Who prefers anonymous self-expression online? A survey-based study of Finns aged 15–30 years

Abstract

The dominance of computer-mediated communication in the online relational landscape continues to affect millions of users; yet, few studies have identified and analyzed characteristics shared by those specifically valuing its anonymous aspects for self-expression. This article identifies and investigates key characteristics of such users through online survey by two samples of Finnish users of social networking sites aged 15–30 years (n = 1013; 544). Various characteristics espoused by those especially valuing anonymity for self-expression online were identified and analyzed in relation to the users in question. Favoring anonymity was positively correlated with both grandiosity, a component of narcissism, and low self-esteem. In addition, users with stronger anonymity preference tended to be younger, highly trusting, having strong ties to online communities while having few offline friends. Findings emphasize the significance of a deeper understanding of how anonymity effects and attracts users seeking its benefits while also providing new insights into how user characteristics interact depending on motivation.

Keywords:

• anonymity
• internet
• youth
• social networking

Introduction

Computer-mediated communication has become a pivotal mechanism in the interactional patterns of much of the world. This rapid advancement of technological innovation in the area of connectivity has enabled a dynamic evolution toward increasingly internet-enabled interaction. This has, in turn, allowed for an enhanced capability toward the development of new social networks through various social networking sites (SNS), allowing users and especially young people to become more self-networking than ever before (Allen, Szwedo, & Mikami, 2012; Davidson & Martelozzo, 2013; Lehdonvirta & Räsänen, 2011; Näsi, Räsänen, & Lehdonvirta, 2011). These SNS can be both specific and general, in that one can focus on specific populations or activities by way of a virtual platform where users can present themselves as they wish to be seen while managing networks and establishing new connections (Ellison, Steinfield, & Lampe, 2007). Networking online is a widespread phenomenon, with SNS use accounting for approximately one-quarter of all time spent on the internet in addition to nearly 80% of internet users reporting the use of SNS, with young people continuing to be its most significant consumers, especially in a relational capacity (Panek, Nardis, & Konrath, 2013; Subrahmanyam & Lin, 2007).

This enhanced capacity in both scope and convenience for interaction through SNS is carried out in a computer-mediated fashion and as such, the fullness of face-to-face cues is typically diminished. There exist various levels of user anonymity online, where cues and consequences of traditional face-to-face communication are altered or even absent. In the anonymous state, there are fundamental alterations to physical feedback loops based on reaction and interpretation of characteristics (Burke, 1991; Carver & Scheier, 1981), along with reputation effects where long-term identifiability is central (Raub & Weesie, 1990). This anonymous aspect of relational internet-use can influence users desiring a relative lack of accountability toward an increased willingness to share aspects of themselves that would otherwise remain private, causing more intimacy in relationships than would occur offline (Shaw & Gant, 2002). Furthermore, the flexibility of self-portrayal in the anonymous setting can enhance expression through feelings of increased autonomy (Allen et al., 2012; McKenna & Bargh, 2000; Shaw & Gant, 2002).

This study analyzes not only SNS use, but rather those users who specifically prefer expression mediated by anonymity online, including on SNS. We seek to clarify which characteristics comprise those SNS users who find it easier to express themselves anonymously. Variables central to past analysis of SNS users include gender, age, trust, strength of offline friendships, narcissism, and self-esteem, and identification with an online community. In this study, these variables will be revisited and applied specifically to those users who prefer anonymous expression in order to determine how each is correlated with such a user. Findings concerning those who prefer a sense of invisibility online through anonymity will then be compared to previous studies concerning general SNS user characteristics. This unique comparison allows for an analysis of how the subjective need for internet anonymity is related to various foundational user characteristics, allowing for a new discussion concerning who prefers anonymity online.

Internet anonymity as a multifaceted interactional phenomenon

Foundational to the attraction of SNS such as Facebook is the relational component, involving the fulfillment of the need for intimacy and belongingness with others (Davidson & Martelozzo, 2013). The stage provided by SNS allows for an abundance of relational possibilities through various avenues for personal expression (Merchant, 2012; Shaw & Gant, 2002). Furthermore, SNS involves a unique mode of interaction, as social presence is modified by technology through the removal of live face-to-face communication. Social presence here refers to the degree to which an interacting partner is experienced as a physical person (Rourke, Anderson, Garrison, 1999). Notably, communication online varies in its degree of social presence. These varying degrees are significant in the understanding of online interaction (Kreijns, Kirschner, Jochems, & Van Buuren, 2004). The modifier of social presence online is often referred to as anonymity, employed to varying degrees depending upon the form of interaction chosen by users.

Anonymity is a complex phenomenon that can be approached from many perspectives. Notably, no widely accepted model of online anonymity has thus far been developed, and as such, there exists room for interpretation in terms of its various forms. Though the intricacies of anonymity online are not central to this study, it is important to note the various modes within which young people operate online in order to better understand any preference for it. The tools provided by this setting act to enhance the motivation of young people toward relational aspects of the internet through a relative freedom from possible constraints, with the depth of interaction dependent upon user preference (Allen et al., 2012; Wolak, Mitchell, & Finkelhor, 2003).

Young people interact online through various forms of anonymity which affect socialization through levels of social invisibility (Keipi & Oksanen, 2014). Visual anonymity is the most common level experienced on a regular basis by users online, namely any situation where one's physical characteristics are hidden even in cases where participants are otherwise known to one another. Here, there is no direct visual feedback during interaction. Simply removing visual cues has a significant positive effect on self-disclosure (Joinson, 2001). Next, pseudonymity refers to interactions where usernames, avatars, or other social profiles are created by the user for a social purpose online. Lastly, full anonymity exists where interactions result in no reputation effects and where users remain unknowable after interaction has concluded. These experiences are text-based without any pseudonym or other long-term username (Joinson, 2001; Pfitzmann & Köhntopp, 2000). Together, various forms of anonymity are integral to the online experience, providing expressional benefit to young people seeking that possibility.

Age, gender, and relational dynamics online

Although SNS users are made up of all ages, adolescents, and young adults aged 16–30 years are the most active by a significant margin (Lehdonvirta & Räsänen, 2011). Furthermore, the need for validation and social support sought online is especially prominent in youth under the age of 16 (Livingstone, 2008). The relational potential provided by SNS to these young people of various ages brings with it a wider array of relatively specific identity or interest-based groups; like-minded users become more accessible online. Here, age is a significant factor in bond strength with an online group; young people identify with online communities more strongly than do older generations (Näsi et al., 2011). The level of independence and individuality of a young person is negatively related to strength of identification with an online community or peer group and as such, less autonomous young people are more likely to be strongly tied to online groups (Lehdonvirta & Räsänen, 2011).

Furthermore, in terms of gender, there exists little difference in quantity of time spent online, but the forms of activity taken have been found to vary between boys and girls under the age of 16 (Livingstone & Helsper, 2010). Furthermore, studies involving emerging adults aged 18–29 are in line with findings of younger adolescent groups in terms of gender-based differences in SNS use; on an interactional level, SNS tends to be used by females more for reinforcing existing relationships, while males are more likely to seek out new friends and romantically motivated attachments (Subrahmanyam, Reich, Waechter, & Guadalupe, 2008). Males tend to be more outwardly focused online, leading to a higher likelihood of encountering harmful material or individuals; as such, they are far more likely than females to take part in risky behaviors online (Livingstone & Helsper, 2010; Pompili et al., 2007). Males also spend significantly more time interacting through gaming online than do females (Jackson et al., 2009), perhaps reflecting a gender-based valuation of the gaming platform's facilitation of interaction. Notably, although the interactional forms and motivations behind social networking vary between genders, its role remains ever-present for both.

SNS open an avenue to young people for bonding social capital through providing social and emotional support by way of peer interaction (Chew, LaRose, Steinfield, & Velasquez, 2011). Central to the draw of online groups are the aspects of common interest or shared identity characteristics; there is a sharing of common objectives and behavioral norms (Ellison et al., 2007). Notably, age is a predictor of user naiveté online, where young people seeking identification and validation are more likely to display trust once relational intimacy is experienced (Livingstone, 2008; Merchant, 2012).

Social trust is often a precondition of whether or not one wishes to remain anonymous or reveal identity in the online settings. Young people who avoid face-to-face interaction are more likely to connect with others online, where some form of anonymity can be used to encourage self-disclosure (Merchant, 2012; Sheldon, 2008). This self-disclosure is a key component to relationship formation and maintenance as it fosters social trust both online and offline (Fogel & Nehmadb, 2009; Sheldon, 2009). Regardless of social disposition, investment in online networks facilitates interpersonal trust, which is ultimately necessary for online communities to survive (Sheldon, 2009; Valenzuela, Park, & Kee, 2009).

The mechanisms of online and offline peer group identification are not dissimilar; group membership and experience online and offline are equivalent in terms of relational tie strength (Lehdonvirta & Räsänen, 2011). Furthermore, both extroverts and introverts are found to transfer behavioral style from the offline setting to the online setting (Wilson, Fornasier, & White, 2010). However, those with fewer offline friends tend to have lower self-esteem and will spend more time on SNS in addition to displaying a more favorable attitude toward the benefits of its use (Mehdizadeh, 2010).

Narcissism and self-esteem online

The need for acceptance in a group or community is foundational in the motivation toward security and self-esteem, and as such, a core goal of many taking advantage of the benefits offered by SNS (Näsi et al., 2011). The online environment, notably, allows for self-presentation to be tailored to the desired audience through the choosing of what aspects of self will be hidden or displayed. The anonymous nature of online interaction allows for expression and experimentation with aspects of self that might otherwise remain hidden (Shaw & Gant, 2002). Central to the relational aspect of SNS is the ability for self-presentation in whichever ways one pleases; here, users are in control of their visible persona, or profile, with full control over whether it is accurate or idealized to artificially enhance one's appeal in the online arena (Anderson, Fagan, Woodnutt, & Chamorro-Premuzic, 2012).

SNS provide a platform for customized self-presentation for the purpose of both maintaining and developing relationships, which makes it attractive to those desiring a wider audience (Wang & Stefanone, 2013). This capacity to control one's self-presentation along with dynamics of the social arena brings into play the draw of those exhibiting traits of narcissism, characterized as the tendency to consider oneself more important than others, seeking admiration, and espousing a grandiose sense of self-importance or uniqueness with leanings toward self-promotion (Ackerman et al., 2011; Carpenter, 2012; Panek et al., 2013). Notably, there exists a strong positive relationship between the level of narcissism of a Facebook user and the amount of activity carried out on the site daily (Mehdizadeh, 2010). Narcissists also display a high level of extroversion and openness, using social relationships as a means of showing off (Carlson, Vazire, & Oltmanns, 2011; Wang & Stefanone, 2013), all effective methods toward gaining attention on SNS. In many ways, the stage provided by SNS is an ideal space for the fulfillment of these needs in individuals seeking such an outlet. Furthermore, narcissists are more likely to be open to new experiences due to low levels of anxiety and as such are enabled by a relative lack of social inhibition (Carpenter, 2012; Sedikides, Rudich, Gregg, Kumashiro, & Rusbult, 2004).

SNS users display higher levels of narcissism and exhibitionism than non-users, as the accessibility of such sites provides a new level of gratification of those needs; self-promotion, as a skill among users, becomes increasingly effective online (Ryan & Xenos, 2011; Wang & Stefanone, 2013). Notably, this manner of behavior that might be seen as an expression of confidence has been linked in fact to the opposite end of the spectrum by sharing behaviors with users displaying low self-esteem. This is in line with work that has shown that the unhealthy self-esteem of narcissism is not equivalent to a healthy high self-esteem, despite similarities in their favorable self-opinions (Baumeister et al., 2003; Bushman & Baumeister, 1998). According to Bushman and Baumeister (1998), ‘High self-esteem means thinking well of oneself, whereas narcissism involves passionately wanting to think well of oneself' (p. 228). Both those with a high level of narcissism and those with low self-esteem indiscriminately seek attention from a wide audience while those with higher self-esteem are primarily concerned with popularity within a smaller chosen relational spectrum (Anderson et al., 2012; Christofides, Muise, & Desmarais, 2009). Furthermore, those with high levels of narcissism and those with low levels of self-esteem are more likely to spend more than one hour per day on SNS such as Facebook (Mehdizadeh, 2010).

Here, those with low self-esteem use SNS information sharing as a method for seeking validation, or gaining acceptance from a particular group (Anderson et al., 2012). The need for belonging or acceptance is closely related to self-esteem, which can act as a measure of one's worthiness to joining a community online; low self-esteem can motivate one toward increased investment in projecting oneself in a light favored by the desired group in order to ensure a sense of security and belonging (Leary, 2007). As such, low self-esteem and a sense of grandiosity combined with the enhanced ability to manipulate and control interaction online can be a significant motivator for displaying an increasingly idealized self-representation through SNS in the hopes of attaining desired attention from other users.

Research questions, data, and methods

The above-discussed findings concerning the various components of SNS use and user characteristics combined with the dynamics of anonymity online provide a framework from which to continue toward the answering of the following research questions:

1. How do psychological characteristics associate with this preference for anonymity online?

2. How does respondents' socio-demographic background associate with the preference for anonymous expression in the online setting?

Given previous research of SNS users, we hypothesize similarities in the variables that will be positively related to preferring anonymous. First, we expect age to be positively correlated with anonymity preference as younger people are more likely to engage online. Furthermore, we expect a strong positive relationship between anonymity preference and displaying grandiosity. Lastly, we expect strength of online community to be positively related to preferring anonymous expression, as most online communities function with a degree of anonymity.

In terms of negative correlations, we hypothesize that those with strong offline friendships will be less likely to prefer online anonymity for self-expression. Furthermore, we expect those preferring anonymity online to be less trusting overall due to a desire to remain hidden, despite benefits of self-disclosure. Furthermore, linked to our hypothesis on the positive relationship between anonymity preference and grandiosity, we expect a negative relationship between anonymity preference and self-esteem. Finally, we expect gender to have at least a marginal association in preferring anonymous expression online to a more outward focus. On average, males are assumed to report stronger anonymity preference than females.

In order to fundamentally examine the relevant relationships between the variables of this study, we use two different datasets drawn from Finnish young people and emerging adults aged 15–30. Both of the data were collected using an identical Younet2013 Survey questionnaire in spring 2013. The data collection was managed with Limesurvey software. The surveys were optimized for both computers and mobile devices. The survey included socio-demographic variables and questions about online activity along with various online risks. We also asked participants about online and offline interactions, social trust, self-esteem, life satisfaction, and violent victimization. Respondents were given a chance to provide feedback concerning the survey to the research team.

1. Facebook data: Respondents for our first survey were recruited using three campaigns targeted at Finnish Facebook users aged 15–30 (n = 1013). The campaigns were launched between 10 April and 18 May 2013. We used four images and four short marketing texts in 15 different combinations to attract users to fill in the YouNet2013 Survey. The three campaigns reached between 432,649 and 528,261 adolescents and young adults, which is approximately half of the Finnish population aged 15–30 (Official Statistics of Finland, 2013). The campaigns received a total of 6074 clicks and generated 1337 survey responses in total. In this study, we excluded respondents who were over 30 years of age (n = 35) or under 15 (n = 121) or who did not answer all of the necessary questions (n = 168). Mean age of the 1013 respondents was 19.15 (SD = 3.90). 63.0% (n = 638) were female and 37.0% male (n = 375).

2. SSI data: Respondents to our second survey (n = 544) were recruited from a demographically balanced panel of Finnish respondents who voluntarily agreed to participate in research surveys. The panel is administered by Survey Sample International (SSI), and potential panel participants are recruited through random digit dialing, banner ads, and other permission-based techniques. E-mail invitations were sent to a sample of panel members stratified to mirror the Finnish population aged 15–30 on age, gender, and geographic region. Our sample quota was estimated to be nationally representative in terms of age and gender. Mean age of the 544 respondents was 22.62 (SD = 4.20); 48.8% (n = 271) were female and 50.2% (n = 273) were male.

We view our two-source population as a strengthening factor for this study, as it allows for us to show whether findings are consistent between samples. This research design was chosen due to difficulty in attracting young respondents in recent years. Furthermore, some of the traditional methods such as mail surveys are not functional, and some others such as phone surveys are less feasible. Online surveys are a way to avoid these problems. As young people are increasingly living their lives online, it follows that surveys designed to reach them can be effectively based there as well (e.g. Farrell & Petersen, 2010; Shropshire, Hawdon, & Witte, 2009). Descriptive statistics of all variables by sample are shown in Table 1.

Table 1. Descriptive statistics for dependent and independent variables.

	Facebook sample	SSI sample
	(n = 1013)	(n = 544)
	M (SD)	M (SD)
Dependent variable
Online anonymity preference	4.98 (3.08)	4.68 (2.83)
(1 = weak … 10 = strong)
Independent variables
	% (n)	% (n)
Gender
Male	37.0 (375)	50.20 (273)
Female	63.0 (638)	49.80 (271)
	M (SD)	M (SD)
Age	19.15 (3.90)	22.62 (4.20)
Trust	5.39 (1.65)	5.66 (1.58)
(1 = weak  …  10 = strong)
Offline interaction	5.29 (1.54)	4.87 (1.57)
(1 = never  …  7 = everyday)
Self-esteem	6.21 (2.67)	6.41 (2.41)
(1 = low  …  10 = high)
Grandiosity	5.90 (2.74)	5.84 (2.51)
(1 = no  …  10 = yes)
Online identification	2.88 (1.18)	2.74 (1.16)
(1 = no  …  5 = yes)

Note: Means and standard deviations (in parenthesis), percentages and number of cases (in parenthesis).

Online anonymity preference: Our dependent variable determines the extent to which participants prioritize anonymous private interactions online. The following question was posed to respondents: ‘I find it easier to talk about private things online when others don't know who I am'. Responses ranged from 1 (‘not very true of me’) to 10 (very true of me’). Mean was 4.98 (SD = 3.08) in the Facebook data and 4.68 in the SSI data.

Our independent measures include a total of seven variables: gender, age, trust, offline interaction, self-esteem, grandiosity, and online identification. Gender corresponds to respondent sex and age to respondent age in years. Trust was measured by using seven questions on participant trust toward others in general, namely their family, close friends, other acquaintances, work or school colleagues, neighbors, and people only met online. Different dimensions of interpersonal trust have previously been tested in various research settings (e.g. Hargittai, Fullerton, Menchen-Trevino, & Thomas, 2010; Henderson & Gilding, 2004).

In the questionnaire, these seven levels of trust were measured based on the following question:

Would you say that the following people can be trusted, or that you can't be too careful in dealing with these people? Assess on a scale from 1–10, where 1 = you can't be too careful and 10 = can be fully trusted.

The seven levels included: (1) ‘People in general', (2) ‘Family members', (3) ‘Good friends', (4) ‘Work colleagues', (5) ‘Other acquaintances', (6) ‘People met only online’, and (7) ‘People met on the streets and in the city'. Crombach's alpha for the seven trust items was .85 in the Facebook sample and .84 in the SSI sample. On the basis of high reliability, we created a new variable by summing up the original values and dividing them by the number of items. The means were 5.39 (1.65) for the Facebook sample and 5.66 (1.58) for the SSI sample.

Offline interaction was measured by using the question: ‘How often do you meet face to face with friends, relatives, or work colleagues for social reasons?' Scale varied from 1 (never) to 7 (every day). Mean was 5.29 (SD = 1.54) in the FB data and 4.87 (SD = 1.57) in the SSI sample. In preceding studies, offline interaction with friends has been proven to be predictor of online behavior (e.g. Ellison et al., 2007; Oksanen & Keipi, 2013).

Self-esteem was assessed by using the Single-Item Self-Esteem measure that has been found to be interchangeable with longer self-esteem measures such as the Rosenberg Self-Esteem Scale (Robins, et al., 2001). The responses to ‘I have high self-esteem' question ranged from 1 (‘not very true of me’) to 10 (‘very true of me’). Mean was 6.21 (2.67) in the Facebook data and 6.41 (2.41) in the SSI data.

Grandiosity was asked with a question modified from the 16-item Narcissistic Personality Index (NPI-16) (Ames, Rose, & Anderson, 2006). Respondents were asked to rate themselves from 1 (‘not very true of me’) to 10 (‘very true of me’) according to the following statement: ‘I think I'm a very special or extraordinary person’. The mean was 5.90 (2.74) in the Facebook data and 5.84 (2.51) in the SSI data. We recognize that grandiosity represents a sub-type of narcissistic personality. Grandiose personality is connected with overt self-enhancements, arrogance, denial of weaknesses, and generally unrealistic view of self, which also manifest in interpersonal problems with others (Dickinson & Pincus, 2003).

Online identification was asked as part of an 11-item set. Respondents were asked how strongly they felt that they were part of various groups: ‘How close do you feel to each of the following?' We used the question concerning online communities. The items have been previously tested and successfully used in research concerning online and offline activities (Lehdonvirta & Räsänen, 2011; Näsi et al., 2011). The scale ranged from 1 (‘not at all’) to 5 (‘very much’) and means were 2.88 (1.18) for the Facebook sample and 2.74 (1.16) for the SSI sample.

Our analysis techniques consist of descriptive statistics and ordinary least squared (OLS) linear regression. Descriptive techniques are first used to offer an overview of the profiles of online anonymity preference between the samples. Thereafter, regression models are used for a more specific evaluation of the effect of the selected background variables. In the analysis, we estimate the mean scores of ‘online anonymity preference' from a set of independent variables. Focusing on the two separate samples, the linear connection between online anonymity preference and independent variables are tested and the strength of the connections is also estimated. The coefficients of the regression equations are presented in both non-standardized (B) and standardized (β) form. We also report t-values, standard errors for each estimated parameter, and the variances accounted for the final models (adjusted R²).

Results

With regard to descriptive statistics given in Table 1, we note that preference for online anonymity is slightly stronger in the Facebook sample when compared to the SSI sample. We also note that the distributions of independent variables also differ, particularly in terms of age and gender. The proportion of over 20-year-old respondents is somewhat underrepresented in the Facebook sample when compared to the SSI sample (M_FB = 19.15; M_SSI = 22.26). Similarly, young males are underrepresented in the Facebook sample (37% male), whereas SSI sample appears to be gender balanced (50% male). Otherwise, there are no notable differences in the distributions of independent variables between the samples.

Given that online anonymity preference displays approximately normal distribution in the samples, multiple OLS regression analysis can be deemed an appropriate method of analysis. We treat the samples separately, because the age and gender structure of the target populations are different from each other. Therefore, the next phase in the analysis was to predict the level of online anonymity preferences using the selected independent variables.

In Table 2, OLS regression model is employed on the Facebook sample. Since we are interested in the linear effects of the independent variables, we entered all predictors and control variables into the model at once. Based on the preliminary bivariate analysis (not shown in the table), all independent variables were significant predictors of online anonymity preferences (at the level of p < .05). As the table shows, each independent variable remains significant in the full model. Despite this, the model explains variance in anonymity preference only moderately (R² = .088, F(7 991) = 14.74, p < .001).

Table 2. Predicting online anonymity preference among 15–30-year-olds, OLS linear regression (Facebook sample).

	B	S.E.	β	t	p
Female	−.44	.20	−.07	−2.20	.028
Age	−.07	.03	−.09	−2.85	.004
Trust	.14	.06	.08	2.19	.029
Offline interaction	−.25	.06	−.12	−3.88	< .001
Self-esteem	−.17	.04	−.15	−4.11	< .001
Grandiosity	.13	.04	.12	3.41	.001
Online identification	.54	.08	.21	6.55	< .001
Constant	6.37	.86		7.40	< .001
Adj. R^2	.088

Note: Unstandardized (B) and standardized regression coefficients (β), standard errors, t-value, and statistical significances.

As for individual predictors, we may note that offline interaction (β = −.12), self-esteem (β = −.20), online identification (β = −.32), and grandiosity (β = −.23) have the strongest effects on anonymity preference (at p < .001 level). As hypothesized, the impact of offline interaction is negative; those meeting friends offline regularly report lower anonymity preference compared to those meeting friends less often. Also, self-esteem has negative impact, indicating that respondents with higher self-esteem report lower anonymity preference. On the other hand, the effects of online identification and grandiosity are positive. In other words, those respondents who identify strongly with an online social group and score high on grandiosity also report stronger online anonymity preference than other respondents.

In terms of socio-demographic controls and other variables, both age (β = −.09) and gender (β = −.07) as well as trust (β = .08) are significantly associated with online anonymity preference (at p < .05 level). Age is negatively associated with online anonymity preference, showing that younger respondents on average prefer anonymity more than older ones. The effect of trust is positive, which can be considered as an interesting finding. Since our measure of trust includes many correlated dimensions of interpersonal trust, it seems that those more willing to rely on others both online and offline are also most freely self-expressive while unidentifiable. This finding points to a new disconnect between anonymity preference and inherent fear of others, as perhaps anonymity's role in allowing one to remain hidden is not actually motivated by social fear, but by some other characteristic. Or, on the other hand, perhaps anonymity is preferred simply because is it inherently tied to one's favorite social networking provider. Males report slightly stronger anonymity preference than do females. The effects of trust and gender are relatively weak, however.

Overall, these findings support many of our assumptions that preference for online anonymity varies according to many individual characteristics. But, is this interpretation also valid when examining our second sample?

 Table 3 shows similar multiple regression model for the SSI sample. Basically, the findings are very similar to what we observed in the Facebook data. All variables are significant with an exception for gender, however. In other words, males do not report higher anonymity preference in this sample. We may also note that most of the effects are stronger here, although the interpretation of the associations must be identical. Age (β = −.09), offline interaction (β = −.12), and self-esteem (β = −.20) are associated negatively with online anonymity preference, whereas trust (β = .19), online identification (β = .32), and grandiosity (β = .23) have positive effects. The full model accounts for a considerable share in anonymity preference variation (R² = .227, F(7 535) = 23.44, p < .001).

Table 3. Predicting online anonymity preference among 15–30-year-olds, OLS linear regression (SSI sample).

	B	S.E.	β	t	p
Female	ns	ns	ns	ns	ns
Age	−.06	.03	−.09	−2.16	.031
Trust	.34	.08	.19	4.24	<.001
Offline interaction	−.22	.07	−.12	−3.05	.002
Self-esteem	−.24	.06	−.20	−4.26	<.001
Grandiosity	.26	.05	.23	5.26	<.001
Online identification	.78	.10	.32	7.78	<.001
Constant	3.45	1.00		3.44	.001
Adj. R^2	.227

Note: Unstandardized (B) and standardized regression coefficients (β), standard errors, t-value, and statistical significances.

Altogether, it is reasonable to conclude that the predictors selected are feasible for explaining the level of anonymity preference across the data. The results thus suggest certain interpretations related to the psychological as well as socio-demographic constraints in the functional internet anonymity.

Discussion

The role of various levels of anonymity in the relational sphere provided by computer-mediated community including SNS continues to affect the way users interact on a global scale; yet, few studies have delved into who specifically seeks out the online experience for the benefits of anonymity and which factors can actually explain the anonymity preference. More often than not, users interact in a setting where some level of anonymity persists, as its effects exist on a scale that various levels of transparency do not necessarily eliminate.

Our study provides a missing link in this understanding of online users who favor the internet's anonymous aspects through a survey of Finnish young people aged 15–30 years. We explored a number of descriptive characteristics, determining the relationship of these variables with those respondents who favored anonymous expression online over a non-anonymous experience. Through the survey, each respondent assessed himself/herself along criteria toward determining whether each of the presented characteristics was relevant to their experience and motivations. The findings of this study were in line with previous studies while also providing new dimensions of comparison between users and their anonymity preference. Here, a new comparison is made possible between those most active in SNS use and those SNS users specifically valuing anonymity. Furthermore, by using two datasets, we were able to confirm corresponding findings between samples. Measures of both were in agreement in terms of the positive relationship between preference for anonymous online expression and individual characteristics displayed.

Respondents reporting stronger anonymity preference were responding to whether they find it easier to express themselves in interaction when they cannot be identified online. Thus, this is a preference for a level of either pseudonymity where interaction is carried out using a relationally instrumental created identity, or full anonymity where no reputation effects would result. The benefits of such a state of anonymous interaction acknowledged by respondents are in line with the findings of anonymity as allowing for enhanced expression of self in ways that might otherwise remain hidden (Shaw & Gant, 2002). This anonymity brings with it a sense of control over what is shared and with whom (Anderson et al., 2012).

Users preferring anonymity showed a positive correlation with espousing a sense of grandiosity linked to narcissism along with having lower self-esteem, a confirmation of our hypothesis. This finding is in line with studies where those displaying narcissism and low self-esteem were more likely to seek attention through platforms offering a wider audience, such as SNS (Anderson et al., 2012; Christofides et al., 2009; Mehdizadeh, 2010). Here, those displaying narcissism and low self-esteem are positively correlated with the desire for expression in a social space relatively free from social constraints; as such, anonymity allows for custom self-representation toward compensating for low self-esteem. We put forth that anonymity preference is used as an expressive tool toward building self-esteem rather than as a retreat from social contact.

In addition to low self-esteem, respondents preferring anonymity tended to have fewer offline interactions with friends. This finding is a confirmation of both our hypothesis and previous work where those with fewer offline friends were found to have lower self-esteem combined with a more favorable attitude toward the benefits of online use along with the tendency to spend more time there (Mehdizadeh, 2010). This shows that the online space can be seen as a place to compensate for offline relationships among those seeking attention of some kind, again linked to our finding on self-esteem.

Respondents preferring anonymity also correlated positively with valuing online communities strongly, a confirmation of our hypothesis. As these users also displayed low self-esteem, this reinforces previous studies where low self-esteem and information-sharing were shown to act as a method for gaining validation from an online community (Anderson et al., 2012). Linked to this valuing of online community and low self-esteem is again the aspect of anonymity, which here can motivate users otherwise complacent to increase efforts in online self-presentation (here, without identification) in order to gain a sense of belonging online through the pseudonymous state (Leary, 2007). Notably, online communities are based in differing levels of anonymity (Keipi & Oksanen, 2014) and as such do not necessarily limit depth of social ties to any significant degree.

Our hypothesis concerning age was also confirmed, in that those desiring to express themselves anonymously tended to be younger; this finding is in line with previous studies where age has been found to be the most significant factor in social tie strength among online communities, especially when combined with the finding that these users also hold a high valuation of these online communities (Näsi et al., 2011). This finding also confirms previous studies showing that the level of independence of a young person is negatively related to strength of identification with an online peer group (Lehdonvirta & Räsänen, 2011). An additional hypothesis was partly confirmed, in that males expressed stronger anonymity preference than females did. However, the effect was generally weak and statistically significant only in the Facebook sample.

We may only speculate here as to why gender showed no effect in the SSI sample while associations with other independent variables were slightly stronger. As recently discussed in literature, different online techniques may attract different-minded respondents (Weinberg, Freese, & McElhattan, 2014). These differences could, of course, result from a number of factors. In our case, the Facebook recruitment attracted younger respondents than did the SSI recruitment. As such, the findings likely reflect the fact that younger people are generally more familiar with social media. Gender may also play a larger role for sensitive issues at a younger age. As already acknowledged, the SSI sample is well balanced in terms of gender, while females are somewhat overrepresented in the Facebook sample.

In both samples, respondents preferring anonymity were also more likely to display stronger trust online, a finding contrary to our hypothesis as it was assumed that desiring anonymity would be associated with a desire to remain relatively hidden while instrumentally present in order to minimize social risk. When combined with the finding of these respondents also being younger, results are in line with studies where age was found to be a predictor of user naiveté online (Merchant, 2012). Furthermore, when considering our finding concerning the valuation of online community, many requiring some degree of anonymity, a further confirmation can be put forth; in the online space where users seek identification and validation through community, trust is more readily displayed by young people as relational intimacy is experienced (Livingstone, 2008).

Our findings show users who prefer to express themselves in an anonymous manner as displaying a greater sense of grandiosity, lower self-esteem, younger age, more trust, fewer offline friends, and a high valuation of online communities. The effect of gender remained weak. The interaction between these characteristics within users is likely dynamic. A sense of grandiosity drives the need for an audience while low self-esteem brings a need for validation. Online community membership is seen as a valuable addition to one's social well-being, with youthful trust being displayed in the hope of connection online.

However, despite the confirmation of previous work, our study contains a set of limitations. Although our sample size is relatively substantial when compared to previous studies, online surveys cannot be said to be nationally representative. Furthermore, we cannot be certain that all respondents understood the issue of anonymity in the same light; as it was not defined in the survey to the extent to which it is here, there is a possibility that a survey with more strict definitions would yield more accurate results. Despite this, our findings based on linear regressions seem reliable when compared to previous research. Furthermore, respondents assessed their preference for online anonymity using only one question; a set of questions dealing exclusively with anonymity may have provided a deeper understanding of user preference for anonymity and understanding of its benefits to them.

Our study represents a baseline on which to build, with next steps delving into user motivation for the anonymity preference, for example. Notably, in building the case of who prefers anonymity, we feel one question concerning anonymity is adequate for both identifying those users and determining user characteristics that are related to that preference. Definitive answers to questions of why those preferring anonymous expression display the characteristics revealed by the study are beyond the scope of the data. This illuminates necessary steps for future research, namely a deeper look into how young people understand and experience the various forms of anonymity and their effects, combined with explanatory analysis of why certain characteristics such as grandiosity are so tied to low self-esteem and anonymous expression. Though the dynamic and evolving nature of online interaction may defy strict definitions, a deeper understanding of the anonymous aspect of the internet and its effect of young people's development is needed. Finally, we encourage a cross-cultural comparison of these findings, as those of this study are limited to the relatively homogenous population of Finland.

Despite these limitations, our study provides an important bridge between various studies concerning characteristics of SNS users along with a connection of those users to the valuation of anonymity. It does this with the new approach of studying a certain type of internet user where all of these characteristics meet and interact. As the preference for anonymity in expression is the key identifier of this type of user, we believe that our findings provide an important stepping stone toward understanding the role of anonymity in societal dynamics, especially online. This study adds to the growing body of work that seeks a deeper understanding of young internet users and the dynamics that occur online.

Conclusions

The role of the internet has been foundational to the methods with which interaction is carried out today. Online communication through social networking and online communities has allowed for a repackaging of how users can present and express themselves to one another. Notably, some level of anonymity is often part of this experience. Although the various levels of user anonymity are often taken for granted in the online setting, their effects must be accounted for, especially in users who express a preference for interacting under their influence. Our results based on a sample of Finnish internet users show various confirmations of previous work concerning SNS use and motivation while combining many of those factors in individuals who prefer the anonymous experience for expression. The dynamics discussed affect all young internet users to some degree, whether directly or indirectly, and as such have become a part of the social framework existing online.

The findings of this study concerning these various dynamics show that they overlap to a great degree, influencing one another to the benefit or detriment of the user. A deeper understanding of users' offline experiences and background would assist in understanding of why anonymity is desired, as that is a question that an identification of user characteristics alone cannot answer. As such, it is critical to find new ways to study how anonymity is utilized both online and offline and why those needs persist in a portion of the young population toward a determination of how social health is helped or hindered by the internet's provision of the anonymous outlet.

In this quantitative study, a profile of who prefers anonymity online for self-expression was identified, a novel and important finding that provides a foundation for new research; ascertaining user motivations for this preference will provide a deeper understanding of these users, yet remains outside the scope of this study. Here, we offer a strong basis for further research, where questions of why anonymity is preferred by certain segments of young internet users might be revealed through qualitative methods, for example, increasing the impact of the research findings.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Teo Keipi, M.Sc., is currently a Ph.D. student at Turku University. His doctoral dissertation on economic sociology analyzes anonymity in online settings. [email: [email protected]]

Atte Oksanen, Dr. Soc. Sci., is an associate professor of social psychology at the University of Tampere, Finland. Dr Oksanen's research focuses on mass violence, emerging technologies, and youth. He has published in a variety of areas including youth studies, drugs and alcohol use, and cultural studies. [email: [email protected]]

Pekka Räsänen, Dr. Soc. Sci., is a professor of economic sociology at the University of Turku, Finland. He has studied a variety of topics connecting with mass violence, culture, and consumer behavior. His research focuses on the ways that contemporary social life is affected by the new information and communication technologies. [email: [email protected]]

Additional information

Funding

This research was funded by the Alfred Kordelin Foundation [funding decision 2012, Keipi Dr. dissertation grant] and the Kone Foundation [project funding decisions 2012–2014, Oksanen & Räsänen, Hate Communities Project].