Adaption and psychometric evaluation of the Presentation of Online Self Scale in adults

Introduction: The Presentation of Online Self Scale (POSS) is an established psychometric tool used to measure how an individual presents oneself in online environments. The POSS was originally used with an adolescent sample. This study aimed to develop a factor structure of the POSS for use with an adult sample as well as examine the reliability and convergent validity of the tool. Methods: A quantitative design was applied. Participants (n 1⁄4 531) aged 18–74 (M 1⁄4 35, SD 1⁄4 12.09) completed the POSS and the Ten-Item Personality Index (TIPI) that assessed personality traits. Results: Exploratory factor analysis revealed a three-factor solution of the POSS with an adult sample. The three factors explained 55% of the variance in POSS scores. All items had loadings ranging from 0.30 to 0.84. The new structure, renamed the Presentation of Online Self Scale for Adults (POSSA), showed satisfactory reliability and convergent validity. Conclusions: The findings of this study suggest that the three factor version of the POSS is a valid and reliable tool for measuring online self-presentation in adults. Self-presentation evolves through developmental stages, thus presenting the need for age appropriate psychometric instruments to fit those stages.


Introduction
People frequently make conscious and deliberate decisions about how to present themselves to others (Baraket-Bojemal, Moran, & Sahar, 2016;Goffman, 1959). Goffman's (1959) dramaturgical approach likens self-presentation to that of an actor playing a role, and requesting that those observing the role-playing take seriously what is being put forward. In terms of everyday self-presentation, this means that when one presents oneself to others there is specific intent behind the formation of an impression. Furthermore, that impression is expected to be believed and viewed as true. However, online environments allow for a new level of flexibility in self-presentation. With the growth in use of Social Networking Sites (SNSs), researchers have become increasingly interested in the similarities and differences between offline and online self-presentation. Some research argues that individuals carefully manage their online self-presentation (Grieve, March, & Watkinson, 2020;Okdie, Guadagno, Bernieri, Geers, & Mclarney-Vesotski, 2011;Tosun, 2012;Yu & Kim, 2020). These self-presentation decisions can be facilitated with greater ease through the technological advancements of the online world and social-networking sites (SNSs) (Nadkarni & Hoffman, 2012). For example, in a study examining false self-presentation on Facebook, it was found that emerging adults (18-25 years of age) will use specific strategies when presenting the self online (Zhao, Grasmuck, & Martin, 2008). However, while individuals may omit or exaggerate certain aspects of the self online, other research has found that individuals are generally consistent offline to online with self-presentation (Back et al., 2010;Fullwood, James, & Chen-Wilson, 2016;Gosling, Augustine, Vazire, Holtzman, & Gaddis, 2011;Marriott & Buchanan, 2014;Strimbu & O'Connell, 2019;Wilson, Gosling, S. D., & Graham, L. T., 2012). This presents two arguments about online self-presentation. One is that online self-presentation is simply an extension of the self. The other argues that online self-presentation is an entirely different self.
These competing arguments pose a question of the degree to which personality plays a part in online self-presentation. Previous research has investigated whether personality traits like neuroticism and extroversion are associated with the presentation of an authentic self online. It has been found that individuals low in extroversion are more likely to engage in self-exploratory behaviors online, whereas individuals high in extroversion are less more likely to show consistency between offline and online presentation (Michikyan, Subrahmanyam, & Dennis, 2014). One interpretation of this is that extroverted individuals tend to view SNS's as more of an extension of their offline world while more introverted individuals who view SNS's as an additional/alternate world to express oneself (Amichai-Hamburger & Vinitzky, 2010).
However, a different study assessing personality and the "true self" online found online self-presentation to be more closely related to personality expression preference than an individual's actual personality (Marriott and Buchanan, 2014). It was found that introverted individuals and those with higher levels of neuroticism prefer to express their "true self" online instead of in face-to-face interactions. This, however, was very heavily influenced by an individual's level of shyness. Individuals who scored low in Conscientiousness and Agreeableness also preferred to express their "true self" online. Where this occurred, the preference for presenting the "true self" online over face-to-face indicates inconsistency between offline and online self-presentation, but suggests that the authentic self is actually expressed online instead of face-to-face. This study also explored how both online friend and offline friends view the individual's online self-presentation. It was found that offline friends had significantly closer ratings to those of the participant than did online friends, suggesting that overall, participants were not representing themselves more authentically online compared to offline.
To date, research has focused primarily on the online selfpresentation of adolescents (Bobkowski, Schafer, & Ortiz, 2016;Khan, Gagne, Yang, & Shapka, 2016), as they typically hold a less stable self-concept than adults (Strasburger, Wilson, & Jordan, 2014.) However, it has been suggested that the years of emerging adulthood offer the greatest opportunity for identity experimentation with regards to love, work, and worldviews (Arnett, 2000). Identity development is particularly pronounced as emerging adults transition to greater autonomy in education or work, where they are keen to explore their new adult identity (Kroger, Martinussen, & Marcia, 2010); they can try to project a more socially attractive self online (Yang & Brown, 2016). However, as adults develop, their identity becomes more congruent, and the gap between the ideal and actual self, i.e. the level of self-discrepancy (Higgins, 1987), becomes less pronounced (Bybee, Piastunovich, & Glick, 1995;Heidrich, 1999). The self-assurance of adults with regards to their identity is evident in online contexts. In a systematic review of 21 studies (N ¼ 7573), findings of authentic/inauthentic online self-presentation were most generalizable to adolescents and emerging adults, less so to adults aged over 25 (Twomey & O' Reilly, 2017). Other studies have indicated that younger adults engage in false self-presentation more than older adults (Wright, White, & Obst, 2017).
It is clear from the research that online presentation is an important task for individuals, featuring in both broad developmental trends as well as individual differences. Understanding and measuring people's online behavior, and specifically their preferences in self-presentation, are important goals. The increasing ubiquity of digital applications, made possible by the steady growth of computing power, has accelerated the integration of people's online and offline worlds. We have entered a time where a majority of SNS users are what are referred to as digital nativesthose who have grown up using digital technology. Society is keeping in touch online, dating online, and holding work meetings and conferences online. All of these interactions are happening in environments where one has to present themselves online. A widely used online selfpresentation measure in cyberpsychology research is the Presentation of Online Self Scale (POSS). This scale was developed by Fullwood et al. (2016) to capture online self-presentation in adolescents. It included four factors of online self-presentation: Multiple Selves, Ideal Self, Consistent Self, and Online Presentation Preference. The Multiple Selves factor captures the presentation of different personas. This can be expressed by acting out different personas across various online platforms or even creating an entirely new online identity. The Ideal Self factor, while similar to Multiple Selves in its presentation of a different version of the self, encompasses the important element of presenting the self that one desires to bethe ideal version of oneself. The Consistent Self factor can be broadly defined as presenting an authentic self. It is situated in opposition to Multiple Selves and the Ideal Self since it expresses the position that one's online character is no different to one's offline character. Finally, the Online Presentation Preference Factor captures the level of preference one has to present the self online over presenting the self offline. In other words, it reflects the comfort one feels showing who they are via online spaces vs. face-to-face. These four factors were found to be a reliable method for describing the adolescent population's online self-presentation. However, when used with an adult sample (aged 18þ) the four factors did not maintain their original structure (Strimbu & O'Connell, 2019). More specifically, the Ideal Self factor was found to be the most unstable with the redistribution of its questions to the other factors. This study aimed to evaluate and validate a version of the POSS measure for future use with an adult sample. Moreover, because the factor structure of the previously examined tool in adults (Strimbu & O'Connell, 2019) was established by using principal component analysis (PCA), the present study aimed to examine the tool by applying factor analysis. PCA is a data reduction method, while factor analysis is a method used to capture and reflect the structure of latent factors underlying data (Ford, MacCallum, & Tait, 1986).

Participants
Five hundred thirty-one participants (377 female; 127 male; 10 nonbinary; 6 transgender female; 4 gender non-conforming; 3 transgender male; 3 other, 1 prefer not to disclose), recruited through social media outreach and online advertisement, completed the POSS. The mean age was 35 (SD ¼ 12.09), ranging from 18 to 74 years. Participants resided across sixty-two countries with majority citizenship reported as the United States of America and/or Ireland (70.5%). Finally, a majority of participants reported as having completed an undergraduate degree or higher (68%). To examine and establish the factor structure of the POSS, the sample was randomly split into two sub-samples. Data from subsample 1 (n ¼ 265) were used to examine the factor structure of the tool by applying Exploratory Factor Analysis (EFA), while data from subsample 2 (n ¼ 266) were used to establish the factor structure by applying Confirmatory Factor Analysis (CFA). Demographic characteristics for the two sub-samples are presented in Table 1.

Procedure
Participants were provided a link or QR code to the online survey (containing the POSS questionnaire) hosted by Qualtrics, a secure online survey platform. The survey could be completed via computer, tablet, or mobile phone. Participants were prompted to complete a Captcha, a response test to verify the user is human. Participants were fully informed with regard to the aims of the study and informed consent was obtained. Demographic information and various measures, including the POSS questionnaire were then completed.

Measures
The Presentation of Online Self Scale (POSS) (Fullwood et al., 2016) is a self-report measure with 21 items divided into four factors: Item 1-9: Ideal Self (IS), Item 10-14: Multiple Selves (MS), Item 15-18: Consistent Self (CS), and Item 19-21: Online Presentation Preference (OPP). Each factor is scored on a 5-point Likert scale from "strongly disagree" to "strongly agree." A higher score represents a higher presentation of each factor. For example, a higher IS score indicates that an individual presents more of an ideal self online when compared to their offline self-presentation. A higher MS score indicates higher online presentation of multiple selves, while a higher CS score indicates higher consistency between one's offline and online self-presentation. Finally, a higher OPP score suggests a stronger preference for online self-presentation over offline self-presentation.
The Ten Item Personality Measure (TIPI) (Gosling, Rentfrow, & Swann, 2003) is a 10-item measure that examines five personality traits: Extroversion, Agreeableness, Conscientiousness, Emotional Stability, and Openness to Experiences. It is a shortened version of the well-known Big Five Personality Measure (Goldberg, 1992). Each personality factor is addressed through an "I see myself as" statement and is scored on a 7-point Likert scale from 1 ¼ "strongly disagree" to 7 ¼ "strongly agree." A total mean score for each factor is then calculated.

Data analysis overview
To explore the tool's factor structure, we subjected data from subsample 1 to EFA using Principal Axis Factoring (PAF) as an extraction method with oblique rotation thus allowing the factors to be correlated. Parallel analysis, scree plots, eigenvalues >1 and item loadings >0.30 were used to determine the number of factors to be retained. Using data from sub-sample, CFA was applied to confirm whether the factor structure emerged from the EFA fits data adequately. The following indices were used to determine the data-model fitting adequacy: Chi-square test (χ 2 ), Tucker-Lewis Index (TLI), Comparative Fit Index (CFI), Root-Mean-Square Error of Approximation (RMSEA) and Standardized-Root-Mean-Square Residual (SRMR). A non-significant χ 2 suggests that there is no significant discrepancy between the observed and estimated covariance matrix. However, χ 2 can be sensitive, thus a ratio of Degrees of Freedom (df)/chi-square value < 2.5 has been suggested as an additional indication of a good model-fit (Kline, 2011). TLI and CFI range from 0 to 1 with values > 0.90 and >0.95 showing good and excellent fit respectively; RMSEA values < 0.08 and SRMR values < 0.08 indicate that the model fits the data adequately (Hu & Bentler, 1999).
Using the whole sample (N ¼ 531) convergent validity was tested by applying Pearson product moment correlations between the POSS factors and the personality factors "Extroversion" and Conscientiousness" of the TIPI. These were the two personality factors selected as the link between them and online self-presentation has been widely reported and replicated in the relevant literature. Previous research found that people scoring higher on conscientiousness are more "cautious in their online self-presentation" (Seidman, 2013), suggesting that they would also be less likely to explore multiple selves in an online context. Furthermore, individuals who are more extroverted use online settings as an extension to their offline self (Michikyan et al., 2014;Tosun & Lajunen, 2010) and present an authentic self online (Marriott & Buchanan, 2014;Seidman, 2013). Therefore, extroverted individuals would show more consistency between their offline and online self compared to introverted individuals. Thus, significant positive correlations were hypothesized between a factor describing consistency of self from offline to online and the TIPI factors "Extroversion" and "Conscientiousness." By contrast, it was hypothesized that a factor encompassing discrepancies from offline to online self-presentation would result in a significant negative correlation to the two TIPI factors.
EFA and correlations were conducted using IBM SPSS software version 24, while CFA was conducted using AMOS software version 24.
While not a primary element of this research, due to ethics recommendations on gender reporting, gender differences were examined. Only 5% of the sample identified as a gender other than male or female, therefore an independent-samples t-test was conducted to compare the new POSS factor structure.

Exploratory Factor Analysis
Using data from the sub-sample 1, the 21 POSS items were subjected to a series of EFA in order to determine the most optimal factor structure. The initial 21 POSS items are provided in Appendix 1. The most optimal factor structure was achieved after removing three items in total. These items were deemed problematic because they either did not load on any of the emerged factors or had high cross-loadings on more than one factor (>0.35). Overall results of the EFA in conjunction with the scree plot and the outcomes of the Parallel Analysis showed that a three-factor solution was the most optimal. The three factors reflect constructs of an Adaptable Self, Authentic Self, and Freedom of Self Online, respectively. This threefactor solution including eighteen out of the 21 items of the original scale explains 55% of the variance in scores of the POSS. Factor loadings, eigenvalues item means, SDs and Cronbach's alphas are presented in Table 2. As Table 2 shows, all items had loadings ranging from 0.30 to 0.84. The internal consistency of the three factors was satisfactory for sub-sample 1 with corrected-item total correlations ranging from 0.37 to 0.59 and 0.30-0.73.

Confirmatory Factor Analysis
Using data from sub-sample 2 (n ¼ 266), CFA was applied to examine the three-factor proposed model fitting adequacy. Overall CFA results revealed that the proposed three-factor model had a very good fitting adequacy to data. However, because item 15 (see Appendix A) demonstrated a quite low loading (β ¼ 0.15, p ¼ .02) on Factor 2: Authentic Self, we removed the item and re-ran the CFA. This resulted in an improved fitting model. Table 3 presents the fit indices before and after removing item 15 from the CFA. Reliability coefficients indicated satisfactory internal consistency for the Authentic Self factor both in sub-samples 1 and 2 (α ¼ 0.74 and α ¼ 0.71 respectively) after removing item 15. Table 4 presents the correlation matrix for the associations between the three-factor POSS and TIPI (Extroversion and Conscientiousness subscales). Correlations revealed that the POSS factors have good convergent validity. Confirming our hypothesis, the POSS Factor 1: Adaptable Self and Factor 3: Freedom of Self Online showed a significant negative relationships with both Extroversion and Conscientiousness scores indicating that more extroverted and conscientious individuals show less discrepancies between their offline and online self- presentation. Furthermore, also confirming our hypothesis, the POSS Factor 2: Authentic Self showed a significant positive relationship with both Extroversion and Conscientiousness scores indicating more consistency between offline and online self-presentation for more extroverted and conscientious individuals.

Discussion
The EFA and CFA findings support the three-factor structure of the POSS for an adult sample, referred to as the Presentation of Online Self Scale for Adults (POSSA) from here on. All items had high loadings under each factor indicating three elements of online self-presentation. The significant correlations between the POSSA factors and TIPI scores, that is, online self-presentation and personality factors (extroversion and conscientiousness), established convergent validity of the POSSA three factor solution. The factors showed satisfactory internal consistency in the present sample and the overall findings indicate that the POSSA is a valid and reliable tool for measuring online self-presentation in adults.

The POSSA
The three POSSA factors were closely related to the original POSS factors: "Multiple Selves," "Consistent Self," and "Online Presentation Preference," with additional items from the original "Ideal Self" factor dispersed among them. This was consistent with previous findings from Strimbu and O'Connell (2019), who found that the "Ideal Self" factor in the POSS did not maintain structure with an adult sample. It also supports the findings that show while SNS's allow users to promote a version of their 'ideal self' online (Manago, Graham, Greenfield, & Salimkham, 2008;Michikyan et al., 2014) and to present multiple selves online (Lin & Wang, 2014), adults actually present a more authentic self in online settings (Twomey & O'Reilly, 2017;Wright et al., 2017) in comparison to adolescents.
The original POSS "Ideal Self" items reloaded on three factors now labeled: Adaptable Self, Authentic Self, and Freedom of Self Online (see Appendix B). The 'Adaptable Self' factor contains all of the original items from the 'Multiple Selves' factor (Fullwood et al., 2016), with the addition of item 8 ("I can escape from myself online") from the 'Ideal Self' factor (Fullwood et al., 2016). The items of the new 'Adaptable Self' factor suggest that the adult SNS user acts out multiple personas online, where they can escape from themselves and adapt their identity to one which is more desirable.
The 'Authentic Self' factor contains 3 of the original 4 items from the 'Consistent Self' factor (Fullwood et al., 2016), with the removal of item 15 ("I feel my personality online is the real me"). The 'Authentic Self' factor also includes item 4 ("The way I present myself online differs significantly from real life"), from the 'Ideal Self' factor (Fullwood et al., Table 2 Exploratory Factor Analysis of the POSS using data from sub-sample 1 (n ¼ 265).   Note: M is average mean; SD is standard deviation; *p < .05, **p < .01; POSS ¼ Presentation of Online Self Scale; FOS ¼ Freedom of Self; TIPI ¼ Ten Item Personality Measure (Ex ¼ Extroversion, C ¼ Conscientiousness).
N. Strimbu et al. Computers in Human Behavior Reports 3 (2021) 100073  2016). The items of the new 'Authentic Self' factor refer to the adult SNS user's authenticity in the online world in comparison to their identity in the offline world. The third factor of the new POSSA, 'Freedom of Self Online', comprises the original items from the 'Online Presentation Preference' factor and four items from the 'Ideal Self' factor (Fullwood et al., 2016). The 'Freedom of Self Online' factor includes items 2 ("I can show my best qualities online"), 5 ("Communicating online allows me to say the things I can't say offline"), 7 ("I feel I can be my Ideal Self online"), and 9 ("I can talk to people who wouldn't usually talk to me in the real world"), from the 'Ideal Self' factor (Fullwood et al., 2016). The new factor reflects the degree to which the SNS user believes in the merits of online self-presentation, particularly the freedom to express oneself.
Like the original POSS, each POSSA factor is scored on a 5-point Likert scale ranging from "Strongly Disagree" to "Strongly Agree." The items under each factor are added to make a total score for each factor where a higher score indicates a higher representation of that factor. For example, there are 6 items in the Adaptable Self factor which equate to an Adaptable Self score range of 7-42. The higher the total score, the more one presents various versions of themselves online.

Limitations and future research
While this study provides support for the psychometric value of the POSSA, this newly structured measure should be tested again across a variety of larger samples. This sample did show significant differences between male and female scores on the Adaptable and Authentic Self factors; however, the effect sizes were small and this study was comprised primarily of females. A more balanced gender ratio would be beneficial. Furthermore, culture has been associated with selfpresentation (Markus & Kitayama, 1991), and any insight into how culture may affect the notion of multiple selves or an ideal self would be of great future contribution on online self-presentation. Finally, self-presentation happens across numerous social networking platforms and mode (written, photo, video, etc.) of self-presentation varies as well as the specific culture of the platform. Therefore, the use of this scale across the different platforms should be assessed.

Conclusion
This study contributed to the literature by creating an alternate version of the POSS to use with an adult sample. A version that has taken into account the evolution of the "Ideal Self" that comes with the transition from adolescence to adulthood. It has presented a valid instrument to measure online self-presentation of adults by capturing elements of authenticity, discrepancy, and comfort with online self-presentation.
16. I feel I can be my ideal self online 17. I can talk to people who wouldn't usually talk to me in the real world *indicates a reversed scored item.