Abstract
Abstract
Social anxiety (SA) and online privacy concerns (OPCs) are conceptually distinct fears, but both may be activated by Internet-based social contexts. Whereas SA is focused on being the object of interpersonal evaluation, OPC is focused on preventing others from gaining unauthorized access to private personal information. No research to date has investigated how SA and OPCs may uniquely or interactively predict individual differences in online interaction anxiety or attitudes and preferences about online communication. Participants (N = 374) completed the Social Phobia Inventory and measures of OPCs, online interaction anxiety, and attitudes related to online communication. The results revealed that SA and OPCs were not correlated with one another; however, they each uniquely predicted significant variance in particular outcomes, with no interactive effects. Findings help to illuminate the ways in which online communication preferences may be differentially shaped by people's levels of SA and OPCs, respectively. Theoretical implications and applications are discussed.
Introduction
S
SA and OPCs are theoretically distinct constructs. First, each is associated with a different object of fear. Whereas the primary object of fear in SA is revealing one's perceived self-flaws to the scrutiny of others, 1 OPC is rooted in fear or distrust of unknown others in cyberspace.2,3 Second, SA can occur in any social situation, whereas OPC is specific to an online medium. In fact, SA predates the advent of the Internet, but OPC only exists as a construct within the context of Internet-mediated activities. Despite these conceptual distinctions, the online context may elicit similar emotional experiences and attitudes toward methods of online communication for people high in either SA or OPCs and even more so for those high in both.
There are a number of ways that people communicate with others using the Internet, such as email, instant messaging, and audio–video chats. By selecting specific methods of interacting with others online, Internet users can manipulate the degree of anonymity (i.e., lack of audiovisual cues identifying the user's identity), asynchronicity (i.e., time lag in communication reciprocity between two parties), control over access to personal information, and privacy protection that their preferred methods of online communication provide. For example, relative to email or other text-based methods of communication, communicating online through video chat provides little anonymity or asynchronicity because of the presence of numerous audiovisual cues and the demands associated with interacting in real time. Thus, people may desire, and develop a preference for, those methods that provide them a greater degree of characteristics they prefer.
It is possible that for people high in either SA or OPCs, the online context elicits heightened state anxiety and similar attitudes toward characteristics of online communication, but for different reasons. Specifically, research suggests that people higher in SA desire anonymity in online social interactions because it reduces the risk of negative self-portrayal and negative evaluation by others.4–7 People higher in OPCs may favor anonymous online interactions because these reduce the risk of revealing personally identifying information. 8 Additionally, people higher in SA may also prefer asynchronous social interactions because in these contexts, they are able to carefully craft their messages to others, thereby minimizing the likelihood of embarrassment or rejection. 5 In contrast, people higher in OPCs may prefer asynchronicity in social interactions because it provides them the opportunity to mindfully process, select, and edit the personal information that they share with others online.9,10 Together, the research suggests that both people higher in SA and those higher in OPCs likely favor online social contexts that allow them to control the personal information that they disclose to others on the Internet. Moreover, those people higher in OPCs likely prefer online social communication methods that include tools and options to protect their privacy given that is their major concern.3,9 Yet, these features also may be attractive to people with higher SA as a means of maintaining control over their online self-presentation.
Given the potential for higher SA and higher OPCs to lead to similar emotions and attitudes (but for different reasons), it is also possible that SA and OPCs interact with one another to moderate those outcomes. Specifically, people high in SA who are also high in OPCs may experience even greater online interaction anxiety and place even greater importance on anonymity, asynchronicity, control over personal information, and privacy protection in their online social interactions than those who are high in SA, but low in OPCs. However, to our knowledge, no previous research has examined the potential unique and interactive effects of SA and OPCs on anxiety or attitudes toward methods of online communication.
In the present study, we evaluated the degree to which SA and OPCs were related to anxiety about online interaction and to attitudes about preferred methods of online communication. We predicted that SA and OPCs would be associated with one another and that as levels of each increased, so too would levels of anxiety in anticipation of online social interactions. Moreover, we expected that higher levels of SA and higher levels of OPCs would each uniquely and significantly predict stronger preferences for anonymity, asynchronicity, control over information, and protection of privacy in online communication. Finally, we hypothesized that OPCs would moderate the relationships between SA and outcomes, such that higher SA participants would report the greatest online interaction anxiety and the strongest preferences for anonymity, asynchronicity, control over information, and protection of privacy in online communication when concurrent levels of OPCs were also high.
Methods
Participants
A total of 374 adult participants, recruited on Mechanical Turk from the United States, consented to, completed, and submitted the survey. Participants ranged in age from 18 to 82 years (M = 36.0, SD = 12.7) and consisted of 30.5 percent males and 68.2 percent females, with 1.4 percent reporting their gender as other. The ethnic background of the sample was predominantly white/Caucasian (75.9 percent), followed by black/African American (9.1 percent), Asian (6.1 percent), Hispanic or Latino (4.8 percent), and other ethnicities (3.4 percent). Almost half of the participants were single (45.2 percent) or married/common-law/engaged/living with a partner (42.7 percent). Most participants identified as heterosexual (85.2 percent). Many participants (54.3 percent) possessed a degree from a college or university or had some college or university education (36.4 percent). The vast majority of participants (96.8 percent) indicated that their first language was English and that they read English very well (97.3 percent). They owned on average 4 Internet-connected devices (M = 3.9, SD = 2.6) and had used the Internet for an average of 15.3 years (SD = 4.7).
Measures
The Demographic Questionnaire included questions about participants' demographic characteristics (e.g., age, gender) and background (e.g., Internet use).
The Social Phobia Inventory (SPIN) 11 is a 17-item measure of SA. Participants rated each of the items on a five-point Likert-type scale ranging from 0 (not at all) to 4 (extremely). Full scale scores range from 0 to 68, with higher scores representing greater SA. The SPIN is a valid and reliable measure of SA. 11 In the current study, the internal consistency was excellent (α = 0.95).
The OPC questionnaire comprised four items adapted from a measure used in commercialization and business research online. 12 Participants rated the extent to which they agreed with each statement on an eight-point Likert-type scale ranging from 0 (strongly disagree) to 7 (strongly agree). Items were summed to create a total score with larger scores representing greater concerns about privacy online. Internal consistency in this study was good (α = 0.86).
Anticipated Online Interaction Anxiety was measured using a single-item anxiety thermometer. 13 Participants reported how anxious they imagined they would feel on an 11-point Likert-type scale ranging from 0 (no anxiety) to 10 (extreme anxiety) with the midpoint labeled moderate anxiety. Before responding to this question, participants read a scenario in which they imagined meeting a new acquaintance and were instructed to rate how anxious they would feel in anticipation of interacting with this person again online/using the Internet.
Online communication attitudes
We administered four items assessing attitudes about preferred features of online communication: importance of online anonymity (allows me the ability to conceal visual aspects of myself), asynchronicity (allows me time to craft the message that I want to send), privacy protection (protects my privacy most effectively), and control (gives me control over how much information the person learns). Participants reported the degree to which each feature was important for selecting their preferred methods of communicating online with the new acquaintance using a seven-point Likert-type scale ranging from not important (1) to extremely important (7).
Procedure
This study was part of a larger online survey study about online and offline social behaviors. 14 We recruited Mechanical Turk users in the United States, who had at least a 95 percent success rate on previous tasks. Following consent, participants completed the Background Questionnaire, SPIN, and OPC. Next, they read a description of meeting a new acquaintance offline with whom they were interested in connecting again over the Internet and they were asked to keep this scenario in mind while rating anticipated online interaction anxiety and online communication attitudes. At the end of the survey, participants were provided debriefing information and $1.00 remuneration (commensurate with Mechanical Turk pay for similar studies at the time). The study received ethics approval from a university ethics committee at the fourth author's institution.
Results
Preliminary analyses
We examined the extent and pattern of missing data at the total and subscale score level using Missing Values Analysis in SPSS v.22 and used expectation maximization to replace missing values. The amount of missing data was relatively small, with the most missing data occurring on the SPIN total score (5.6 percent missing). There were 25 multivariate outliers; we ran all analyses with and without these outliers and results were consistent both ways; thus, we report the results using the sample with no outliers (N = 349). 15 Means and standard deviations for each of the variables and bivariate correlations between variables are reported in Table 1. As shown in Table 1, SA and OPCs were not correlated with one another. Because of the number of analyses conducted, the threshold of significance was set at p < 0.01 to control for type 1 error.
Note. N = 349. *p < 0.05; **p < 0.01; ***p < 0.001.
OPC, online privacy concern; SA, social anxiety; SD, standard deviation.
Primary analyses
We calculated bootstrapped moderated linear regressions with 5,000 iterations using the PROCESS 16 macro for SPSS to determine the extent to which participants' level of SA and OPCs, separately and together, predicted anticipated online interaction anxiety, and the importance of features in preferred methods of online communication. The model and variable coefficients, as well as corresponding significance levels, are reported in Table 2. All the models were statistically significant at the omnibus level, accounting for 9–27 percent of variance across the models, with the main effects of SA and/or OPCs explaining unique variance in each of the examined dependent variables. However, the addition of the SA-OPC interaction term did not account for a significant increase in explained variance over and above the main effects in any of the models. As shown in Table 2, with respect to the main effects, results indicated that SA (but not OPCs) significantly predicted anticipated online interaction anxiety, and that OPCs (but not SA) predicted importance of control over information and privacy protection in preferred methods of online communication. Finally, both SA and OPCs uniquely predicted the importance of anonymity and asynchronicity in preferred methods of online communication.
Note. N = 349. SA × OPC = interaction between SA and OPC.
Discussion
To our knowledge, this was the first study to examine the unique and interactive effects of adults' SA and OPCs on online interaction anxiety and attitudes about online communication methods. We found that people who reported higher SA did not also report higher OPCs, supporting a clear distinction between these two types of concerns. Moreover, the pattern of results suggested that higher SA, but not higher OPC, was associated with greater anticipated anxiety in online social interactions. It is possible that online social interactions provoke greater anxiety for people higher in SA because the context is clearly related to key fears in SA—fears of being negatively evaluated by others. 1 The potential threats in online social interaction likely are not as clearly related to the core fear in OPCs, fear of someone accessing and misusing personal information 2 ; thus, they did not elicit anxiety. In contrast, higher OPCs, but not higher SA, predicted stronger importance toward maintaining control over information and protecting one's privacy. This finding suggests that control over information and protection of one's privacy are more similar to the core fear in OPCs than in SA. These findings are consistent with research in which only SA or OPCs were examined and with research on the appeal of anonymity and asynchronicity online for people high in SA or OPCs separately.2–10
We also found that higher levels of OPCs and higher levels of SA both uniquely and independently predicted greater importance of anonymity and asynchronicity in preferred methods of online communication, over and above the contribution of the other. However, contrary to our hypotheses, these two distinct types of concerns did not interact to moderate the importance of anonymity and asynchronicity, that is, people higher in both OPCs and SA did not necessarily report even greater importance of anonymity or asynchronicity than did those high in one or the other. These findings suggest that SA and OPCs may activate similar concerns about online social contexts; however, these concerns may occur for different reasons. For example, people higher in SA and those higher in OPCs may both place importance on anonymity and asynchronicity in their methods of online communication, but people higher in SA may do so because they believe that anonymity protects them from other people's evaluations, whereas people higher in OPCs may do so because they believe that anonymity protects their identity from being revealed or stolen.
Our findings point to a number of considerations for research on online social interaction as well as for the study of telepsychology and the delivery of Internet-based psychotherapies. 17 First, the results suggest that researchers who study online social activities and social networking may benefit from examining the role of SA and OPCs on people's online behaviors. Given the potential for both or either type of concern to independently influence similar online experiences, researchers will benefit from assessing both constructs rather than focusing exclusively on either one. Second, researchers and clinicians interested in online therapies likely will benefit from assessing the role of SA and OPCs in people's experiences and outcomes of online therapies. People higher in SA or higher in OPCs may make choices about participating in Internet-based psychotherapies related to their experiences and attitudes about methods of online communication that may hinder their ability to fully engage in or benefit from such activities.
The findings also suggest that therapists need to consider people's affective and cognitive experiences during online social interactions in their assessment of SA to better understand the role that the Internet plays in enhancing or maintaining people's anxious experiences. People high in SA may experience less overall anxiety when interacting in online than offline contexts18,19; however, our results suggest that they do still experience some anxiety during online social interactions. Research is needed on the contextual causes, moderators, and consequences of online interaction anxiety for socially anxious individuals. Our findings also highlight the types of attitudes that may interfere with or act as safety behaviors—strategies that anxious people use in anxiety-provoking situations to cope with or avoid feared outcomes 20 —during online interactions for people high in either SA or OPCs. For example, focusing and placing importance on particular features of online social communication at the expense of other features or methods may prevent people high in SA or OPCs from engaging in opportunities for pursuing and developing online social connections. This limited engagement may hinder their social flexibility and limit the rewards they are able to obtain from Internet-based social communication. Therapists could elicit and then work with these attitudes to enhance people's experiences online, which may in turn facilitate improved social interactions offline, especially for people high in SA. 18
This study is not without its limitations. As a first direct investigation of adults' SA and OPC experiences, it requires replication and extension. The brevity of the OPC questionnaire is a limitation. Research suggests that OPC likely is multidimensional with components related to information and expressive concerns. 2 We only measured information concerns; expressive concerns may be more important for SA. Researchers should investigate the relationship between SA and different types of OPCs. Additionally, our outcome variables were based on single-item ratings, which have unknown reliability and may have been influenced by measurement error or social desirability bias. We also only examined SA and OPCs as they related to one online social context (i.e., interacting with a new acquaintance). There are many online social activities and different types of social networking sites that may or may not be anxiety provoking for socially anxious people and for people who are concerned about online privacy. Researchers should examine how different contexts provoke these different types of concerns and their related beliefs and behaviors. Finally, our participants were mostly white, English-speaking, fairly well-educated American women recruited through Mechanical Turk (an online platform) who reported SA and OPCs on a continuum. Our participants reported moderate levels of SA based on published norms,15,21 with an average score on the SPIN of 20.87 and a distribution that ranged from scores of 12 and below, which are normally indicative of low SA, to scores of 30 and above, which normally resemble those of individuals with SA disorder.22,23 However, the extent to which our results may generalize to clinical samples and other populations is unknown.
Despite these limitations, our findings contribute novel insights to the literature on SA, OPCs, and online social behaviors and suggest directions for future research on Internet-based social communication and the burgeoning field of telepsychology. Measuring these constructs in tandem within the context of future research will help to enhance our understanding of the unique and interactive effects of SA and OPCs on the ways that people choose to use the Internet and social networking sites to interact with others and on ways that clinicians may be able to tailor the delivery of online therapies to accommodate particular concerns of the individual user.
Footnotes
Acknowledgments
The research reported in this article was conducted during the first author's postdoctoral fellowship and third author's Honor's thesis at the University of Waterloo. The study and manuscript preparation were partially supported by funding from the Social Sciences and Humanities Research Council of Canada (no. 435-2012-0017), Canada Research Chairs Program (nos. 950-203384 and 950-223145), and the University of Ottawa.
Author Disclosure Statement
No competing financial interests exist.
