Iterative Analysis of Interviews About in ,

© 2021 SAGE Publications, Ltd. All Rights Reserved. This PDF has been generated from SAGE Research Methods Datasets. SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1 Iterative Analysis of Interviews About , Denmark

Student Guide

Introduction This dataset demonstrates how to conduct an iterative analysis on interview data. In iterative analysis, the researcher alternates between emergent meanings found in data and existing concepts, theories, and ideas from the academic literature (Tracy, 2020). Iterative analysis is useful for developing new insight into a topic that builds on an existing body of research. These data are drawn from field research conducted by Elizabeth Wilhoit Larson. Following research on bike commuters in the American Midwest, where biking to work is unusual, Larson was interested in studying a place where biking is normal. This led her to Copenhagen, Denmark, a city with world-class where about half of its 1 million residents use bicycles every day (Denmark, 2016). The goal of this study was to understand how communication plays a role in making cycling in Copenhagen an orderly and organized experience. Semi-structured interviews with residents of Copenhagen and ethnographic observations were used to collect data.

Semi-Structured Interviews In qualitative research, interviews are seen as a way to co-construct knowledge through the interaction between a researcher and participant. The researcher is not seen as extracting knowledge from the participant, but knowledge and

Page 2 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1 data are created through conversation. Sidney Webb and Beatrice Webb (1932) have defined a research interview as “a conversation with a purpose.” In a semi- structured interview, the researcher has something in mind that they would like to learn from a research participant. To this end, the researcher comes prepared with some questions they plan to ask. However, the interview is only semi-structured because the interviewer might ask additional questions, change the order that questions are asked in, or not ask certain questions based on the progression of the interview. The researcher seeks to understand the world from the participant’s point of view and uses the interview questions to help the participant articulate their perspective. With semi-structured interviews, the researcher makes in-the- moment decisions about how to advance the interview and which questions to ask in order to best understand the participant’s experience. Semi-structured interviews also allow the participant to guide the interview more, potentially resulting in a more complex understanding of participant’s experiences (Tracy, 2020).

Data Exemplar: Semi-Structured Interviews on Cycling With Residents of Copenhagen, Denmark Data collection took place in Copenhagen, Denmark, one of the best cities in the world for cycling (Københavns Kommune, 2017). I had previously studied people who ride bikes to work in the American Midwest—where riding bikes for transportation is unusual (Wilhoit & Kisselburgh, 2015), and wanted to understand what makes a city like Copenhagen so good for biking, particularly in terms of how rules and norms for cycling are communicated. I collected data through participant observation. I spent January–July 2013 in Copenhagen and collected extensive ethnographic data during that time, both biking around the city myself as an active participant and observing others as they navigated the cycling infrastructure and culture of the city. I wrote field notes after most bike rides I took (Emerson et al., 2011). Additionally, I conducted semi-structured interviews (n = 23) with residents

Page 3 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1 of Copenhagen, so that I could better understand the experience of cycling. I also conducted one interview with an official from the cycling division of the city government. In the interviews, I asked questions about their cycling routines, challenges they encountered, how they learned the rules and norms of cycling in Copenhagen, and their reflections on the culture of cycling in Denmark. In this dataset, I focus on the interview data.

Iterative Analysis Iterative analysis is a framework for analyzing qualitative data in a way that alternates between meanings emerging from the data (emic readings) and use of existing literature, theories, and explanations (etic readings). Iterative analysis has been synthesized by Tracy (2020), drawing on the work of other scholars like Dennis Gioia and colleagues (see Gioia et al., 2013) and Srivastava and Hopwood (2009). Tracy was motivated to articulate this form of analysis because she noted that there is a difference between what many researchers say they are doing for their analysis and what they actually do. Tracy (2020) writes, “I would estimate at least 80% of qualitative articles say, ‘I used a version of grounded theory and the constant comparative method for analyzing my qualitative data.’” However, she continues to state that most scholars who claim to use grounded theory to analyze their data are not actually using this method. Grounded theory (Glaser & Strauss, 1967) is an inductive approach to data analysis in which one’s findings emerge from the data rather than questions the researcher had going into the project. Tracy (2020) has explained that most researchers who claim to be doing grounded theory are actually drawing on existing frameworks, theories, and questions in their data analysis.

Based on this observation, Tracy (2020) synthesized various approaches to data analysis to develop iterative analysis, which she suggests more closely matches what most qualitative researchers actually do when they analyze data. Iterative

Page 4 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1 analysis is a reflexive process. Through repeated engagement with the data, researchers gradually gain a more focused understanding of the data (Srivastava & Hopwood, 2009). Iterative data analysis begins with coding. Coding involves assigning a word or phrase to a portion of data to capture the essence of that segment of data. Primary-cycle coding involves examining the data and assigning emergent codes that represent the data (Tracy, 2020). As a researcher gets further into the data analysis process, they will create a codebook that summarizes and defines each code. Once primary-cycle coding has been completed, the researcher moves to secondary-cycle coding (Tracy, 2020). In this cycle of coding, the researcher moves beyond purely descriptive codes to categorize, combine, and synthesize the first-level codes. It is in the second- level coding that the researcher will often turn to disciplinary concepts as codes, specifically looking for concepts from relevant theories in the data. Through activities like memoing, finding additional data that might counter one’s emerging findings, and proposing possible interpretations of the data, the researcher creates an in-depth understanding of the data and relationships between codes in this stage. At this point, the researcher can return to initial research questions and begin to formulate answers.

Stage 1: Organizing and Preparing Data The first step of data analysis is simply to gather and organize one’s data (Tracy, 2020). This might involve transcribing interviews, organizing field notes, or deciding which documents should be part of the analysis. For this project, organizing the data involved transcribing the interviews. Additionally, I used software to aid my analysis, and I needed to import the interview transcripts into the software.

Stage 2: Primary-Cycle Coding The next step is to begin coding the data. In primary-cycle coding, the researcher

Page 5 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1 seeks to identify meaning in a unit of data (Saldaña, 2016). While it is important to be open to many possible meanings in the data, the researcher will often have their initial research question or goal in mind at this stage. The table below shows some examples of how I identified first-order codes. As I read through the data, I sought to capture the essence of different parts of the data. The amount of data that gets a code can vary in length—sometimes one word will be coded, while other times it is an entire paragraph. As you can see below, the codes are very open-ended and emergent, and they encapsulate what is in the data.

For this project, I used NVivo 11, a qualitative data analysis software, to help me code. Software is useful because it can help you manage a large number of codes and then sort by them. For example, once I was done with coding, I could easily pull up a list of all data that I had coding with, “everyone bikes” to see how often participants talked about this concept. However, you can also code manually by writing codes on printed transcripts or in documents. Many researchers find it useful to print out the transcripts and use different colored pens to mark different codes.

Looking at the codes below, it is also important to note that a code does not need to be one word. A code can be a word, a phrase, or even a sentence that gets at the meaning present. At some point during the coding process, it becomes important for the researcher to create a codebook that lists and defines the codes used. A codebook helps the researcher to code consistently across the data and is especially useful if multiple researchers are working together.

Data (Participant names are pseudonyms) Codes

Researcher: So then when you moved to Copenhagen, tell me about getting A. Learning to cycle assimilated into biking here. around Lotte: It actually came pretty quick I think (A), uh, I had, I had a bike from home, I Copenhagen was brought it here and, I can remember, it was a little bit scary to drive the first couple of easy. times cause (B), so much traffic and stuff, not used to that, I think just walking around B. Cycling in

Page 6 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1

is just pretty scary also, so, but then everybody else is biking, all my friends are biking, so also when we’re going like to town or doing stuff, it’s always by bike (C). Copenhagen was initially scary. Were there any like unwritten rules or other things about biking that it like took C. Everyone bikes. some time to figure out when you came here? D. Some of the rules were unclear at Um. Uh. I cannot really remember, but I can remember that I, cause I am used to driving by car, so it was sometimes like, okay can I do this on a bike or not or like if first. it’s okay to, there’s red, but only for pedestrians and then you, do you need to stop or E. Cycling in do you just drive? (D) Or, of course you can’t, but I don’t think like, or at least I didn’t Copenhagen is notice, because it just comes so naturally (E). natural.

F. Usually chooses biking even Søren: We’ve always had a car, but my parents never used it to go to work for though owns instance. It’s always been bikes, it’s, well that’s continued with me as well (F). a car. Because I mean I’m 28 and I don’t have a car and I don’t have a need for it as well. G. Cars are Uh, I occasionally borrow my parents’ car, that’s about it. And even if I was given the difficult in opportunity to buy a car really cheap and get it parked and all that, I still wouldn’t Copenhagen. have a need for it and it would be more difficult to use that than the bike in H. Cars are Copenhagen (G). And more expensive as well (H). (I) expensive. I. Biking is better than driving.

Stage 3: Secondary-Cycle Coding In secondary-cycle coding, the researcher starts looking at the data in a higher level way that goes beyond the first-cycle descriptive coding. There are several ways to do this. First, one might look at the relationships between codes. Specifically, one can group together different codes to create a hierarchy of codes (Tracy, 2020). For example, participants often compared different modes of transportation in the interviews, and I was able to put several codes (i.e., biking is better than public transit, walking is better than biking, biking is better than driving) under a broader code of “comparing modes of transportation.”

A second way to conduct secondary-cycle coding that was especially insightful for this project is to begin using theoretical or disciplinary terminology (Tracy, 2020).

Page 7 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1 As I reviewed my first-cycle codes, I found that many of them related to specific material features that were used strategically to communicate to cyclists what they should do while riding a bike. The disciplinary term “affordances” describes features like these, and I returned to the data to find examples of affordances that participants described. As I revisited the data to code for affordances, I also noted that participants discussed both material affordances (things like speed bumps, textured roads that were difficult to bike over quickly, and footrests at red lights to help you stop and start biking more easily) and symbolic features like signs that were also supposed to communicate to cyclists about how to act. I noted that participants often said they ignored signs while they did comply with the material features. This observation, discovered in second-cycle coding, became the main point of my publication from these data. The table below shows examples of this level of coding.

Following this observation in secondary-cycle coding, I returned to the literature to read more about material communication and how affordances might be conceptualized as communication. Working between the existing literature and my findings, I was able to articulate a new kind of nonhuman communication in which aspects of the built environment are able to communicate to people about how they should act.

Data Codes

Researcher: Have you ever gotten a ticket for biking?

Lotte: Uh, no, but I was once told by a policeman that I had to uh, get down from my bike because I was driving on a big like, Vesterbrostorv, so what’s it called like a big A. Ignoring square? sign with Square, yeah. written rules And that was not okay, but everybody’s biking there, like all the time, and me, I didn’t get a ticket, he was just like, “Okay, you’re not allowed to bike here.” “Oh I didn’t notice!”

Page 8 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1

So then you just got off and walked the rest of the way?

Right next to the big sign of a bike with a cross over it. And I do it every morning, the same place (A). What do I care what the police officer says?

Sigrid: I mean a lot of it is about infrastructure. Cause the reason why anyone can drive a bike to work and not get dead in Copenhagen is because there’s bike paths

everywhere and there’s lights and signals to make sure the bikes can actually make B. Affordances their crossing even though there’s cars needing to go right or whatever (B). So, there’s to facilitate a lot of that. Definitely. And the fact that everybody else expects bike traffic as well. cycling Cause everybody who’s driving cars here as well, uh, are probably riding bikes at some point, so.

Summary Iterative analysis (Tracy, 2020) is an open and useful method for analyzing qualitative data. The researcher does multiple rounds of coding, both attending to emergent concepts from the data as well as guiding theoretical questions and concepts. The emphasis on both emic and etic understandings of the data are helpful for creating new theoretical understanding.

Reflective Questions

1. What are the benefits of integrating emerging ideas from the data with existing theoretical constructs? 2. What descriptive (first-cycle) codes do you identify in the data? 3. What analytic (second-cycle) codes do you identify in the data? 4. What role does existing theory and disciplinary knowledge play in iterative analysis?

Further Reading Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3, 77–101. https://doi.org/10.1191/1478088706qp063oa

Page 9 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1 Brinkmann, S., & Kvale, S. (2015). InterViews: Learning the craft of qualitative research interviewing (3rd ed.). SAGE.

Tracy, S. J. (2010). Qualitative quality: Eight “big-tent” criteria for excellent qualitative research. Qualitative Inquiry, 16(10), 837–851. https://doi.org/10.1177/ 1077800410383121

Wilhoit, E. D. (2017a). Affordances as material communication: How the spatial environment communicates to organize cyclists in Copenhagen, Denmark. Western Journal of Communication. https://doi.org/10.1080/ 10570314.2017.1306098

Wilhoit, E. D. (2017b). Photo and video methods in organizational and managerial communication research. Management Communication Quarterly, 31, 447–466. https://doi.org/10.1177/0893318917704511

References Denmark. (2016) Copenhageners love their bikes. http://denmark.dk/en/green- living/bicycle-culture/copenhageners-love-their-bikes

Emerson, R. M., Fretz, R. I., & Shaw, L. L. (2011). Writing ethnographic fieldnotes (2nd ed.). University of Chicago Press.

Gioia, D.A., Corley, K.G., & Hamilton, A.L. (2013). Seeking qualitative rigor in inductive research: Notes on the Gioia methodology. Organizational Research Methods, 16(1), 15–31. https://doi.org/10.1177/1094428112452151

Glaser, B., & Strass, A. (1967). The discovery of grounded theory. Aldine de Gruyter.

Københavns Kommune. (2017). Cykelstrategien 2011–2025. http://kk.sites.itera.dk/apps/kk_pub2/?mode=detalje&id=818

Page 10 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark SAGE SAGE Research Methods Datasets Part 2021 SAGE Publications, Ltd. All Rights Reserved. 1 Saldaña, J. (2016). The coding manual for qualitative researchers (3rd ed.). SAGE.

Srivastava, P., & Hopwood, N. (2009). A practical iterative framework for qualitative data analysis. International Journal of Qualitative Methods, 8(1), 76–84. https://doi.org/10.1177/16094069090080010

Tracy, S. J. (2020). Qualitative research methods: Collecting evidence, crafting analysis, communicating impact (2nd ed.). Wiley-Blackwell.

Webb, S., & Webb, B. (1932). Methods of social study. Cambridge University Press.

Wilhoit, E. D., & Kisselburgh, L. G. (2015). Collective action without organization: The material constitution of bike commuters as collective. Organization Studies, 36, 573–592. https://doi.org/10.1177/0170840614556916

Page 11 of 11 Iterative Analysis of Interviews About Cycling in Copenhagen, Denmark