140. Rodriguez, H., Snyder, M., Uhlén, M., Andrews, P., Beavis, R., Borchers, C., Chalkley, R. J., Cho, S. Y., Cottingham, K. & Dunn, M. (2009). Recommendations from the 2008 international summit on proteomics data release and sharing policy: the amsterdam principles. Journal of Proteome Research, 8(7), 3689-3692. Roßmann, J., Blumenstiel, J. E. & Steinbrecher, M. (2014). Why do respondents break off web surveys and does it matter? Results from four follow-up surveys. International Journal of Public Opinion Research, DOI: 10.1093/ijpor/edu025. Rosen, R. J. (2012). The Case (Study) of Arsenic Life: How the Internet Can Make Science Better [Online]. The Atlantic. Available: http://www.theatlantic.com/technology/archive/2012/07/the-case-study-of- arsenic-life-how-the-internet-can-make-science-better/259581/ [Accessed 24 April 2015]. Rowlands, I. & Nicholas, D. (2006). The changing scholarly communication landscape: an international survey of senior researchers. Learned Publishing, 19(1), 31-55. Rowlands, I., Nicholas, D. & Huntington, P. (2004). Scholarly communication in the digital environment: What do authors want? Learned Publishing, 17(4), 261-273. RussellGroup. (2012). The social impact of research conducted in Russell Group universities, Russell Group Papers issue 3 [Online]. Available: http://russellgroup.org/SocialImpactOfResearch.pdf [Accessed 19 August 2014]. Sale, J. E., Lohfeld, L. H. & Brazil, K. (2002). Revisiting the quantitative-qualitative debate: Implications for mixed-methods research. Quality and Quantity, 36(1), 43-53. 189

Sanders, D. & Brynin, M. (1998). Ordinary least squares and logistic regression analysis. In: Scarbrough, E. & Tanenbaum, E. (eds.) Research Strategies in the Social Sciences. Oxford: Oxford University Press. Schaefer, D. R. & Don, A. D. (1998). Development of a standard e-mail methodology: Results of an experiment. The Public Opinion Quarterly, 62(3), 378-397. Schafer, J. B., Konstan, J. & Riedl, J. (1999). Recommender systems in e-commerce. In: Proceedings of the 1st ACM conference on Electronic commerce, 158-166. Schauder, D. (1993). Electronic publishing of professional articles: Attitudes of academics and implications for the scholarly communication industry. Proceedings of the IATUL Conferences. Scheliga, K. & Friesike, S. (2014). Putting open science into practice: A social dilemma? First Monday, 19(9). Schwarz, G. J. & Kennicutt, R. C. J. (2004). Demographic and citation trends in astrophysical journal papers and preprints. Arxiv preprint astro-ph/0411275. Seale, C. (1999). The Quality of Qualitative Research. London: SAGE. Seife, C. (2015). Science’s big scandal [Online]. Slate. Available: http://www.slate.com/articles/health_and_science/science/2015/04/fake_peer_review_scientific _journals_publish_fraudulent_plagiarized_or_nonsense.html [Accessed 28 April 2015]. Selltiz, C., Jahoda, M., Deutsch, M. & Cook, S. W. (1959). Research Methods in Social Relations. New York: Holt, Rinehart and Winston. Selwyn, N. (2009). Faceworking: exploring students' education-related use of Facebook. Learning, Media and Technology, 34(2), 157-174. Shanahan, M. (2011). Science blogs as boundary layers: Creating and understanding new writer and reader interactions through science blogging. Journalism, 12(7), 903-919. Shema, H., Bar-Ilan, J. & Thelwall, M. (2012). Research blogs and the discussion of scholarly information. PLoS One, 7(5), e35869. Shema, H., Bar-Ilan, J. & Thelwall, M. (2014). Do blog citations correlate with a higher number of future citations? Research blogs as a potential source for alternative metrics. Journal of the Association for Information Science and Technology, 65(5), 1018-1027. Shih, C. (2010). The Facebook era: Tapping online social networks to market, sell, and innovate. Pearson Education. Sismondo, S. (2004). An Introduction to Science and Technology Studies. Oxford: Blackwell Publishing. Smith, A. (2013). Smartphone ownership 2013 [Online]. PewResearch Internet Project. Available: http://www.pewinternet.org/2013/06/05/smartphone-ownership-2013/ [Accessed 17 June 2014]. Smith, A. & Rainie, L. (2010). Overview: The people who use Twitter. Retrieved March, 23, 2011. Solomon, D. J. & Björk, B. C. (2012). Publication fees in open access publishing: Sources of funding and factors influencing choice of journal. Journal of the American Society for Information Science and Technology, 63(1), 98-107. Sonnert, G. & Holton, G. (1996). Career patterns of women and men in the sciences. American Scientist, 63- 71. Spier, R. (2002). The history of the peer-review process. Trends in Biotechnology, 20(8), 357-358. Statista. (2014). Tablet user penetration rate in the United Kingdom from 2012 to 2018 [Online]. Statista: The Statistics Portal. Available: http://www.statista.com/statistics/249124/tablet-penetration-in- the-united-kingdom/ [Accessed 17 June 2014]. Steen, R. G. (2011). Retractions in the scientific literature: do authors deliberately commit research fraud? Journal of Medical Ethics, 37(2), 113-117. 190

Stewart, J., Procter, R., Williams, R. and Poschen, M. (2013). The role of academic publishers in shaping the development of Web 2.0 services for scholarly communication. New Media & Society, 15(3), 413- 432. Stone, S. (1982). Humanities scholars: Information needs and uses. Journal of Documentation, 38(4), 292- 313. Stonehill-Publicity, M. (2010). The top 10 most popular social networks [Online]. Available: http://prmotion212.wordpress.com/2010/10/15/the-top-10-most-popular-social-networks-2010/ [Accessed 16 May 2014]. Styles, K. (2013). 7 in 10 people in the UK now own a smartphone [Online]. MobileMarketing. Available: http://mobilemarketingmagazine.com/7-10-people-uk-now-own-smartphone/ [Accessed 17 June 2014]. Swan, A. (2006). The culture of open access: Researchers’ views and responses. In: Jacobs, N. (ed.) Open Access: Key Strategic, Technical and Economic Aspects. Chandos. Swan, A. & Brown, S. (2004). Authors and open access publishing. Learned Publishing, 17(3), 219-224. Swan, A. & Brown, S. (2005). Open access self-archiving: An author study [Online]. Key Perspectives. Available: http://cogprints.org/4385/ [Accessed 11 June 2014]. Swan, A., Needham, P., Probets, S., Muir, A., Oppenheim, C., O'Brien, A., Hardy, R., Rowland, F. & Brown, S. (2005). Developing a model for e-prints and open access journal content in UK further and higher education. Learned Publishing, 18(1), 25-40. Tacke, O. (2011). Open Science 2.0: How Research and Education Can Benefit from Open Innovation and Web 2.0. On Collective Intelligence, 76, 37-48. Tashakkori, A. & Teddlie, C. (2003). Handbook of Mixed Methods in Social & Behavioral Research. Thousand Oaks: Sage. Tatum, C. & Jankowski, N. W. (2010). Openness in scholarly communication: Conceptual framework and challenges to innovation [Online]. Available: http://digital-scholarship.ehumanities.nl/wp- content/uploads/Tatum-and-Jankowski-openness-in-scholarly-comm-v202.pdf [Accessed 21 December 2011]. Terras, M. (2012). The verdict: is blogging or tweeting about research papers worth it? [Online]. The Impact Blog, LSE. Available: http://blogs.lse.ac.uk/impactofsocialsciences/2012/04/19/blog-tweeting- papers-worth-it/ [Accessed 13 June 2014]. Thanuskodi, S. (2011). Web 2.0 awareness among library and information science professionals of the engineering colleges in Chennai city: A survey. J Communication, 1(2), 69-75. The Daily Texan. (2013). Despite flaws, open access is worth the price [Online]. Available: http://www.dailytexanonline.com/opinion/2013/10/11/despite-flaws-open-access-is-worth-the- price [Accessed 18 May 2015]. The Economist. (2013). Trouble at the lab [Online]. The Economist. Available: http://www.economist.com/news/briefing/21588057-scientists-think-science-self-correcting- alarming-degree-it-not-trouble [Accessed 24 April 2015]. Thelwall, M. & Kousha, K. (2014a). Academia.edu: Social network or academic network? Journal of the Association for Information Science and Technology, 65(4), 721-731. Thelwall, M. & Kousha, K. (2014b). ResearchGate: Disseminating, communicating, and measuring scholarship? Journal of the Association for Information Science and Technology. DOI: 10.1002/asi.23236 Thompson, J. W. (2002). The death of the scholarly monograph in the Humanities? Citation patterns in literary scholarship. Libri, 52(3), 121-136. Tucker, J. (2009). Motivating Subjects: Data Sharing in Cancer Research. PhD Thesis, Virginia Polytechnic Institute and State University. 191

Veletsianos, G. & Kimmons, R. (2012). Networked participatory scholarship: Emergent techno-cultural pressures toward open and digital scholarship in online networks. Computers & Education, 58(2), 766-774. Virdi, J. (2010). The growth of HPS blogging [Online]. Newsletter of the History of Science Society. Available: http://hssonline.org/publications/Newsletter2010/October-blogging.html [Accessed 8 February 2012]. Watts, J. (2007). Disseminating research results in the 21st century. Suppression and Detection Research and Applications--A Technical Working Conference (SUPDET 2007). Weber, N. M., Piwowar, H. A. & Vision, T. J. (2010). Evaluating data citation and sharing policies in the environmental sciences. Proceedings of the American Society for Information Science and Technology, 47(1), 1-2. Weisberg, H. F. & Bowen, B. D. (1977). An Introduction to Survey Research and Data Analysis. San Francisco: WH. Freeman and Company. WellcomeTrust. (2014). Charity alliance launches fund to make charitably funded research open access [Online]. Available: http://www.wellcome.ac.uk/News/Media-office/Press- releases/2014/WTP057254.htm [Accessed 10 Oct 2014]. Weller, K., Dröge, E. & Puschmann, C. (2011). Citation analysis in Twitter. Approaches for defining and measuring information flows within Tweets during scientific conferences. Proceedings of Making Sense of Microposts. Whitley, R. (1972). Black boxism and the sociology of science: A discussion of the major developments in the field. The Sociological Review Monograph, 18, 61-92. Wilbanks, J. (2006). Another reason for opening access to research. BMJ: British Medical Journal, 333(7582), 1306-1308. Wilkinson, D. & Thelwall, M. (2011). Researching personal information on the public web methods and ethics. Social Science Computer Review, 29(4), 387-401. Williams, P., Stevenson, I., Nicholas, D., Watkinson, A. & Rowlands, I. (2009). The role and future of the monograph in arts and humanities research. Aslib Proceedings, 61(1), 67-82. Willinsky, J. (2010). The stratified economics of open access. OJS på dansk, 1(1). Willis, G. B. (2005). Cognitive Interviewing: A Tool for Improving Questionnaire Design. Thousand Oaks, CA: Sage. Wing, K. (2014). Scientists across disciplines must get to a better agreement on social media metrics [Online]. The Impact Blog, LSE. Available: http://blogs.lse.ac.uk/impactofsocialsciences/2014/09/29/what- are-scientists-doing-on-twitter/ [Accessed 10 Oct 2014]. Wolfendale Committee (1995). Final Report. London: Office of Science and Technology. Wouters, P. F. (1999). The Citation Culture. PhD thesis, The University of Amsterdam. Wright, K. B. (2005). Researching internet-based populations: Advantages and disadvantages of online survey research, online questionnaire authoring software packages, and web survey services. Journal of Computer-Mediated Communication, 10(3). Xia, J. F. (2007). Assessment of self-archiving in institutional repositories: Across disciplines. Journal of Academic Librarianship, 33(6), 647-654. Xia, J. F., Myers, R. L. & Wilhoite, S. K. (2011). Multiple open access availability and citation impact. Journal of Information Science, 37(1), 19-28. Xia, J. F. & Sun, L. (2007). Assessment of self-archiving in institutional repositories: Depositorship and full- text availability. Serials Review, 33(1), 14-21. Xie, Y. & Shauman, K. A. (1998). Sex differences in research productivity: New evidence about an old puzzle. American Sociological Review, 847-870. 192

You, J. (2014). The top 50 science stars of Twitter [Online]. Science. Available: http://news.sciencemag.org/scientific-community/2014/09/top-50-science-stars-twitter#full-list [Accessed 10 Oct 2014]. Zeller, R. & Carmines, E. (1978). Statistical analysis of social data. Chicago: Rand McNally College Publishing Company. Zika, E., Kaye, J., Brand, A. & Ibarreta, D. (2008). Sample, data use and protection in biobanking in Europe: Legal issues. Pharmacogenomics, 9(6), 773-781. Ziman, J. M. (1987). An Introduction to Science Studies: The Philosophical and Social Aspects of Science and Technology. Cambridge: Cambridge University Press. Zimmerman, A. S. (2008). New Knowledge from Old Data: The Role of Standards in the Sharing and Reuse of Ecological Data. Science, Technology & Human Values, 33(5), 631-652.

193

Appendix 1 Questions for interviewees who were users of social media for research

Questions about using social media for research: How long have you been a researcher? Which subject area and which institution are you based? Which social media site do you use the most generally? (e.g. Facebook, blog, Twitter, Youtube) Which social media site do you use for research purpose? (e.g. Twittter, blogs, Facebook, Academia.edu, MethodSpace, etc.) And Why? Do you use Facebook for research? For example, search events or research groups? Do you write or read other research blogs? If yes, on which website? Why did you start to use blogs and Twitter? How long have you used blogs and Twitter for your research? How often do you use blogs and Twitter in the past 1 month? What you do hope to get out of your blogging and tweeting? Who do you think are the audience for your blogs and Twitter? Have you achieved what you hoped for since starting using blogs and twitter? How do you use blogs and Twitter? What did you find useful? What strategy did you use to promote your blogs and Twitter profile? Do you use blog and twitter together? For example, do you use Twitter to promote your blog profile? Or do you use two or more social media tools to cooperate with each other? On blogs/Twitter, do you post links to papers that you find interesting? Or papers you have published or abstract that have been accepted by conferences? Do you announce when you start a research project on blogs/Twitter or other social media site? Do you announce conferences or workshops you are going to attend on Twitter or other social media site? Do you use twitter during a conference? If yes, could you tell me about your experience? Do you blog or tweet about research findings before they are published in a journal or conference paper? If yes, why? If no, why not? What would you feel if someone you collaborate with announces the research findings on social media (e.g. blog or tweet) before final publication? What is your opinion of the ownership and appropriate use of your online content by others? Would you be worried that your online content be misused or misquoted by others? 194

Do you use the Common licences? Do you have any concern of privacy issues using social media? Do you have separate professional and personal identities on different social media sites? If yes, why? If no, why not? Have you heard of researchblogging.org or any other aggregator of research blogs? It is an aggregator of science blogs including social science. Bloggers who discuss peer-reviewed research can register with the aggregator and they mark relevant posts in their blog, these posts appear on the aggregator’s site. To your knowledge, do the professors in your department use social media for research? For example, have you seen their profiles on Academia.edu? To your knowledge, do PhD students and early career researchers in your department use social media for research?

Questions about open access and publication: How do you normally search for scholarly paper in your field? (e.g. Google, Google scholar, Twitter, blog, University library sites, Web of Science, etc.) Do you use any subject-specific forums and discussion boards? Who do you think should be the audience of your research output? To whom, you want your research outcome to be seen? (e.g. policy maker, general public, service users, etc.) Have you heard about open access journal? Have you ever published in Open access journal (such as PLoSOne or other author-pay model open access journal)? If so, can you tell a bit about your experience? If not, will you be interested in publishing in open access journal when you may need to pay author fee for OA journal? Have you heard about open access repositories (such as ArXiv and Manchester eScholar)? Have you deposited your paper in open access repositories? If not, will you deposit papers (published or working papers) in open access repositories in the future? Have you ever used data from open access data centre? (such as the national funding body’s site) Will you consider share you data in an open access site? And why? What’s your general opinion about openness to scientific knowledge /research outputs? Should science be open to all? Have you heard about anyone stealing ideas for publication? 195

Appendix 2 Questions for interviewees who were non-users of social media for research

Are you active on social media site like Facebook for personal use? How long have you been a researcher? Which subject area and which institution are you based? Why do you choose not to use blogs/twitter for your research? What do you think of the use of social media? Do you know any potential benefit of using social media for your research? What do you think of the claimed benefits? Please comment on them. (e.g., find information, get new ideas, get inspired, find peers for co-operation, disseminate scientific knowledge to a wider audience, use it as notice board on conference? ) Does your university or Department promote using social media for research purpose? What will convince you to use social media for your research? Do you think some changes or development will persuade you to use blogs and twitter it in the future? Questions about open access and publication Have you heard about open access journal? Have you ever published in Open access journal? Or will you be interested in publishing in open access journal? Have you heard about open access repositories? Have you deposited your paper in open access repositories? Or will you deposit papers (published or working paper) in open access repositories? Have you ever used data from open access data centre? (Such as the national Funding body’s site) Will you consider share you data in an open access site? And why? What’s your general opinion about open access to research outputs?

196

Appendix 3 Survey invitation email

From: [email protected] [mailto:[email protected]] Sent: 01 July 2013 12:03 To: XXX Subject: How academics communicate their research.

Dear colleague,

As the central part of my PhD I'm conducting a survey on academics’ attitudes towards and experiences of publishing and communicating their research.

You have been selected for this study as being a UK based academic researcher in one of the sampled universities.

I realise that taking time out from your work to complete this survey is not a top priority. However, I would be extremely grateful if you would. It should only take you around 10 minutes and your input is very important.

You will also be entered into the prize draw for £50 of book vouchers.

All responses will be treated confidentially. This study was approved by the Ethics panel, School of Social Sciences, University of Manchester.

The survey can be accessed here: https://www.surveymonkey.com/s/PK2CRFT

Many thanks in advance for your time and valuable contribution to my PhD project.

If you have any enquires or would like to be removed from the sample list, please write to me directly

[email protected]

More information about my PhD project can be found at my university research profile: http://www.socialsciences.manchester.ac.uk/disciplines/sociology/postgraduate/current/yzhu/

Yimei Zhu PhD Student Sociology & CCSR Arthur Lewis Building University of Manchester M13 9PL Tel: 0161 306 6948 Twitter: @yimeizhu

197

Appendix 4 Survey questions Research Communication Survey

Welcome

If you are a UK based academic from one of the sampled universities (see Q1), you are invited to participate in my survey. This survey intends to find out academics’ attitudes and experiences with publishing and communicating their research.

Thank you very much in advance for your time and valuable contribution to my PhD project. It should only take you 10 minutes and all responses will be treated confidentially. You will also be entered into the prize draw for £50 of book vouchers.

To begin, I have some questions about your basic information.

1. Are you primarily based in one of the sampled universities? (Please tick one)

o Cardiff University

o Durham University

o Imperial College London

o King's College London

o Newcastle University

o Queen's University Belfast

o University of Edinburgh

o University of Glasgow

o University of Leeds

o University of Manchester

o University of Warwick

o University of York

o Other (please specify)

2. What is your PRIMARY research discipline (according to 2014 REF categories)?

A 1 Clinical Medicine 2 Public Health, Health Services and Primary Care 3 Allied Health Professions, Dentistry, Nursing and Pharmacy 4 Psychology, Psychiatry and Neuroscience 5 Biological Sciences 6 Agriculture, Veterinary and Food Science B 7 Earth Systems and Environmental Sciences 198

8 Chemistry 9 Physics 10 Mathematical Sciences 11 Computer Science and Informatics 12 Aeronautical, Mechanical, Chemical and Manufacturing Engineering 13 Electrical and Electronic Engineering, Metallurgy and Materials 14 Civil and Construction Engineering 15 General Engineering C 16 Architecture, Built Environment and Planning 17 Geography, Environmental Studies and Archaeology 18 Economics and Econometrics 19 Business and Management Studies 20 Law 21 Politics and International Studies 22 Social Work and Social Policy 23 Sociology 24 Anthropology and Development Studies 25 Education 26 Sport and Exercise Sciences, Leisure and Tourism D 27 Area Studies 28 Modern Languages and Linguistics 29 English Language and Literature 30 History 31 Classics 32 Philosophy 33 Theology and Religious Studies 34 Art and Design: History, Practice and Theory 35 Music, Drama, Dance and Performing Arts 36 Communication, Cultural and Media Studies, Library and Information Management

3. Do you teach (lectures and/or tutorials)?

o Yes

o No

Copy of RIN WEB2.0 Draft v5 4. How old are you? o Under 25 o 25-34 o 35-44 o 45-54 o 55-64 o 65 and over

199

5. How many years have you worked as an academic researcher (including the period doing a research degree, eg,. PhD)?

6. What is your grade, or approximate equivalent?

o Professor o Reader o Senior Lecturer o Senior Researcher o Lecturer o Research Fellow/post-doc o PhD candidate o MPhil/MSc/MA student o Other (please specify)

7. What is your gender? o Female o Male o Other

8. Do you use any of the following devices?

Yes No

Personal laptop Smart phone IPad or other Tablet

Your Research and Publication

In this section I wish to know about your attitudes and experiences in searching and disseminating information for research purposes.

9. How important are the following INFORMATION RESOURCES for your research?

Very important/Fairly important/Not very important/Not at all important/N/A Peer Reviewed journals

Conference papers/presentations

Academic Books

Institutional web pages

Research blogs

Twitter 200

Facebook

Research social networks (e.g. ResearchGate & Academia.edu)

Online forums & groups

Podcasts

Youtube videos

Email lists

Films, audio, artwork or other non-textual sources

Other (please specify other means and how important they are)

10. Who should know about your research findings? (Tick all that apply) o Other researchers o Policy makers o General public o Service users o Other (please specify)

11. How important are the following DISSEMINATION means for communicating your research?

Very important/Fairly important/Not very important/Not at all important/N/A Peer reviewed journal

Conference paper/presentation

Academic Book

Institutional web page

Personal web page

Research blog

Twitter

Facebook

Research social network (e.g. ResearchGate & Academia.edu)

Online forums & group

Podcast 201

Youtube video

Email list

Films, audio, artwork or other non-textual output

Other (please specify other means and how important they are)

12. Regarding your most RECENT peer-reviewed publication, did you do anything to promote it? (Tick all that apply) o No. o Yes, I told colleagues in person. o Yes, I emailed the paper to colleagues/ added the web link in my email signature. o Yes, I used Twitter to post the web link to the online article. o Yes, I posted the article title and link in my research blog. o Yes, I posted the article title and link on Facebook or other social networks. o I haven't published anything yet. o Other (please specify)

13. Have you published an article in a journal that is open-access(OA)?(OA journal means the articles in this journal are freely accessible online to the public by the publishers, which often requires author-pay, in contrast to subscription based journal that requires reader- pay. Authors can also choose to pay a fee to make their articles open-access in partially OA journals. ) o Yes o No, but I plan to in the future. o No, I have no plan to publish in OA journal. o No, not sure about this. o Not applicable (I haven't or don't publish research articles) o Other or/and comments

14. If yes, did you or your institution pay for the author fee in your last OA publication? o Yes, my institution/research fund paid. o Yes, I paid out of my own pocket. o No, our collaborators paid. 202

o No, the publishers waved the fee. o No, no fee was required.

15. In general, do you prefer to publish research articles in open-access journals rather than subscription-based journals if they have similar reputation or ranking of citation impact? o Yes, I prefer OA journals even if I personally have to pay author fee. o Yes, I prefer OA journals only if I personally don’t have to pay author fee. o No, I prefer conventional subscription-based journals. o I don’t have a preference, it all depends on which journals have higher reputation in my field. o Don’t know enough information about this matter. o Not applicable (I don't produce research articles) o Other or/and comments

16. Are you aware of open-access repositories for depositing research articles?

o Yes

o No

17. Have you deposited your research articles (published or working paper) in an open-access online repository (e.g. institutional repository, disciplinary repository & your own website) o Yes o No, but I plan to in the future. o No, I have no plan to do so. o No, not sure about this. o Not applicable (I haven't or don't produce research articles) o Other or/and comments

18. If yes, did you do anything to promote the most RECENT deposited article? (Tick all that apply) o No. o Yes, I told colleagues in person. 203

o Yes, I emailed the web link to colleagues/ added the web link in my email signature.

o Yes, I used Twitter to post the web link to the deposited article.

o Yes, I posted the article title and link in my research blog.

o Yes, I posted the article title and link on Facebook or other social networks.

o Other or/and comments

19. How important do you think it is, in general, to make research articles freely accessible online to everyone? o Very important o Fairly important o Not very important o Not at all important

Any reason of this (please specify)

20. In your research work, have you used secondary research data from an online repository that were collected by other researchers?

o Yes

o No

21. Have you deposited your OWN primary research data in an online repository that can be reused by other researchers?

o Yes

o No

If yes, any reason (eg., journal requirement)

22. If yes, did you do anything to promote the most RECENT primary research data that you deposited online for reuse? (Tick all that apply)

o No.

o Yes, I told colleagues in person.

o Yes, I emailed the information about data to colleagues.

o Yes, I used Twitter to post the web link to my data. 204

o Yes, I posted the data information and link in my research blog.

o Yes, I posted the data information and link on Facebook.

o Other or/and comments

23. If no, in the future, will you share you primary research data in an online repository that can be reused by others? (Tick all that apply)

o Yes, I plan to.

o I don't produce primary data

o Not sure

o No, I want to secure publication

o No, my primary data is not reusable

o No, because of ethical issue

o Other or/and comments

24. How important do you think it is, in general, to make research data available online for reuse? o Very important o Fairly important o Not very important o Not at all important

Any reason of this (please specify)

25. Are you aware of Research Council UK (RCUK) Policy on Open Access to the outputs of RCUK-funded research which came into effect on 1 April 2013?

o Yes

o No

o Heard of, but not sure about the detail

o Other (please specify)

26. How often do you use the following online tools to SEARCH for scholarly articles? 205

Never/Sometimes/Often/Always

Google Search

Google Scholar

Web of Science

Scopus

Mendeley

Twitter

Research blogs

Library database

NCBI

PubMed

Other (please specify other tools and how often you use those)

Using Social Media for research

In this section I wish to find out whether or/and how you use social media. Social media refers to a group of online applications that allow the creation and exchange of User Generated Content. Social media users can create, share, and exchange information and ideas in virtual communities. Examples of social media tools include social networking sites, blogs, microblogs, wikis, photo and video sharing sites and many more.

27. How often do you use the following online services in your RESEARCH Work?

Never/Sometimes/Often/Always

Twitter

Facebook

Google Drive/Doc

Research Blogs

Youtube or other video sharing services

Podcast

Webinars

Linked.In

Academia.edu 206

Researchgate.net

Mendeley or other reference sharing sites

Delicious or other bookmarking sites

Paper.li

Scoop.it

Others (please specify other online services and how often you use those)

28. How often do you do any of the following in your RESEARCH work?

Never/Sometimes/Often/Always

Read research blogs

Comment on other research blogs

Post updates of ongoing research on a research blog

Gather research information on Twitter

Post updates of ongoing research on Twitter

Gather research information on social networking sites (e.g. Facebook & ResearchGate)

Post updates of ongoing research on social networking sites

Watch videos on Youtube

Post videos on Youtube

Read slides on SlideShare or similar

Post slides on SlideShare or similar

Ask or read research issues in online forums

Comment on research issues in online forums

Read a public wiki (e.g. wikipedia)

Contribute to a public wiki

Other or/and comments

29. If you write a research blog, what kind of information do you blog about and why?

207

30. Do you post your real name on your social media sites?

Yes /No, I prefer anonymity/I don’t use it in my research work

Blog

Twitter

Other or/and comments

31. Why did you start to use social media tools in your research work? (if applicable)

32. Have you attended any social media training courses organised by your institution? o Yes o No, but I plan to in the future. o No, I'm not interested. o No, I'm not aware of any training available. o Other (please specify)

33. Are you aware of any encouragement or promotion by your institution in terms of using social media to communicate & help your research?

o Yes

o No

34. Have you received any recommendation from colleagues of using social media to communicate & help your research?

o Yes

o No 208

35. Outside of work, how often do you do any of the following for LEISURE purposes?

Never/Sometimes/Often/Always

Post and/or send message on Facebook

Post and/or send message on Twitter

Read other blogs

Comment on other blogs

Write a blog

Watch videos on YouTube

Post videos on YouTube

Share slides/photos on content sharing sites

Use other social networking sites

Others (please specify other social media tools you use for leisure)

Research Impact and Relevant Attitudes

Now, the last section asks about your research impact and your attitudes towards open science and using social media to communicate your research.

36. Will you be assessed in the 2014 REF? (Research Excellence Framework is the new system for assessing the quality of research in UK higher education institutions)

o Yes

o No

o Don’t Know

37. Do you have a permanent academic job?

o Yes

o No

o Prefer not to say

o My primary job is outside Academia 209

38. Have you created a profile in Google Scholar for yourself?

o Yes

o No

39. How many research outputs have you published in the last 5 years? (Please give an estimate)

40. If you have publications, how many total citation counts have your publications received? (Please give an estimate number. You may search yourself on Google Scholar and create a profile which will give you the total citation)

41. What's your H-index if known?

42. Finally, I would like to know to what extent you agree or disagree with the following statements?

Strongly agree/Agree/Neither disagree nor agree/Disagree/Strongly disagree

Articles that are made open-access will receive more citations.

Blogging or tweeting about my publication will increase citations.

Communicating research on social media may leak results to competitors.

Research published on social media cannot be trusted as not being peer-reviewed.

Communicating research on social media benefits the public.

Communicating research on social media accelerates scientific discovery.

Communicating research on social media risks my good ideas being stolen.

Communicating research on social media may result in plagiarism.

Using social media helps me find collaboration opportunities.

Using social media increases my chances of getting funding.

Using social media promotes my professional profile.

Using social media benefits my career.

210

43. Any additional comments about open access scholarly communication and using social media to communicate your research?

44. If you would like to be entered into the prize draw for £50 of book vouchers, please add you name and email address.

Name

Thank you so much for completing this survey. Your input contributes to my PhD project as well as the understanding of science communication and publishing! All responses will be treated confidentially! If you are happy with your participation, please click 'Done'!