The Influence of Social Distance and Attitudes on Processing Health Messages About Electronic

The Influence of Social Distance and Attitudes on Processing Health Messages about Electronic

Cigarettes on Social Media

Thesis

Presented in Partial Fulfillment of the Requirements for the Degree Master of Arts in the Graduate School of The Ohio State University

Shelby Wilcox, B.S.

Graduate Program in Communication

The Ohio State University

2019

Thesis Committee

Dr. Richard Huskey, Adviser

Dr. David DeAndrea

Abstract

Online contexts, especially social networking sites, are becoming a widely available space to disseminate health information and target specific populations for health campaigns. Limited evidence for health message engagement in these contexts exists. This study draws on the elaboration likelihood model and construal level theory to predict processing time and recall when individuals are presented health messages from various sources and of differing viewpoints. Participants (n = 159) were shown messages about electronic cigarettes, designed to look like tweets, from socially close and socially distant others. Processing times were highest for pro-attitudinal messages while messages from socially close sources were more likely to be recalled, and increasing social distance increased the difference in processing times for pro- and counter-attitudinal messages. We demonstrate the applicability of behavioral measures in online studies, while finding that attitudes, social distance, and their interaction affect message processing.

Keywords: Elaboration likelihood model, construal level theory, social distance, message processing, health messages

Acknowledgements

This thesis would not be possible without the tireless and unwavering support of my mentors, friends, and loved ones.

First, I must acknowledge and thank my thesis advisor, Dr. Richard Huskey. His excitement for the project never faltered, bringing a can-do, positive attitude every step of the way. Such encouragement pushed me to take risks and face challenges with confidence. Regardless of the hour or how much Dr. Huskey had on his plate, he made mentoring a priority. Without his strong stewardship and passion for research, this project and completing my master’s would not have been possible.

I would also like to thank Dr. David DeAndrea for serving as a committee member. Dr. DeAndrea’s pragmatic reasoning and impressive knowledge of theory greatly informed core arguments and the study design of the finished project. If a problem ever arose, he had a solution. He balanced critical concerns with generous praise to ensure the success of the project, as well as my personal success. Much of my confidence in my abilities to complete my thesis and program is owed to Dr. DeAndrea’s support.

I would also like to acknowledge the many female mentors that have supported me in my academic career. Thank you to: Dr. Lynn Borden for allowing me to work on her research team and helping me see the difference that research can make in someone’s life; Dr. Monica Luciana for inspiring my love of research, neuroscience, and empowering me to pursue graduate school; Dr. Teresa Lynch for modeling what it means to uplift others, and also motivating me to continue in academia; and Tatyana Matveeva who puts everyone before herself, and inspires me to be the best mentor and graduate student I can be. This project is a result of the impact their mentoring had on me.

Finally, I must express my profound gratitude to my partner, his family, my family, my friends, and the friendly faces in Derby Hall. Their enthusiasm and love were a guiding force throughout this project. Long days and late nights working on this project were made possible with the meals, care, laughter, joy, dog walks, and patience they provided.

iii

Vita

2016...... B.S Psychology, University of Minnesota, Twin Cities

Publications

Huskey, R., Wilcox, S., & Weber, R. (2018). Network neuroscience reveals distinct neuromarkers of flow during media use. Journal of Communication, 68 (5), 872-895. doi:10.1093/joc/jqy043

Fields of Study

Major field: Communication

Table of Contents

Abstract ...... ii Acknowledgements ...... iii Vita ...... iv List of Tables ...... v List of Figures ...... vi Introduction ...... 1 Overview of Motivational Influences on Message Processing ...... 3 Methods...... 12 Results ...... 19 Discussion ...... 22 Conclusion ...... 29 References ...... 30 Appendix A: Example Stimuli ...... 38

List of Tables

Table 1: Means and Standard Errors for Dependent Variables, by Condition...... 39

List of Figures

Figure 1: Sample Stimuli from the Study ...... 38

Figure 2: Interaction Between Attitude and Social Distance on Processing Time in Seconds. .... 40

Figure 3: Interaction Between Attitude and Social Distance on Sensitivity (A’) and Criterion

Bias (B”) ...... 41

vii

Introduction

Exposure to health messages occurs frequently within our everyday information environment. Considering that cognitive resources are required to process each message we are exposed to, and that cognitive resources are capacity limited (Lang, 2000, 2006, 2009), not every message can be processed with the same amount of time and effort. Motivation and ability to process messages is of key importance as to how messages are selected for further processing and therefore what types of health messages we are likely to be exposed to and engage with

(Capella, Kim, & Albarricin, 2015).

In online contexts, specifically social networking sites (SNSs), we are exposed to a number of health messages originating from a variety of different sources (Duggan & Smith,

2014). On a typical SNS an individual may see health related posts from government institutions, businesses, family, friends, or old acquaintances. Recent health campaigns are increasingly utilizing SNSs to target the public (Cugelman, Thelwall, & Dawes, 2011). Meta-analyses demonstrate the efficacy of online health campaigns by indicating they are cost effective while being more successful at targeting their audience than traditional health campaign methods

(Cugelman, Thelwall, & Dawes, 2011; Webb, Joseph, Yardley, & Michie, 2010). Previous findings suggest that likelihood of exposure can be determined by certain characteristics of message senders, message receivers, and the message itself, all of which shape which messages are likely to be engaging, accessible, and persuasive (Capella, Kim, & Albarracin, 2015).

While there are a number of message features and individual differences that shape message processing, the elaboration likelihood model (ELM; Petty & Cacioppo, 1986) provides

1 a theoretical framework for understanding motivation and ability to process a message. Two widely studied factors are message source, and if the message receiver views content as pro- or counter-attitudinal (Petty & Cacioppo, 1986). Findings from the ELM literature can be applied to an online context which may explain how exposure to a message from varying sources will affect the way a message receiver processes information. Doing so allows for an evaluation of the extent to which pro- and counter-attitudinal messages, message source social distance, and their interaction affects processing time and recall of health messages. Determining these effects is important in understanding which messages are more likely to engage an audience while also motivating positive health behaviors (Capealla, Kim, & Albarricin, 2015).

This study aims to explore how characteristics of health messages on SNSs affect an individual’s motivation for engaging with health messages. We first describe the ELM and the model’s assumptions of when message processing is more motivationally relevant, particularly as related to attitude consistency (i.e., is the message pro- or counter-attitudinal). Next, we use the literature on construal level theory (CLT) to explain the motivational influences of social distance between message sharer and receiver on message processing. We then lay out an argument for how social distance and attitudes jointly affect motivational influences of message processing.

Overview of Motivational Influences on Message Processing

According to the ELM, how we process information determines the degree to which a persuasive message is likely to be influential, and which messages are more likely to result in lasting attitude and behavior change (Petty & Cacioppo, 1986). The ELM describes a pathway to attitude change after message exposure which is contingent on the elaboration of a given message. Elaboration, as defined by Petty and Cacioppo (1986), represents the degree to which issue-relevant thinking occurs during or after message exposure. The likelihood that an individual will elaborate on message content depends on their motivation and ability to do so.

High elaboration indicates careful deliberation and scrutiny of issue-relevant processing compared to low elaboration. Importantly, individuals do not always elaborate on messages. In fact, the ELM specifies two types of ways in which messages might be processed: centrally or peripherally.

Central route processing occurs when there is increased motivation to process and high ability to process a message and involves increased elaboration. Peripheral processing occurs when ability or motivation to process a message is low. Less cognitive effort is exerted for peripheral processes compared to central route processing, and therefore, results in less message elaboration. Messages processed via the central route typically are more persuasive and lead to more consistent, long-lasting attitude changes compared to the peripheral route (Petty &

Cacioppo, 1986). Motivation to process messages depends on a number of cues; specifically we focus on message source cues and attitudes toward message content.

Attitudes Motivate Message Elaboration

Attitudes guide message processing as a function of both message content and one’s held beliefs. Fazio, Powell and Williams (1989) demonstrated that evaluation of a message occurs 3 when message content activates vivid memories of attitude related-content. Without prior exposure or information about message content, individuals will not have formed attitudes towards the message and will therefore be unable to extensively evaluate it. Importantly, individuals can hold strong or weak attitudes about issues (Fazio, 1995; Krosnick & Petty, 1995;

Levitan & Visser, 2007). Strong attitudes are those that are considered important and reflect high personal attachment, care, and concern, while also being more psychologically and motivationally relevant (Boninger, Krosnick, Berent, & Fabrigar, 1995). Stronger attitudes are less likely to change, while weaker attitudes are less resistant to change (Krosnick & Petty,

1995).

The ELM makes a number of predictions for how individuals decide to accept or reject a message by determining what a message is conveying and one’s motivation to hold accurate attitudes. Motivation to process a message depends on if the message is pro- or counter- attitudinal (Petty & Cacioppo, 1986). Counter-attitudinal messages are inconsistent with one’s beliefs and result in unfavorable thoughts and message resistance (Clark, Wegener & Fabrigar,

2008). Being presented counter-attitudinal information increases the likelihood that an individual defensively processes the message; that is, engages in processing that protects one’s preexisting attitudes. Defensive processing takes many forms, including message disengagement by limiting available cognitive resources to highly aversive stimuli or by generating counterarguments against the message (Clayton, Lang, Leshner, & Quick, 2018; Huskey, Mangus, Turner, &

Weber, 2017). The more counter-attitudinal message content is perceived to be, the more individuals are likely to defend against the content by withdrawing or counterarguing.

Messages that contain content about an issue individuals perceive to be strongly related to their held attitudes are more likely to be impactful and encourage behavior change (Fazio & 4

Powell, 1989). Strong attitudes about message content increase information processing and are more likely to result in behavioral intentions in line with the preexisting attitude. However, this is only true for pro-attitudinal messages. Capella, Yzer, and Fishbein (2003) found that drug users exposed to messages that contained strong counter-attitudinal information were more at- risk of abusing drugs than if the drug users had never been exposed to the anti-drug messages.

Such findings suggest individuals who hold strong counter-attitudinal beliefs about message content are less likely to be persuaded or adopt the advocated health behavior.

Ultimately, existing evidence demonstrates that individuals are more likely to elaborate on counter-attitudinal messages compared to pro-attitudinal messages in order to defend their strongly held attitudes (Clark, Wegener, & Fabrigar, 2008). For instance, Edwards and Smith

(1996) found participants spent more time processing counter-attitudinal messages rather than pro-attitudinal messages. Further findings found counterarguing leads to improved recall of message content (Cacioppo & Petty, 1979; Ditto & Lopez, 1992). Eagley, Kulsea, Brannon,

Shaw, and Hutson-Comeaux (2000) conducted a series of studies finding that counterarguing (a high-elaboration form of defensive processing) enhanced memories of counter-attitudinal message content even after two weeks. Taken together, these results provide evidence that individuals are motivated to defend their attitudes against counter-attitudinal messages by scrutinizing available information (see Edwards & Smith, 1996).

When individuals are more motivated to process a message, message content is made more salient and accessible (Newby-Clark, McGregor, & Zanna, 2002). In other words, conditions in which individuals are highly motivated to process a message suggests individuals will be able to recall more information from a message (Krosnick, 1989). And, as demonstrated

5 above, being presented with counter-attitudinal information increases the motivation to elaborate on a message, therefore:

H1: Individuals will spend more time processing information in counter-attitudinal messages compared to pro-attitudinal messages.

H2: Individuals will recall more information from counter-attitudinal messages compared to pro-attitudinal messages.

Social Distance Motivates Message Processing

The attitudes individuals hold are dependent on the attitudes of important others (Fazio &

Powell, 1989). More recent work demonstrates that attitudes held within one’s social network affects the strength of one’s attitude - as being surrounded by others whose attitudes are congruent with our own strengthens our attitude and lessens our susceptibility to persuasion

(Visser & Mirabile, 2004). Therefore, not only the content of the message but also the message sender’s relationship to the message receiver may affect a receiver’s motivation to process a message. Indeed, sources are highly influential in affecting how we process message content (see

Petty & Cacioppo, 1986) and a substantial amount of literature addresses source characteristic effects on message processing.

Message source effects have been measured in terms of likability, similarity, and credibility, which have all been found to affect message evaluation and persuasiveness (Eagly &

Chaiken, 1993; Petty, Wegener, & White, 1998). Furthermore, limited cognitive effort is required to evaluate a message source suggesting that source evaluations are made easily, and nearly automatically (Petty, Wegener, & White, 1998). For individuals that have low motivation to scrutinize a message, increasing source credibility increases message-relevant thinking

(Heesaker, Petty, & Cacioppo, 1979). When evaluating messages in a non-critical, low 6 elaborative manner, people are more likely to accept messages from likable sources and reject messages from unlikable sources (Chaiken, 1980; Lim, 2016). Accordingly, the likability of message sources influences message scrutiny as well as message acceptance (Chaiken, 1980;

Petty & Cacioppo 1986; Lim, 2016; Roskos-Ewoldsen, Bichsel, & Hoffman, 2002).

Summarizing these findings indicates that the more positively a message receiver evaluates a message sender, the higher message acceptance is. Recent evidence corroborates that idea showing that despite being exposed to a variety of sources and opinions in online contexts, we select and engage with messages from sources we evaluate to be most like us (Song, Cho, &

Benefield, 2018). Atkin and Rice (2013) also state that manipulating message source changes how people view the similarity, credibility, and relevancy can change how people engage with a message.

While researchers have looked at the influence of a variety of source characteristics on message processing, the closeness, or social distance, of an other to oneself is not looked at within the ELM, but may be useful in explaining source effects. Here social distance is defined as “a subjective perception or experience of distance from another person or other persons”

(Magee & Smith, 2014, pg. 2). One way of conceptualizing social distance is through construal level theory (CLT; Trope & Liberman, 2003). Essentially, CLT suggests that we evaluate the world in terms of what we know using our memories, hopes, and past experiences. Using our previous experiences to predict and speculate about the world, we form mental representations to categorize and explain phenomena. CLT is based on the concept of intertemporal processing, in that options are compared in terms of now and near versus then and later (Maglio, Trope, &

Liberman, 2013). Thus, psychological distance referred to in CLT is a measure to quantify the subjective differences between two experiences or objects in which clear, detailed, and highly 7 accessible mental representations are considered nearer, while abstract, speculative, and inaccessible mental representations are more distal (Trope & Liberman, 2003).

Distance can be represented in various ways such as geographically, temporally, or as discussed here, socially. Each distance described is represented similarly on a neural level

(Buckner & Carroll, 2007; Spreng, Mar, & Kim, 2009; Tamir & Mitchell, 2011) and evaluating social distance is as implicit as evaluating any other type of distance (Bar-Anan, Liberman,

Trope, & Algom, 2007). High distance leads to more abstract, or high-level evaluations of a target (Fujita, Eyal, Chaiken, Trope, & Liberman, 2007) while decreases in psychological distance is associated with increases in personal relevancy between oneself and a target (see

Fujita, Eyal, Chaiken, Trope, & Liberman, 2007). The more removed an individual becomes from the target of evaluation, the less reliable and available a detailed representation of that target becomes (Liberman, Trope, & Stephan, 2007). Others that are hard to evaluate and have less significance to oneself, have less clear mental representations, and are perceived as more socially distant (Liberman, Trope, & Stephan, 2007).

Importantly, as social distance decreases between message senders and receivers, the number of positive traits a receiver attributes to the sender, typically, increases (Machunsky,

Toma, Yzerbyt, & Corneille, 2014). One implication is that, as social closeness increases, the extent to which people project their own attitudes on that person increases. Said differently, individuals are more likely to assume their best friend would feel more similar about an issue than they would assume their classmate would feel towards that same issue (Machunsky, Toma,

Yzerbyt, & Corneille, 2014). The mental representations we form of, and use to evaluate, socially close others have higher personal relevancy and are more accessible compared to socially distant others. 8

Evaluating social distance entails calculating the distance between oneself and another using dimensions such as, group affiliations (Johnson, & Croson, 2006), felt closeness and familiarity with others (Aron, Aron, & Smollan, 1992), and perceived differences between one’s self and an other (Eveland, Nathanson, Detenber, & McLeod, 1999; Fiedler, 1953; Liviatan,

Trope, & Liberman, 2008; Meirick, 2005; Stephan, Liberman, & Trope, 2011). Decreased social distance between two people increases likelihood of each person behaving, and holding attitudes, more similar to the other (Liviatan, Trope, Liberman, 2008). This is supported by evidence showing that individuals who are socially closer tend to process messages more similarly than those who are socially distant (Parkinson, Kleinbaum, & Wheatley, 2018). Therefore, perceived social distance may be an underlying factor that explains why source likability, credibility, and group affiliations all influence a message receiver’s motivation to process health messages.

The ELM suggests that higher personal relevance, likability, and accessibility leads to increased motivation to engage with messages, which should lead to greater message elaboration.

If socially close others are more motivationally relevant than socially distant others, then:

H3: Participants will spend more time processing messages from socially close message senders compared to socially distant message senders.

H4: Participants will recall more information from socially close message senders compared to socially distant message senders.

Effects of Both Social Distance and Attitudes on Message Processing Motivation

To-date, little is known about how pro- or counter-attitudinal messages from a target source interact with social distance and the resulting effects on processing motivation. In some cases, individuals may be more motivated to process messages from socially distant sources. For example, previous research shows when people are presented information that results in 9 conflicting mental representations, such as a strong argument from a less credible source, they are motivated to elaborate (Lim, 2016; Ziegler & Diehl, 2001; Ziegler, Diehl, & Ruther, 2002).

In the context of the present study, when a message prompts two conflicting mental representations, the perceived importance of the conflicting information source decreases as social distance increases (Maglio, Trope, & Liberman, 2013). Such findings suggest if a message from a highly distant source contains content contrary to an individual’s pre-existing attitudes then individuals are more likely to discount, or less likely to argue against counter-attitudinal content. Similarly, individuals spend more time elaborating on counter-attitudinal messages from liked sources compared to disliked sources (Fujita et al. 2007). Ziegler and Diehl (2001) demonstrated that individuals are more likely to scrutinize, and therefore elaborate on, dislikable expert sources when those sources share messages similar to our own attitudes. Social distance, which covaries with likability (see Liviatan, Trope, Liberman, 2008) may cause similar effects on message processing.

In sum, the ELM suggests that less motivationally relevant sources sharing counter- attitudinal information will result in limited message elaboration. On the other hand, the ELM suggests that more motivationally relevant sources sharing counter-attitudinal information will increase elaboration. Using the Holbert and Park (2019) interaction typology, the proposed interaction is cleaved convergent. Specifically, as social distance decreases, the motivation to process counter-attitudinal messages decreases while motivation to process pro-attitudinal messages increases. In addition, the largest difference in the effects of attitudes on message processing will occur when social distance is low and the difference will decrease as social distance decreases. Therefore,

H5: Processing time will be longest for counter-attitudinal messages from socially close senders and shortest for pro-attitudinal messages from socially close senders.

H6: Recall will be highest for counter-attitudinal messages from socially close senders and lowest for pro-attitudinal messages from socially close senders.

Methods

Participants

For the pre-test, A total of 265 participants were recruited from Amazon Mechanical

Turk (MTurk). Responses were removed if they were indicated to be a spam or a duplicate response, failed the CAPTCHA, or did not complete the study, resulting in a final total of 196 responses. For the main study, a total of 260 undergraduate students at The Ohio State University participated with a final of n = 159 for analyses. Due to an error in the survey resulting in a failure to capture baseline processing times, 50 participant responses were excluded. Another 51 responses were excluded due to failing to complete the survey, being an outlier in their total survey processing time, or being flagged as a spam response. The number of participants was calculated using G*Power software (Faul, Erdfelder, Lang, & Buchner, 2007, 2009) with a 2 x 2 repeated measures ANOVA in which α = 0.05, power = 0.9, and a priori effect size of r = 0.10.

The necessary effect size was obtained from Rains, Levine, & Weber (2018) meta-analysis finding that messages from higher credibility sources (Stiff, 1986) were found to have an effect size of r = 0.10. Participants were over 18 and provided informed consent approved by the university’s IRB.

Design and Materials

Design. The study involved two parts, a stimulus validation pre-test and a main data collection. The main data collection experiment was a 2 (low vs high distance) x 2 (pro- vs counter-attitudinal) repeated measures design. A covariate measuring processing time to non- health messages was included as a measure of baseline message processing.

Pre-test. Participants completed the study online via MTurk. Participants consented to the study and reported their demographic information. Participants evaluated (1) the perceived 12 argument strength (Zhao, Strasser, Cappella, Lerman, & Fishbein, 2011) of text-based messages and (2) the likability (Roskos-Ewoldsen Bichsel, & Hoffman, 2002) and credibility (Chaiken &

Maheswaran, 1994) of fictitious twitter accounts sources. First, individuals were randomly presented five pro and five counter electronic cigarettes messages (from a total of 20 possible messages), all messages appeared in random order. Second, individuals were randomly presented

10 sources (from a total of 20 possible sources). Following the evaluations, participants answered a questionnaire about their smoking and electronic cigarette habits. The study took 15 minutes to complete and participants were awarded $2.08.

Main study experiment. Participants completed the study online through an Ohio State

University recruitment website using a personal laptop or desktop. Participants first consented to the study and reported their demographic information. To familiarize participants with the experimental procedure, they completed a training task where they were shown non-health related tweets. From there, participants were informed that they would be shown a series of tweets one-at-a-time. Participants were randomly assigned to one of four counterbalanced orders that fully crossed low and high social distance sources with and pro- or counter-attitudinal content. Stimuli determined from the pre-test to be similar on perceived argument strength, credibility, and reliability were used. Participants viewed stimuli one at a time through a

Qualtrics Survey and had the ability to go on to the next screen at their choosing by pressing the spacebar key. The time spent on each page was recorded in milliseconds (see Edwards & Smith,

1996). After each tweet, participants completed a counterarguing questionnaire as a manipulation check. After viewing all tweets participants reported perceived social distance of message sharer as another manipulation check. Participants were asked a series of questions about their attitudes and behaviors in regards to e-cigarette use. The final part of the study involved a signal detection 13 task. Finally, participants reported any last questions and guessed at the study’s purpose. In exchange for their time, participants received 0.5 credits of class credit.

Stimuli. Text-based messages were adapted from an online health campaign Know the

Risks (https://e-cigarettes.surgeongeneral.gov) and online advertising messages from JUUL, an electronic cigarette company. The Know the Risks campaign features messages about the risks of electronic cigarette use and are written for sharing on various social media platforms. To fit the context of the study, messages underwent minimal editing to provide clarification. For example, two messages were combined to create one message matched on word count and reading ease

(see Figure 1). Messages were classified as pro- or counter-attitudinal based on e-cigarette use.

Messages containing anti-vaping content was classified as counter-attitudinal for e-cigarette users and pro-attitudinal for non e-cigarette users. Messages containing pro-vaping content was classified as counter-attitudinal for non e-cigarette users and pro-attitudinal for e-cigarette users.

The social distance of the message sender was manipulated for the main study. Message sender was indicated by an image of a building and affiliated name in the form of a twitter profile. Low social distance message senders were indicated as either, Ohio State students,

Columbus community members, or Ohioans via their twitter handle. Profile photos were buildings near The Ohio State campus. For high social distance, message sender was manipulated by showing buildings outside of Ohio with twitter handles indicating affiliations to distant locations (i.e. Bahamas). OSU participants should evaluate sources with these characteristics as more socially close considering that people that are physically closer to a city report the city as more important and emotionally significant (Ekman & Bratfisch, 1965). To make the social distance manipulation more salient, OSU affiliation was primed using messages with non-health content about Columbus and student life during the training task. 14

Pre-Test Measures

Pre-test manipulation checks. Previous research shows that other factors like argument strength (Petty & Cacioppo, 1986; Zhao et al., 2011) shape message elaboration. Perceived argument strength of each message was evaluated using a 10-item perceived argument strength scale ranging from 1 - strongly disagree to 5 - strongly agree (Zhao, Strasser, Cappella, Lerman, and Fishbein, 2011). Cronbach’s alpha for perceived argument strength measure of each message was equal or greater than .79.

The credibility and likability of the social distance stimuli was also measured. Credibility was measured with items from Chaiken and Maheswaran (1994) asking participants to rate trustworthiness, credibility, reliability, and expertise of sources. Response options scaled from 1

- very low to 5 - very high. All Cronbach’s alpha for each source credibility measure was equal or greater than .90. Likability was measured with items from Roskos-Ewoldsen Bichsel, and

Hoffman, (2002) which asks participants to rate a sources likability, appeal, perceived positivity, and goodness with a scale from 1 at the lowest to 7 at the highest. Cronbach’s alpha for each source’s likability measure was equal or greater than .89.

Main Test Measures

Counterarguing questionnaire. Participants reported agreement with messages on a 7- point likert scale from disagree completely (1) to agree completely (7) to assess attitude directionality (Visser & Mirabile, 2004). Additionally, a 3-item questionnaire from Silvia (2006) was used to assess participants reactance and counterarguing response. Items include, “Did you criticize the message you just saw while you were viewing it?” “Did you think of points that went against what was being said while you were viewing the message?” and “While viewing the message, were you skeptical of what was being said?” Response options range from 1 - No, 15 not at all to 5 - Yes, very much. Cronbach’s alpha for each counterarguing scale was equal to or greater than .807.

Social distance. Similarity and interpersonal closeness to each target condition was also measured. Similarity was measured by having participants indicate how similar the message sender was to themselves, and how close they felt to the message sender with response options being “1 - not at all” to “9 - very much” (see Livitan, Trope, & Liberman, 2008). Interpersonal closeness was measured using Aron, Aron, and Smollan’s (1992) Inclusion of Others in Self

Scale. Participants choose from a selection of circles differing in their degree of overlap with 1- separate to 7- completely overlapping to indicate the closeness of the target message sender to themselves. Cronbach’s alpha for each social distance scale was greater than or equal to .883.

Electronic Cigarette Use. To obtain a profile of participant’s smoking behaviors, participants reported how many times in the last 30 days they used an e-cigarette. E-cigarette users were defined as participants who used e-cigarettes in the past 30 days as defined in similar previous research (see Pei, Schmälzle, O’Donnell, Kranzler, & Falk, 2019). Participants labeled as e-cigarette users were coded to agree (pro-attitudinal) with pro-vaping messages and disagree

(counter-attitudinal) with anti-vaping messages. Participants labeled as non e-cigarette users were coded to disagree (counter-attitudinal) with pro-vaping messages and agree (pro-attitudinal) with anti-vaping messages.

Signal Detection Task. A signal detection task was administered to measure message recall. The task involved showing participants 32 messages, 16 messages clipped from health messages shown to participants in the study (target messages) and 16 messages clipped from health messages not shown to participants (foil messages). Both were 10 word fragments with the target messages adopted from the 8 stimuli messages and the foil messages adopted from the 16 unused messages from the pre-testing. Participants randomly viewed one message at a time on screen for 8 seconds and selected ‘yes’ or ‘no’ to indicate if they had previously seen the fragment, or not.

Data analysis

Missing values for responses to counterarguing and social distance measures underwent

Hotdeck imputation (see Myers, 2011). Hypotheses related to message processing time were evaluated using a repeated measures ANCOVA with attitude (pro or counter) and social distance

(low or high) as within-subjects factors, and average processing time for non-health measures as a baseline measure and covariate. Following guidelines from Ratcliff (1993), processing times were cleaned by replacing outliers above the 95% confidence interval for each message with the value 2 standard deviations above the mean of each message. Baseline processing time was mean centered. For means and standard errors for dependent measures by condition, see table 1.

Hypotheses related to message recall were evaluated using a repeated measures ANOVA with attitude (pro or counter) and social distance (low or high) as within-subjects factors. Signal detection theory (see Shapiro 1994) was applied to determine response accuracy in a recall test.

Following procedures by Stainslaw and Todorov (1999), sensitivity (A’) and criterion bias (B”) will be calculated. A’ represents a measure of how well participants distinguish between targets and foils, with larger values suggesting higher sensitivity (better memory; Shapiro, 1994). B” represents a threshold measure wherein larger positive values represents a higher, more conservative threshold for participants to indicate they remember the information and larger negative values represent a lower, more liberal threshold for participants to indicate they remember the information. All pairwise comparisons were Bonferroni corrected to control

17 family-wise error rates. For ANOVA and ANCOVA models, results from the multivariate tests are reported as they are more robust against violations of normality and sphericity.

Open Science Practices

Study design, hypotheses, and analysis plan were pre-registered in accordance with Open

Science Practices. Furthermore, all data, materials, and code used for this study are available on an Open Science Framework project page

(https://osf.io/84dhx/?view_only=369df030060348fa9e358d22f2e0fac8).

Results

Pre-test

Text based messages were evaluated on perceived argument strength and sources were evaluated on their credibility and likability for selection in the main study. Each stimulus was viewed by between 42 - 54 participants. The dataset was split by e-cigarette use (users/non- users) and the mean and 95% confidence intervals of perceived argument strength for each message and perceived likability and credibility for each source were calculated. Messages and sources selected for final the stimuli creation did not differ (as indicated by overlapping confidence intervals) on perceived argument strength, credibility, or likability both within and between e-cigarette users. A total of eight messages (four pro- and four counter-attitudinal) and eight sources (four socially close and 4 socially distant) were selected for the main study.

Main study Manipulation checks. Results of a paired-samples t-test demonstrated that participants, on average, significantly counterargued more when presented counter-attitudinal messages (M =

3.32, SD = .84) than pro-attitudinal messages (M = 2.50, SD = .85), t(158) = 8.74, p < .001,

Cohen’s d = .98. In addition, results of a paired-samples t-test demonstrated that participants, on average, significantly felt closer to sources categorized socially close (M = 4.01, SD = 2.00) compared to sources categorized as socially distant (M = 2.78, SD = 1.64), t(158) = 8.62, p <

.001, Cohen’s d = .67.

Hypothesis Testing. Hypotheses 1 and 3 were tested using a 2 (attitude) × 2 (distance) repeated-measures ANCOVA with baseline processing time as a covariate (see figure 2). We

19 expected that processing times would be longest for counter-attitudinal messages (H1) and socially close sources (H3). Results showed that attitude significantly affect processing time

(F(1, 157) = 10.19, p = 0.002, Wilks’ λ = .94), with longer processing times for pro-attitudinal

(M = 9.64, SE = .19) compared to counter-attitudinal messages (M = 9.06, SE = .19). Although the result was significant, it was in the opposite direction of H1, so H1 was not supported.

Distance did not significantly affect processing time (F(1, 157) = 0.22, p = 0.64, Wilks’ λ = .99).

Therefore, H3 was not supported.

We posited a cleaved convergent interaction (see Holbert and Park, 2019) for H5, such that the effect of attitudes decreases as social distance decreases. Instead, a significant cleaved divergent interaction was found such that a counter-attitudinal and pro-attitudinal message from a close source resulted in similar processing times, but counter-attitudinal messages from a distant source resulted in significantly less processing time than pro-attitudinal messages from a distant source (F(1, 157) = 4.07, p = 0.05, Wilks’ λ = .98). Therefore, the results, although significant, do not support H5.

Hypotheses 2 and 4 were tested using a repeated measures ANOVA to determine two measures of recall based on signal detection theory, recognition accuracy (A’) and criterion bias

(B”) (see figure 3). We expected that recognition accuracy would be highest, and that criterion bias would be lowest for counter-attitudinal messages (H2) and socially close sources (H4).

Results indicated that participants, on average, did not differ in their recognition accuracy (A’)

(F(1,157) = 0.88, p = 0.35, Wilks’ λ = .99) or criterion bias (B”) (F(1,157) = 0.57, p = 0.45,

Wilks’ λ = .99) of counter-attitudinal and pro-attitudinal messages. Therefore, H2 was not supported. However, in support of H4 participants had higher recognition accuracy (F(1,157) =

15.42, p < 0.001,Wilks’ λ = .99) and lower criterion bias (F(1,157) = 16.37, p < 0.001, Wilks’ λ

= .91) for socially close compared to socially distant sources.

We posited a transverse negative contributory interaction for H6, with processing time highest for counter-attitudinal messages from socially close sources and highest for pro- attitudinal messages from socially distant sources. No significant interaction between attitude and distance was determined for recognition accuracy (F(1,157) = 0.95, p = 0.33, Wilks’ λ = .99) or criterion bias (F(1,157) = 0.05, p = 0.81, Wilks’ λ = 1.00). Therefore, H6 was not supported.

Discussion

We applied the elaboration likelihood model (ELM) and construal level theory (CLT) to better understand how people engage with health messages, specifically how people’s motivation shapes message processing. In this study, participants viewed health messages related to e- cigarettes that either supported e-cigarette use or advocated against e-cigarette use from socially close or distant sources. Here, we discuss the findings of this study and the implications of the results in terms of theoretical and practical contributions.

Effects of Attitudes on Message Processing

While we found a significant relationship between attitude consistency and processing time, the relationship was in the opposite direction we predicted. When participants were presented a pro-attitudinal message, they spent a significantly longer time engaging with the message compared to counter-attitudinal messages. These results may be due in part to the findings that e-cigarette users, in general, appear to have an attentional bias to e-cigarette content. Specifically, positive attitudes towards e-cigarette use has been previously associated with longer time looking at stimuli that contain cues related to e-cigarettes (Lochbuehler,

Wileyto, Tang, Mercincavage, Cappella, & Strasser, 2018). E-cigarette users in this study have comparatively longer processing times for pro-attitudinal (t(157) = -.72, p = .47) and counter- attitudinal messages from socially close sources than non e-cigarette users (t(157) = -.57, p =

.57). E-cigarette users, however, spent less time processing pro-attitudinal messages (t(157) =

2.57, p = .01) and more time processing counter-attitudinal messages from socially distant compared to non e-cig users (t(157) = -3.53, p = .001). This pattern of results seems to support 22 the notion that e-cigarette users are more sensitive to e-cigarette cues, regardless of source characteristics, making it more difficult to detect differences in processing times between conditions.

Interestingly, attitudes did not affect message recall despite participants spending more time engaging with pro-attitudinal sources. One explanation may be that people are equally likely to remember pro-attitudinal and counter-attitudinal messages if presented in low-arousing, congenial formats (Eagley, Kulsea, Brannon, Shaw, & Hutson-Comeaux, 2000; Liu & Bailey

2018). On the other hand, results confirmed our hypothesis (H4) that individuals recalled more from socially close sources compared to socially distant sources despite distance not significantly affecting processing time (H2). People seem better at discriminating between new information from old information (A’) and had a lower threshold for recalling old information (B”) when the information came from socially close sources. Our results are in line with previous research suggesting that we place more importance on socially close others and discount the important of socially distant sources (Machunsky, Toma, Yzerbyt, & Corneille, 2014).

Theoretical Contributions

According to the ELM, we would expect that individuals are more motivated to elaborate on messages salient to their attitudes (Petty & Cacioppo, 1986). Messages that focus on issues that people view as less salient to their attitudes are associated with more elaboration (measured by thought-listing) for pro-attitudinal messages compared to counter-attitudinal messages (Clark et al., 2008). Considering that participants here spent more time engaging with pro-attitudinal messages, our findings suggest that participant‘s attitudes towards e-cigarettes were not particularly salient. These findings also suggest people may prefer engaging with messages that

23 align with their attitudes. Said differently, people seem more averse and likely to disengage with messages they find contradictory to their attitudes.

Furthermore, attitudes affected processing time but not recall which supports previous research on message elaboration that pro-attitudinal and counter-attitudinal information are equally likely to be recalled (Bigsby, Monahan, & Ewoldsen, 2017). However, these findings contradict previous experimental research that people are less likely to remember (as measured by recognition or signal detection theory tasks) information from threatening messages that elicit stronger counterarguing responses (Clayton et al., 2018; Clayton, Keene, Leshner, Lang, and

Bailey, 2019). The important difference between when individuals are more likely to recall counter-attitudinal or pro-attitudinal messages seems to be if a message elicits a strong emotional or threatening response. Another possible difference to explain discrepant findings is whether a study includes smokers only. Our more generalizable study using non-smokers and smokers might yield different findings than studies using only non-smokers that find a positive relationship between counterarguing and recall. Notably, few studies that apply an ELM approach have measured recall using signal detection theory. Including behavioral measures of memory like sensitivity (A’) and criterion bias (B”) in future studies may better clarify how differences processing motivation affects memory of information.

CLT suggests that differences in distance are cognitively evaluated similarly with closer distances being perceived as more important than farther distances (Trope & Liberman, 2003).

Previous research has shown that source characteristics (e.g., credibility, likability) operate as heuristic cues that bias people towards peripheral processing (Petty, Wegener, & White, 1998).

In our study, source cues seem to bias people towards central processing. Even when holding credibility and likability constant, we show that social distance of a source is a distinct 24 characteristic, and encompasses multiple dimensions like felt closeness, similarity, and self-other overlap. Differences in perceived social distance may underpin how source characteristics are evaluated. We provide a novel test of how social distance cues affects sensitivity (A’) to information and confidence in reporting familiarity with information (B”). In line with CLT predictions that socially close sources are more relevant and mentally accessible, our findings suggest information from socially close sources is more mentally accessible (better recall).

Clearly, social distance impacts message processing by affecting message receivers memory.

Researchers interested in how sharing behaviors occur in social networks or the effects of interpersonal discussion may want to consider using a CLT framework.

The interaction between attitudes and social distance has important implications for the

ELM and CLT. We observed a significant interaction between pre-existing attitudes and social distance for processing time. Specifically, attitudes weakly affect processing time when a message is from a socially close source but attitudes strongly affect processing time from a socially distant source. As discussed in Holbert and Park (2019), this cle 9aved divergent interaction carries significant weight because competing effects are observed based on the level of the factor. Attitudes become more motivationally relevant when the source becomes less motivationally relevant during message processing. By comparison, the interaction between attitudes and social distance on A’ and B” have a divergent positive contributory relationship

(see Holbert & Park, 2019), although this finding was not statistically significant. It seems that attitudes are equally motivationally relevant during recall regardless of the source. In sum, attitude effects on message processing in online context may be particularly susceptible to social distance effects. Such effects should be considered carefully when predicting message processing using an ELM framework. 25

Practical Contributions

Social networking sites and interpersonal discussion increase the possible reach and exposure to health campaign messages while reinforcing target health messages (Brennan,

Durkin, Wakefield, & Kashima, 2016; Hwang, 2010; Kranzler, Schmälzle, O’Donnell, Pei, &

Falk, 2019 Southwell & Yzer, 2009). Our results indicate that individuals are more likely to process messages from close sources, especially if the message is pro-attitudinal. Seeing counter- attitidunal messages from distant sources are more likely to be glanced over and forgotten which is particularly problematic for health campaigns aimed at changing health attitudes or behaviors.

More alarming is that exposure to tweets that contain pro-substance use content is positively correlated to engaging in risky behavior (Moreno & Whitehill, 2014). That finding combined with findings from this study suggest targeting individuals with messages perceived as more relevant may not only be remembered more, but impact behavioral choices. While previous studies suggest that government organizations like the CDC are perceived as highly credible (for review, see Dutta-Bergman, 2003), messages from such sources may not be as engaging, or memorable to target populations as socially close sources. Instead, health campaigns may benefit by targeting specific populations and creating accounts that appear similar to the target audience.

Indeed, a health message from someone perceived as more homogenous to oneself is more likely to result in attitudes and behavioral intentions that align with the message content (Morgan,

Miller, & Arasaratnam, 2002). A fruitful area for further investigation is the interaction between socially close sources sharing credible sources such as the CDC. Creating messages that are motivationally relevant will result in individuals spending more time engaging with a message and remembering the message.

Limitations 26

Some aspects of the study may limit the conclusions made. We did not specifically measure attitude toward e-cig use in this study. Instead, we categorized attitudes by measuring e- cig use behavior. Although unlikely, it’s possible participants attitudes towards e-cig use may differ from their behavior. Attitudes about substance can be difficult to measure. Numerous scales and behavioral measures have been developed to measure attitudes about smoking or substance use, with little consensus on which measures are best (White, Chan, Repetto, Gratale,

Capella, & Albarracin, 2019). Previous research though supports the decision to link behavior to attitudes finding that individuals are much more likely to behave and have behavioral intentions similar to their attitudes (Ajzen, 1991). Furthermore, attitudes have been demonstrated to be predictive of substance use (Perkins, 2007). Another possible limitation of the study is that only a relatively small number of e-cig users (n = 38) participated. However, our results are reflective of previous findings that among college students, about 14% report using e-cigarettes within the past 30 days (Littlefield, Gottlieb, Cohen, & Trotter, 2015). Furthermore, this limitation is mitigated considering the study is conceptualized on a strong theoretical basis and an a-priori power analysis derived from a meta-analysis (Rains et al., 2018).

Future Directions

Future studies may benefit from exploring how counter-arguing impacts message recall and processing time. Counter-arguing may not be an accurate indicator of individuals motivation to process a message, and messages that elicit high counter-arguing may actually impede health campaigns efficacy. Exploring how message recall and processing time affects retransmission of information may be another fruitful area of future studies. Currently, exciting new work is looking at information sharing and message processing on a neural level (see Cooper, Garcia,

Tompson, O’Donnell, Falk, & Vettel, 2019; Scholz, Baek, O’Donnell, & Falk, 2019). 27

Retransmission of health related content appears driven by an individual’s desire to defend their attitudes and engage with attitude consistent content (Capella et al., 2015). Impression and relationship management motives also affect retransmission, and the potential audiences (i.e. email vs twitter) further affect retransmission with individuals sharing more attitude consistent content with audiences perceived as more like the self (Kim et al., 2015). While we use a self- report to measure social distance, or how self-like an other is, a more thorough measure of social distance may be social network organization (see Parkinson, 2016). Extending this study on a larger scale and looking at how individuals are related to sources based on their social network organization may strengthen the findings here.

Conclusion This study examined how attitude accessibility and social distance of message senders affects processing time and message recall. We take a novel approach using online studies to collect behavior data, and made all of the necessary code and data to extend and replicate this research available using open science practices (see Poldrack, 2012). Messages from socially close sources are recalled more, and messages perceived as pro-attitudinal result in longer time spent processing the message. Although the ELM suggests that attitudes are a key determinant in message related outcomes, such as elaboration and memory, attitudes only affected processing time. Health campaigns should be made to appear socially close to target populations considering those messages were more likely to be remembered and result in longer processing times. Taken together, this evidence suggests further examination between social distance effects on message processing, and differences between message processing and message elaboration in online context.

References

Anderson, S., & Hauck, W. W. (1983). A new procedure for testing equivalence in comparative

bioavailability and other clinical trials. Communications in Statistics—Theory and

Methods, 12, 2663–2692. doi:10.1080/03610928308828634

Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of other in the self scale and the structure

of interpersonal closeness. Journal of Personality and Social Psychology, 63, 596-612.

Atkin, C. K., & Rice, R. E. (2013). Theory and principles of public communication campaigns.

In R. E. Rice & C. K. Atkin (Eds.), Public communication campaigns (pp. 3–19).

Thousand Oaks, CA: Sage

Bar-Anan, Y., Liberman, N., Trope, Y., & Algom, D. (2007). Automatic processing of

psychological distance: Evidence from a Stroop task. Journal of Experimental

Psychology: General, 136, 610–622. doi:10.1037/0096-3445.136.4.610

Blackwelder, W. C. (1982). “Proving the null hypothesis” in clinical trials. Controlled Clinical

Trials, 3, 345–353. doi:10.1016/0197-2456(81)90059-3

Buckner, R. L., & Carroll, D. C. (2007). Self-projection and the brain. Trends in Cognitive

Sciences, 11, 49–57. doi:10.1016/j.tics.2006.11.004

Cappella, J.N., Yzer, M.C., Fishbein, M. (2003). Using beliefs about positive and negative

consequences as the basis for designing message interventions for lowering risky

behavior. In: Romer D. editors. Reducing Adolescent Risk: Toward an Integrated

Approach, pp. 210–220. Thousand Oaks, CA: Sage Publications

Clark, J. K., Wegener, D. T., & Fabrigar, L. R. (2008). Attitudinal ambivalence and message-

based persuasion: Motivated processing of proattitudinal information and avoidance of

counterattitudinal information. Personality and Social Psychology Bulletin, 34(4), 565-

577. doi:10.1177/0146167207312527

Cohen, J., Mutz, D., Price, V., & Gunther, A. (1988). Perceived impact of defamation: An

experiment on third-person effects. Public Opinion Quarterly, 52(2), 161-173.

Cugelman, B., Thelwall, M., & Dawes, P. (2011). Online interventions for social marketing

health behavior change campaigns: a meta-analysis of psychological architectures and

adherence factors. Journal of Medical Internet Research, 13(1), e17.

doi:10.2196/jmir.1367

Duggan, M., & Smith, A. (2014). Social media update 2013. Pew Research Center. Retrieved

June 28th, 2018: http://www.pewinternet.org/files/2013/12/PIP_Social-Networking-

2013.pdf

Dutta-Bergman, M. (2003). Trusted online sources of health information: Differences in

demographics, health beliefs, and health-information orientation. Journal of Medical

Internet Research, 5(3), e21. doi:10.2196/jmir.5.3.e21

Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Harcourt Brace Jovanovich

College Publishers.

Eagly, A. H., Kulesa, P., Brannon, L. A., Shaw, K., & Hutson-Comeaux, S. (2000). Why

counterattitudinal messages are as memorable as proattitudinal messages: The importance

of active defense against attack. Personality and Social Psychology Bulletin, 26(11),

1392-1408. doi:10.1177/0146167200263007

Edwards, K., & Smith, E. E. (1996). A disconfirmation bias in the evaluation of arguments.

Journal of Personality and Social Psychology, 71(1), 5-24.

Ekman, G., & Bratfisch, O. (1965). Subjective distance and emotional involvement. A

psychological mechanism. Acta Psychologica, 24, 430–437. doi:10.1016/0001-

6918(65)90027-2

Eveland, W. P., Jr., Nathanson, A. I., Detenber, B. H., & McLeod, D. M. (1999). Rethinking the

social distance corollary: Perceived likelihood of exposure and the third-person

perception. Communication Research, 26, 275-302. doi:10.1177/009365099026003001

Farr, J. N., Jenkins, J. J., & Paterson, D. G. (1951). Simplification of Flesch Reading Ease

Formula. Journal of Applied Psychology, 35(5), 333-337. doi:10.1037/h0062427

Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical

power analysis program for the social, behavioral, and biomedical sciences. Behavior

Research Methods, 39, 175-191. doi:10.3758/BF03193146

Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using

G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods,

41, 1149-1160. doi:10.3758/BRM.41.4.11

Fazio, R. H., Powell, M. C., & Williams, C. J. (1989). The role of attitude accessibility in the

attitude-to-behavior process. Journal of Consumer Research, 16(3), 280-288.

doi:10.1086/209214

Fiedler, F. E. (1953). The psychological-distance dimension in interpersonal relations. Journal of

Personality, 22, 142-150. doi:10.1111/j.1467-6494.1953.tb01803.x

Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7(2), 117-140.

doi:10.1177/001872675400700202

Fujita, K., Eyal, T., Chaiken, S., Trope, Y., & Liberman, N. (2008). Influencing attitudes toward

near and distant objects. Journal of Experimental Social Psychology, 44(3), 562-572.

doi:10.1016/j.jesp.2007.10.005

Hwang Y. (2010). Social diffusion of campaign effects. Communication Research, 39(1), 120-

141. doi:10.1177/0093650210389029

Johnston, L. D., Bachman, J. G., & O’Malley, P. M. (1982). Student drug use, attitudes, and

beliefs: National trends, 1975-1982.

Jones, B., & Rachlin, H. (2006). Social discounting. Psychological Science, 17(4), 283-286.

doi:10.1111/j.1467-9280.2006.01699.x

Krosnick, J. A. (1989). Attitude importance and attitude accessibility. Personality and Social

Psychology Bulletin, 15(3), 297-308. doi:10.1177/0146167289153002

Lang, A. (2000). The limited capacity model of mediated message processing. Journal of

Communication, 50(1), 46–70. doi: 10.1111/j.1460-2466.2000.tb02833.x

Lang, A. (2006). Using the limited capacity model of motivated mediated message processing to

design effective cancer communication messages. Journal of Communication, 56(1),

S57–S80. doi:10.1111/j.1460-2466.2006.00283.x

Lang, A. (2009). The limited capacity model of motivated mediated message processing. In R. L.

Nabi & M. B. Oliver (Eds.), The SAGE Handbook of Media Processes and Effects (pp.

193–204). Thousand Oaks, CA: Sage Publications.

Levine, T. R., Weber, R., Park, H. S., & Hullett, C. R. (2008). A communication researchers’

guide to null hypothesis significance testing and alternatives. Human Communication

Research, 34, 188–209. doi:10.1111/j.1468- 2958.2008.00318.x

Liberman, N., Trope, Y., & Stephan, E. (2007). Psychological distance. Social psychology:

Handbook of Basic Principles, 2, 353-383.

Littlefield, A. K., Gottlieb, J. C., Cohen, L. M., & Trotter, D. R. (2015). Electronic cigarette use

among college students: Links to gender, race/ethnicity, smoking, and heavy drinking.

Journal of American College Health, 63(8), 523-529.

doi:10.1080/07448481.2015.1043130

Liviatan, I., Trope, Y., & Liberman, N. (2008). Interpersonal similarity as a social distance

dimension: Implications for perception of others’ actions. Journal of Experimental Social

Psychology, 44(5), 1256-1269. doi:10.1016/j.jesp.2008.04.007

Lochbuehler, K., Wileyto, E. P., Tang, K. Z., Mercincavage, M., Cappella, J. N., & Strasser, A.

A. (2018). Do current and former cigarette smokers have an attentional bias for e-

cigarette cues?. Journal of Psychopharmacology, 32(3), 316-323.

doi:10.1177/0269881117728418

Machunsky, M., Toma, C., Yzerbyt, V., & Corneille, O. (2014). Social projection increases for

positive targets: Ascertaining the effect and exploring its antecedents. Personality and

Social Psychology Bulletin, 40(10), 1373-1388. doi:10.1177/0146167214545039

Maglio, S. J., Trope, Y., & Liberman, N. (2013). Distance from a distance: Psychological

distance reduces sensitivity to any further psychological distance. Journal of

Experimental Psychology: General, 142(3), 644-657. doi:10.1037/a0030258

Meirick, P. C. (2005). Rethinking the target corollary: The effects of social distance, perceived

exposure, and perceived predisposi•tions on first-person and third-person perceptions.

Communica•tion Research, 32, 822-843. doi:10.1177/0093650205281059

Myers, T. A. (2011). Goodbye, listwise deletion: Presenting hot deck imputation as an easy and

effective tool for handling missing data. Communication Methods and Measures, 5(4),

297-310. doi:10.1080/19312458.2011.624490

Newby-Clark, I. R., McGregor, I., & Zanna, M. P. (2002). Thinking and caring about cognitive

inconsistency: When and for whom does attitudinal ambivalence feel uncomfortable?.

Journal of personality and social psychology, 82(2), 157- 166. doi:10.1037/0022-

3514.82.2.157

Poldrack, R. A. (2012). The future of fMRI in cognitive neuroscience. Neuroimage, 62(2), 1216–

1220. doi:10.1016/j.neuroimage.2011.08.007

Parkinson, C. M. (2016). How we connect: Neural mechanisms underpinning human social

networks. Dartmouth College.

Parkinson, C., Kleinbaum, A. M., & Wheatley, T. (2018). Similar neural responses predict

friendship. Nature Communications, 9(1), 1–14. doi:10.1038/s41467-017-02722-7

Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. Advances

in Experimental Social Psychology, 19, 123-205. doi:10.1016/S0065-2601(08)60214-2

Petty, R. E., Wegener, D. T., & White, P. H. (1998). Flexible correction processes in social

judgment: Implications for persuasion. Social Cognition, 16, 93-113.

doi:10.1521/soco.1998.16.1.93

Rains, S. A., Levine, T. R., & Weber, R. (2018). Sixty years of quantitative communication

research summarized: lessons from 149 meta-analyses. Annals of the International

Communication Association, 42(2), 105-124. doi:10.1080/23808985.2018.1446350

Ratcliff, R. (1993). Methods for dealing with reaction time outliers. Psychology Bulletin. 114,

510–532. doi:10.1037/0033-2909.114.3.510 35

Rozeboom, W. W. (1960). The fallacy of the null hypothesis significance test. Psychological

Bulletin, 57, 416–428. doi:10.1037/h0042040

Song, H., Cho, J., & Benefield, G. A. (2018). The dynamics of message selection in online

political discussion forums: Self-segregation or diverse exposure?. Communication

Research, 1-28. doi:10.1177/0093650218790144

Spreng, R. N., Mar, R. A., & Kim, A. S. (2009). The common neural basis of autobiographical

memory, prospection, navigation, theory of mind, and the default mode: A quantitative

meta-analysis. Journal of Cognitive Neuroscience, 21, 489–510.

doi:10.1162/jocn.2008.21029

Stephan, E., Liberman, N., & Trope, Y. (2011). The effects of time perspective and level of

construal on social distance. Journal of Experimental Social Psychology, 47, 397-402.

doi:10.1016/j.jesp.2010.11.001

Tamir, D. I., & Mitchell, J. P. (2011). The default network distinguishes construals of proximal

versus distal events. Journal of Cognitive Neuroscience, 23, 2945–2955.

doi:10.1162/jocn_a_00009

Trope Y, Liberman N. (2003) Temporal construal. Psychological Review,110, 403–421.

doi:10.1037/0033-295X.110.3.403

Tryon, W. W. (2001). Evaluating statistical difference, equivalence, and indeterminacy using

inferential confidence intervals: An integrated alternative method of conducting null

hypothesis statistical tests. Psychological Methods, 6, 371–386.

doi:10.3102/1076998609332753

Visser, P. S., & Mirabile, R. R. (2004). Attitudes in the social context: the impact of social

network composition on individual-level attitude strength. Journal of Personality and

Social Psychology, 87(6), 779- 785. doi:10.1037/0022-3514.87.6.779

Webb, T. L., Joseph, J., Yardley, L., & Michie, S. (2010). Using the internet to promote health

behavior change: a systematic review and meta-analysis of the impact of theoretical

basis, use of behavior change techniques, and mode of delivery on efficacy. Journal of

Medical Internet Research, 12(1), e4. doi:10.2196/jmir.1376

Weber, R., & Popova, L. (2012). Testing equivalence in communication research: Theory and

application. Communication Methods and Measures, 6(3), 190–213.

doi:10.1080/19312458.2012.703834

Westerwick, A., Kleinman, S. B., & Knobloch-Westerwick, S. (2013). Turn a blind eye if you

care: Impacts of attitude consistency, importance, and credibility on seeking of political

information and implications for attitudes. Journal of Communication, 63(3), 432-453.

doi:10.1111/jcom.12028

White, H. A. (1997). Considering interacting factors in the third-person effect: Argument

strength and social distance. Journalism & Mass Communication Quarterly, 74(3), 557-

564. doi:10.1177/107769909707400309

Ziegler, R., & Diehl, M. (2001). The effect of multiple source information on message scrutiny:

The case of source expertise and likability. Swiss Journal of Psychology, 60, 253-263.

doi:10.1024//1421-0185.60.4.253

Ziegler, R., Diehl, M., & Ruther, A. (2002). Multiple source characteristics and persuasion:

Source inconsistency as a determinant of message scrutiny. Personality and Social

Psychology Bulletin, 28(4), 496-508. doi:10.1177/0146167202287007 37

Appendix A: Example Stimuli

Figure 1. Sample stimuli from the study, A) depicts a non-health message shown during the training task, B) depicts a socially close source and pro e-cigarette message, and C) depicts a socially distant source and anti e-cigarette message.

Table 1. Means and standard errors for dependent variables, by condition.

Figure 2. Interaction between attitude and social distance on processing time in seconds. Results of repeated measures ANCOVA show that attitudes and the interaction between attitudes and distance significantly influence processing time at the .05 level.

Figure 3. Interaction between attitude and social distance on sensitivity (A’) and criterion bias

(B”). Results of repeated measures ANOVA show that distance significantly influences A’ and

B” at the .05 level, but not attitudes or their interaction.