Selfie Appearance Investment and Peer Feedback Concern: Multi-Method Investigation of Adolescent Selfie Practices and Adjustment

Abstract

Sharing “selfies” on social media is common among adolescents. The frequency with which adolescents post selfies may be less important than behaviors and cognitions underlying selfie-posting, and these practices may differ by gender. This multi-method study explored selfie practices in a school-based sample of 639 adolescents (M_age=17.6; 53.5% female). Participants completed self-report measures of selfie practices, body esteem, depressive symptoms, and peer behaviors. In addition, a subset of participants’ social media pages (n = 245) were observationally-coded for numbers of selfies, followers, and likes. Factor analyses revealed two distinct selfie practices: selfie appearance investment and selfie peer feedback concern. Girls posted selfies more frequently, and reported greater levels of appearance investment and concern over peer feedback on selfies compared to boys. Multiple group structural equation models indicated that for boys and girls, selfie appearance investment was associated with depressive symptoms. For girls only, selfie peer feedback concern was associated with excessive reassurance-seeking and lower body esteem. No associations were revealed between observationally-coded measures of selfie-posting frequency and psychosocial outcomes. Overall, findings suggest that frequency of selfie-posting may be less relevant for understanding adolescent adjustment than investment in and concern over the selfie-posting experience.

Social media use is nearly ubiquitous among U.S. adolescents (Rideout & Robb, 2018). Social media users commonly share “selfies,” or photos of the self in which the individual is the focal subject, typically taken using a mobile device. Such photos are often carefully curated and typically receive peer feedback in the form of “likes” and “comments” (Dhir et al., 2016). The impact of such posts on adolescents’ peer relationships, mental health, and body image is not clear. Research on the link between adolescents’ frequencies of selfie-posting and psychosocial risk factors yielding mixed results with some studies finding positive effects, some negative effects, and some no effects at all (Chang et al., 2019; McLean et al., 2015). An emerging consensus among social media researchers suggests the need for more nuanced investigations in this area that move beyond measures of whether or how often selfie-posting occurs, and instead consider specific behavioral and cognitive factors that may explain why selfie-posting may be helpful or harmful (McLean et al., 2019). The current study identifies such specific behaviors and cognitions related to selfie-posting, explores how these constructs differ by gender, and examines associations between selfie practices and psychosocial factors.

Focus on Physical Appearance During Adolescence: Theoretical Perspectives

Adolescence is a developmental period of substantial biopsychosocial change, including pubertal changes, identity exploration, and increased salience of peer relationships (see Dahl et al., 2018). Many adolescents experience the imaginary audience, in which they believe they are the focus of their peers’ attention (Elkind, 1967). The developmental features of adolescence contribute to an increased focus on physical appearance. A peer appearance culture develops, in which peers regularly discuss their attractiveness and how to enhance it (Jones et al., 2004). Attractiveness becomes closely tied to adolescents’ popularity (Kleiser & Mayeux, 2020) and dating and sexual relationships (Ha et al., 2010), perhaps motivating adolescents to focus on how they look. Importantly, this focus on appearance is exacerbated for adolescent girls, who, according to objectification theory, are socialized to equate their self-worth with attractiveness (Fredrickson & Roberts, 1997). As a result, adolescent girls frequently engage in self-objectification, with implications for body image, wellbeing, and mental health (see Daniels et al., 2020). Yet adolescent boys also experience appearance concerns, often reflecting gender-specific appearance ideals (e.g., muscularity; Baker et al., 2019). The gender intensification hypothesis posits that gender roles and the prioritization of attractiveness for girls become particularly salient during adolescence (Hill & Lynch, 1983). These developmental and gender-related processes may be especially relevant in the context of social media.

Selfies and the Social Media Context

Social media sites represent a unique social context for adolescents, with a set of features that shape the way youth present themselves and interact with peers (e.g., boyd, 2010; Nesi et al., 2018). These features likely influence both the experience and psychosocial effects of posting selfies. For example, social media sites tend to offer quantifiable peer feedback (e.g., numbers of likes, comments), and involve content that is publicly shared, permanently available, and immediately accessible. Selfies may represent a confluence of physically attractive self-presentations and the peer appearance culture (Jones et al., 2004). Visual social media sites exacerbate adolescents’ emphasis on appearance by encouraging adolescents to carefully curate their posted selfies, presenting oneself as maximally attractive to garner peer approval (Yau & Reich, 2019). Moreover, exposure to peers’ idealized selfies can lead to appearance comparisons (Chang et al., 2019; Kleemans et al., 2018). The unique features of social media, and the ways these features intersect with adolescent peer appearance culture, make it critical to examine how selfie-posting beliefs and behaviors impact adolescents’ well-being.

Selfies and Psychosocial Adjustment

Given the importance of examining specific selfie practices in connection to adolescent adjustment, it is not surprising that prior work focusing exclusively on the frequency of selfie posting has produced inconsistent results. Correlational studies of adolescents and young adults have found associations between higher frequency of selfie posting and negative outcomes, including body dissatisfaction and overvaluation of shape and weight (McLean et al., 2015), self-objectification (Zheng et al., 2019), and body shame (Salomon & Brown, 2019). One experimental study of college women found that those assigned to post a selfie felt more anxious, less confident, and less physically attractive (Mills et al., 2018). Alternatively, some studies have found associations between selfie posting and positive outcomes, specifically higher body esteem or satisfaction (Chang et al., 2019; Cohen et al., 2018), and at least one longitudinal study of adolescents found no association between selfie-posting and body dissatisfaction (Wang et al., 2019).

Such results highlight the need to consider how and why adolescents differ in their vulnerability to the negative effects of selfie-posting. The Differential Susceptibility to Media Effects Model (Valkenburg & Peter, 2013) argues that media effects depend on individual vulnerabilities (i.e., dispositional, developmental, and social factors), types of media use, and cognitive and emotional responses to that use. Emerging work has identified two interrelated aspects of selfie-posting that appear particularly relevant for adolescents: investment in one’s appearance in selfies (reflected in photo editing, photo curation, and worries about appearance) and concern over peers’ feedback on one’s selfies (i.e., likes, comments).

Preliminary studies demonstrate the importance of selfie investment and photo-editing for adolescent girls’ body image and wellbeing. Body dissatisfaction is associated with higher levels of investment (e.g., effort toward choosing photos to share), while dietary restraint is linked to both investment and photo editing (McLean et al., 2015). Girls who engage in more frequent photo-editing report higher body dissatisfaction (Wang et al., 2019 and disordered eating (Lonergan et al., 2020), and lower body esteem, mediated by greater social media-based peer appearance comparisons (Chang et al., 2019). These editing and investment practices are associated with depressive symptoms (Lamp et al., 2019) in women, and body dissatisfaction in both women and men (Lonergan et al., 2019). Adolescent girls also report high levels of appearance-related social media consciousness (ASMC), or concerns about the social media audience and efforts to improve one’s attractiveness in selfies; higher levels of ASMC are associated with depressive symptoms and disordered eating (Choukas-Bradley et al., 2020).

Beyond appearance investment, social factors play an important role in adolescents’ selfie practices (McLean et al., 2019). Qualitative work among girls highlights the central role of peer feedback in driving selfie behaviors (Burnette et al., 2017; Chua & Chang, 2016). As with appearance investment, there is likely considerable individual variability in adolescents’ peer feedback concerns, with these individual differences affecting associations between selfie posting and psychological outcomes. For example, a greater desire for online popularity and peer approval among adolescents is associated with poorer self-esteem (Meeus et al., 2019) and greater engagement in health-risk behaviors (Nesi & Prinstein, 2019). Further, higher investment in selfie feedback is associated with greater body surveillance (Butkowski et al., 2019) and body dissatisfaction (Lonergan et al., 2019).

Given these initial findings related to selfie appearance investment and peer feedback concern, an important next step is to examine the implications of these two selfie practices on emotional (i.e., depressive symptoms), appearance-related (i.e., body esteem), and social (i.e., problematic social behaviors) adjustment. Very few prior studies have examined these two distinct practices independently, despite the fact that they may occur at different rates and differentially affect adolescent adjustment. Further, few studies have quantitatively examined selfie-specific peer feedback concern in relation to adolescent adjustment or directly observed selfie-posting behaviors. Studies also have been primarily conducted among young adults, despite adolescents engaging in more selfie-related behaviors than adults (Dhir et al., 2016). Social media may tap into the unique developmental processes that occur during adolescence (Nesi et al., 2018), with unique implications for mental health problems (depression, anxiety) that have increased in prevalence since the advent of social media (Twenge et al., 2018).

Gender Differences in Selfie Practices

While the majority of studies examining selfie practices have been conducted with girls and young women, the limited research investigating both boys’ and girls’ selfie-related experiences has revealed important gender differences. Girls are more likely to take, post, and edit selfies (Dhir et al., 2016; Fox & Vendemia, 2016). Girls also place more importance on appearing attractive in social media photos compared to boys (Yau & Reich, 2019), and demonstrate greater investment in receiving positive peer feedback (Nesi & Prinstein, 2019; Yau & Reich, 2019). With selfie appearance investment and concern about peer feedback possibly driving the negative effects of selfie behaviors, girls may be at greater risk. Indeed, “self-objectifying social media use” (including posting selfies) is associated with greater body shame via increased body surveillance, with this mediation effect stronger among girls (Salomon & Brown, 2019). These findings are consistent with objectification theory, which highlights how girls and women are socialized to be especially concerned about physical appearance (Fredrickson & Roberts, 1997).

However, boys are also concerned about their physical appearance (Jones et al., 2004; Baker et al., 2019), with prior work showing associations between social media use and appearance concerns among both boys and girls (Rodgers et al., 2020; de Vries et al., 2014; de Vries et al., 2016). Social media may emphasize attractiveness across all genders and thus increase boys’ focus on physical appearance (Manago et al., 2015). Although girls generally indicate greater online appearance concerns, boys also report investment in their appearance and selfies (Yau & Reich, 2019). Indeed, among boys, selfie editing has been cross-sectionally associated with appearance comparisons and disordered eating behaviors (Lonergan et al., 2020; Mingoia et al., 2019) and longitudinally associated with self-objectification (Wang et al., 2019).

Given initial evidence for gender differences in selfie behaviors and outcomes, it is critical to explore these processes among both girls and boys. Few studies have investigated boys’ selfie practices, and even fewer have compared selfie-related constructs by gender. Moreover, few studies have incorporated both self-reported measures and direct observation of youths’ actual social media profiles (Nesi & Prinstein, 2019), thus it remains unclear how these metrics of selfie behavior may vary by gender.

Current Study

The current study addresses limitations of prior literature by offering a multi-method investigation of selfie practices in a large, diverse sample of adolescent girls and boys. We examine adolescents’ behaviors and cognitions related to selfie-posting, and how these constructs differ by gender. We also examine associations between selfie practices and relevant psychosocial risk factors, including maladaptive peer behavior (i.e., excessive reassurance seeking), appearance concerns (i.e., body esteem), and depressive symptoms. Furthermore, we explore observationally-coded social media pages among a subset of the sample. We offer preliminary descriptive data on the frequency and prevalence of selfie posting, average numbers of followers, and “likes” per photo. Finally, we explore associations between observationally-coded selfie variables, self-reported psychosocial outcomes, and self-reported selfie practices.

We hypothesize that: 1) two aspects of selfie practices will emerge: selfie appearance investment and selfie peer feedback concern, from the Selfie Practices Scale; 2) compared to boys, girls will post selfies more frequently (based on both self-report and observationally-coded data) and report greater levels of appearance investment and feedback concern; and 3) the frequency of selfie-posting will be unrelated to psychosocial outcomes (based on both self-report and observationally-coded data), but that greater selfie investment and peer feedback concern will be associated with higher levels of depressive symptoms, excessive reassurance seeking, and lower body esteem, controlling for selfie-posting frequency.

Method

Participants

Participants were 639 adolescents (M_age=17.6) recruited from rural, lower-middle class high schools (11^th and 12^th grades) in the southeastern United States. Data collection for this study occurred during the fifth wave of a longitudinal study examining adolescent peer relationships and health-risk behaviors. At baseline, letters were sent home with students to obtain consent from parents; 900 students consented to participate (full recruitment details are available at [reference removed]), of which 648 (72.0%) participated at the fifth wave. Attrition was due to: study withdrawal (n=26), school withdrawal/expulsion (n=30), moving (n=104), death (n=2), or absent, unavailable, or other unknown reason (n=50). Of the 648 who participated at this time period, 9 did not complete any of the relevant study measures, resulting in a final sample of 639 students: 53.5% female, 45.8% male, and 0.6% other gender identities, and 46.5% White, 22.2% Black, 23.9% Hispanic/Latinx, and 7.4% other races/ethnicities. A total of 583 participants (91.2%) reported having or having access to a smartphone.

Procedure

Participants completed a computerized self-report survey in a classroom setting in April 2016. Participants were compensated with a $10 gift card. All procedures were approved by the university’s Institutional Review Board. Of all Instagram users at the time of data collection (n = 528, 82.6% of total sample), 46.4% (n=245) agreed to provide their Instagram user names and consented to be followed by an account created for this study. Among the 528 participants who had Instagram, those who agreed to observational coding of their Instagram pages (n=245) were compared to those who did not (n=283). There were no significant differences between these groups on self-reported frequency of selfie posting, overall appearance investment, body esteem, depressive symptoms, excessive reassurance seeking, age, or race/ethnicity. However, those who agreed to observational coding were more likely to be female, χ²=9.33, p=.002. Additionally, those who agreed to observational coding reported higher levels of selfie appearance investment, t(526)=−2.33, p=.02 and selfie peer feedback concern, t(526)=−2.07, p=.039. However, these differences were no longer significant after controlling for gender, B=.06, p=.178 and B=.07, p=.11, respectively, indicating that these differences were likely driven by the larger proportion of females sharing their Instagram pages.

Measures

Selfie Practices Scale

The 8-item Selfie Practices Scale was developed for the current study to assess participants’ selfie practices, including behaviors and cognitions identified in prior studies, such as investment in one’s appearance in selfies, use of filters and photo editing apps, importance of receiving likes and feedback, and concern about peers’ perceptions of one’s selfies (Brown et al., 1990; Mabe et al., 2014; McLean et al., 2015). See Table 1 for full item list. Response options ranged from 1 (Never) to 5 (Always). A factor analysis supported the expected two-factor structure (described in the Results).

Table 1.

Factor loadings (full sample) and frequencies of item endorsement (by gender) on the Selfie Practices Scale

	Full Sample	Girls (n = 344)			Boys (n = 295)
		Never or Rarely	Sometimes	Often or Always	Never or Rarely	Sometimes	Often or Always
	Factor Loadings	n (%)	n (%)	n (%)	n (%)	n (%)	n (%)
Factor 1: Selfie Appearance Investment
1. I use filters or photo editing apps to make photos of myself look better on social media.	.80	169 (49.1)	106 (30.8)	69 (20.1)	220 (74.6)	57 (19.3)	18 (6.1)
2. I pose in certain ways to make my body look better for photos on social media.	.62	200 (58.1)	91 (26.5)	53 (15.4)	226 (76.9)	55 (18.7)	13 (4.4)
3. I take multiple versions of photos of myself and select the best one to post to social media.	.78	104 (30.2)	81 (23.5)	159 (46.2)	201 (68.1)	57 (19.3)	37 (12.5)
Factor 2: Selfie Peer Feedback Concern
4. If no one comments on my photo about my appearance, I worry that it is not a good photo of me.	.75	256 (74.4)	52 (15.1)	36 (10.5)	244 (83.0)	40 (13.6)	10 (3.4)
5. Before I post a photo of myself on social media, I check with friends about whether I should post it.	.59	188 (54.7)	91 (26.5)	65 (18.9)	226 (76.9)	45 (15.3)	23 (7.8)
6. If I post a photo of myself on social media and it has not gotten very many likes/comments, I might ask my friends to like or comment on it.	.86	256 (74.4)	56 (16.3)	32 (9.3)	239 (81.6)	40 (13.7)	14 (4.8)
7. If I post a photo of myself on social media and it has not gotten very many likes/comments, I might take it down or delete it.	.83	226 (65.7)	70 (20.4)	48 (14.0)	235 (80.2)	40 (13.7)	18 (6.1)
8. It is important to me that people like and comment on photos of me that I post to social media.	.87	241 (70.1)	68 (19.8)	35 (10.2)	225 (76.8)	51 (17.4)	17 (5.8)

Note. Items were measured on a 5-point Likert scale: 1 (Never), 2 (Rarely), 3 (Sometimes), 4 (Often), and 5 (Always); scale points are collapsed (i.e., Never or Rarely; Often or Always) for ease of presentation. For males, n = 294 for items 2,4, and 5 and n = 293 for items 6, 7, and 8. Invariance testing supported metric invariance, indicating no difference in factor loadings by gender; thus, factor loadings were set equal across genders in final structural models.

Frequency of Selfie Posting

A single item assessed participants’ self-reported frequency of posting selfies on social media: “How often do you post a picture of yourself alone (no one else in the photo) on social media?” Item responses ranged from 1 (Never or Almost Never) to 6 (Multiple times per day). Response options were adapted from prior national surveys assessing frequencies of social media posting behaviors (Rideout, 2015).

Frequency of Time Spent on Social Media

A single item asked participants to self-report, on average, how much time they spend using social media on a typical day, with response options ranging from 1 (Less than 1 hour) to 11 (10 hours or more).

Appearance Investment

Overall appearance investment (i.e., not specific to selfies) was measured using the 12item Appearance Orientation subscale of the Multidimensional Body-Self Relations Questionnaire (MBSRQ; Brown et al., 1990). Responses were provided on a scale ranging from 1 (definitely disagree) to 5 (definitely agree). Item wording was edited slightly to match language currently used by adolescents (e.g., “grooming products” changed to “beauty or appearance-related products”). The MBSRQ has demonstrated excellent reliability and validity (Brown et al., 1990). Reliability was good in this sample (α=.82).

Depressive Symptoms

Participants reported on depressive symptoms using the 13-item Short Mood and Feelings Questionnaire (SMFQ; Messer et al., 1995). Items assess the degree to which participants agree with statements on a 3-point Likert-type scale 0 (not true), 1 (sometimes true), 2 (mostly true) over the past two weeks. An example item is: “I felt miserable or unhappy.” Reliability was excellent in the current sample (α=.94).

Body Esteem

Body esteem, or one’s self-evaluation of their body and appearance, was measured using the 10-item Appearance Evaluation subscale of the Body-Esteem Scale for Adolescents and Adults (BESAA; Mendelson et al., 2001). Adolescents reported agreement with statements regarding physical appearance (e.g., “I’m pretty happy about the way I look”) on a scale from 1 (never) to 5 (always). The scale has high internal consistency and test-retest reliability (Mendelson et al., 2001), and showed good reliability in this sample (α=.85).

Excessive Reassurance Seeking

Excessive reassurance seeking was measured using the Revised Excessive Reassurance Seeking Scale (Nesi & Prinstein, 2015; Joiner & Metalsky, 1995). It consists of 10 items relevant to adolescents (e.g., “I often ask people if they like me”) rated on a 5-point scale from 1 (Not at all true) to 5 (Extremely true). Reliability was excellent in this sample (α=.93).

Supplemental observationally-coded measures

An observational coding system was developed for analyzing participants’ Instagram pages (reference removed). All photos posted by participants during the three months prior to baseline data collection were coded. Of the 245 participants who provided access to their Instagram pages, 196 (80.0%) posted at least one photo during this period, with 1629 photos coded across all participants. Seven coders were trained over six weeks prior to coding photos. Then, each coded 51 practice photos, selected to ensure a range of content, with coders required to achieve K = .80 reliability with a master coder across all codes. The coding team discussed questions throughout the coding process, with the master coder (consulting with the lead author) making final decisions. Only photos (i.e., no posts containing videos, text, or memes) were coded. For each photo posted during the three-month period, coders indicated whether the participant: 1) was in the photo, and 2) was alone (i.e., no other people in photo). Photos that met these criteria were considered “selfies,” as long as participants appeared to be their current age (i.e., photos of the participant as a child were excluded). A sum of the total number of selfies posted during the three-month period was calculated, as well as a proportion of number of selfies to total photos posted during that period. Coders noted the number of likes on each photo, and an average was taken across all photos. The participants’ number of followers was also obtained.

Analysis Plan

Primary analyses were conducted with the full sample (n = 639). Descriptive statistics were obtained in SPSS 24.0. Structural equation models were constructed in MPlus 8.0 (Muthén & Muthén, 2017), using full information maximum likelihood estimation to handle missing data. First, an exploratory factor analysis was run on the Selfie Practices Scale. The resulting two-factor solution was then tested within a multiple group structural equation model (SEM) framework for measurement invariance by gender in three steps (testing for configural, metric, and scalar invariance). After establishing the measurement model, the structural model was constructed, again using a multiple group approach (see Figure 1). Exploratory descriptive analyses were also conducted with the subset of the sample (n = 245) with observationally-coded data including: 1) number of selfies posted, 2) proportion of all photos that were selfies, 3) number of followers, 4) average number of likes per photo, 5) percentage of participants with at least one posted selfie. To investigate associations between observationally-coded variables and self-reported outcomes, bivariate correlations were conducted separately by gender.

Figure 1.

Multiple-group structural equation model by gender of selfie practices (Selfie Appearance Investment and Peer Feedback Concern) predicting psychosocial outcomes.

Results

Descriptive Statistics

Participants self-reported posting selfies between once per week and once per month on social media (M=2.29), with 62.9% of participants reporting posting selfies at least once per month (Table 2); 20.5% reported posting selfies a few times per week or more. For observationally-coded selfie variables, 194 (79.18%) of the 245 participants with coded Instagram pages posted at least one photo during the coding period, and 135 (55.1%) posted at least one selfie. Participants received an average of 94.43 likes across photos (M=93.96 removing outliers¹) and had an average of 1008.17 followers (M=954.63 removing outliers; Table 2).

Table 2.

Means and standard deviations of study variables with gender comparisons

	Full Sample	Girls	Boys
	M (SD) / n (%)	M (SD) / n (%)	M (SD) / n (%)	t (df) / χ2
Self-Reported Measures
1. Selfie Appearance Investment	2.29 (1.06)	2.69 (1.03)	1.84 (0.92)	--
2. Selfie Peer Feedback Concern	1.83 (0.95)	2.00 (0.97)	1.63 (0.90)	--
3. Frequency of selfie-posting	2.29 (1.37)	2.65 (1.38)	1.89 (1.25)	−7.23 (622.97)***
4. Post selfies once per month or more	399 (62.93%)	259 (76.2%)	139 (48.3%)	52.34***
5. Appearance Investment	2.96 (0.71)	3.12 (0.71)	2.77 (0.66)	−6.31 (627)***
6. Body Esteem	3.49 (0.76)	3.40 (0.78)	3.60 (0.72)	3.30 (630)**
7. Depressive Symptoms	0.49 (0.53)	0.60 (0.56)	0.36 (0.47)	−5.86 (637.45)***
8. Excessive Reassurance Seeking	1.48 (0.69)	1.55 (0.69)	1.41 (0.69)	−2.56 (637)**
Observationally-Coded Measures
9. Selfies (sum)	2.24 (3.89)	2.83 (4.48)	1.21 (2.26)	−3.72 (236.67)***
10. Selfies (proportion)	0.36 (0.34)	0.42 (0.35)	0.24 (0.29)	−3.94 (160.05)***
11. Likes (mean)	94.43 (91.20)	99.53 (103.51)	84.22 (63.93)	−1.21 (190)
12. Followers	1008.17 (1005.29)	1001.37 (685.73)	1016.27 (1428.47)	0.11 (240)
13. Posted at least one selfie	135 (55.10%)	96 (61.5%)	38 (44.18%)	6.76**

^*p < .05;

^**p < .01;

^***p < .001. Scalar invariance by gender was not established for selfie practices measures (selfie appearance investment and selfie peer feedback concern), thus means were not compared between gender groups. Number of participants who completed the self-reported frequency of selfie posting item was 634 (340 girls, 288 boys, 4 other gender identities, 2 missing data on gender). Note that self-reported frequency of selfie-posting was rated on a 6-point Likert scale: 1 (Never or almost never), 2 (About once per month), 3 (About once per week), 4 (A few times per week), 5 (Everyday or almost everyday), 6 (Multiple times per day). Number of participants with observationally-coded measures was 245 (156 girls, 86 boys, 2 other gender identities, 1 missing data on gender). After winsorization procedures to handle outliers, Likes: M = 93.96, SD = 78.51; Followers: M = 954.63, SD = 700.65.

Selfie Practices Scale: Factor Analysis (Hypothesis 1)

To address Hypothesis 1, exploratory factor analysis was conducted using goemin rotation, a type of oblique rotation, to allow for correlations between factors. The optimal solution was identified based on chi-square difference tests to compare model fit, as well as consideration of Eigenvalues and interpretability of factors. Ultimately, a two-factor solution was retained. Compared to the one-factor solution, the two-factor solution showed significant improvement in model fit, Δχ²(7)=218.58, p<.001. Although model fit improved for the three- versus two-factor solution as well, Δχ²(6)=56.73, p<.001, a three-factor model contained two factors each with only two items, which were not interpretable. In addition, eigenvalues for the first (λ=4.99) and second (λ=0.88) factors were close to one, whereas the eigenvalue for the third factor was not (λ=0.53). The two-factor solution showed excellent fit, χ²(13)=78.55; CFI=.98,TLI=.96; RMSEA=.089, SRMR=.021. Geomin rotated factor loadings on each factor were good, with the majority of factor loadings above .78, and all greater than .59 (Tabachnick & Fidell, 2013). In line with Hypothesis 1, the two factors that emerged from the data were: (1) Selfie Appearance Investment, describing the use of strategies to maximize physical attractiveness in selfies, and (2) Selfie Peer Feedback Concern, describing behaviors and beliefs that reflect concern about peers’ feedback on one’s selfies.

Gender Differences: Descriptive Statistics and Tests of Measurement Invariance (Hypothesis 2)

In line with Hypothesis 2, girls endorsed all items on the Selfie Practices Scale and reported posting selfies at higher frequencies than boys (Table 1). For observationally-coded measures, a higher percentage of girls (64.5%) posted a selfie compared to boys (45.2%). Girls posted an average of 2.83 selfies during the coding period (3.52 among those who posted at least one photo), which was greater than boys’ average number of selfies posted (1.21; and 1.51 among those who posted at least one photo).

A multiple group structural equation model (SEM) was constructed to test for measurement invariance in the Selfie Practices Scale by gender (girls versus boys), including configural, metric, and structural invariance across gender groups, comparing model fit indices (model chi-square value, CFI, TLI, SRMR, and RMSEA) to determine whether constraining each set of parameters to be equal across groups resulted in a significant decrement in fit (see Table 4). Configural invariance, or equivalence of model form, was confirmed by acceptable overall model fit statistics for the multiple-group, two-factor model. Metric invariance, or equivalence of factor loadings across gender, was also supported. Scalar invariance was examined by constraining all item intercepts to equality across genders, and resulted in a significant decrement in fit compared to the metric invariance model. Thus, these constraints were systematically lifted, using modification indices as a guide, until adequate model fit was achieved. Ultimately, item intercepts for four items (items 1, 2, 3, and 5) were left to vary freely across gender, establishing partial scalar invariance. Fit indices for this model showed good fit, χ²(50)=194.65; CFI=.96, TLI=.95; RMSEA=.095, SRMR=.064. Full measurement invariance was not supported, factor means on the Selfie Practices Scale were unable to be compared across genders, and gender moderation was not tested for associations between selfie practices and outcomes.

Table 4.

Tests of Measurement Invariance by Gender in the Selfie Practices Scale

Model	χ 2	df	CFI	TLI	RMSEA	SRMR	Model Comparisons	Δχ2	Δdf	p	ΔCFI	ΔTLI	ΔRMSEA	ΔSRMR
M1:Configural Invariance	151.47	38	0.965	0.949	0.097	0.033	--	--	--	--	--	--	--	--
M2: Metric Invariance	171.05	46	0.962	0.953	0.092	0.052	M2 vs M1	19.58	8	.012	−0.003	0.004	−0.005	0.019
M3: Scalar Invariance	355.64	54	0.908	0.904	0.132	0.129	M3 vs. M2	184.59	8	<.001	−0.054	−0.049	0.04	0.077
M3b: Partial Scalar Invariance	194.65	50	0.956	0.950	0.095	0.064	M3b vs. M2	23.6	4	<.001	−0.006	−0.003	0.003	0.012

Note: Comparison of all model fit indices were examined due to the sensitivity of the chi-square change statistic to large sample sizes (Bentler & Bonnett, 1998). Note that Chen (2007) recommends no more than −.01 change in CFI, paired with changes in RMSEA of −.015 and SRMR of −.030 (for metric invariance) and −.015 (for scalar invariance).

χ² = chi-square; df = degrees of freedom; CFI = Comparative Fit Index; TLI = Tucker Lewis Index; RMSEA = Root Mean Square Error of Approximation; SRMR = Standardized Root Mean Square Residual; Δ = change in parameter. Model 1 (M1) = baseline model (without invariance); Model 2 (M2) = invariant factor loadings; Model 3 (M3) = invariant factor loadings and item intercepts; Model 3b (M3b): invariant factor loadings and item intercepts for only items 4, 6, 7, and 8; see Table 1).

Associations between selfie practices and psychosocial outcomes (Hypothesis 3)

Given that full measurement invariance was not supported across gender groups, associations between selfie practices and outcomes were examined separately for girls and boys in a multiple-group model (Figure 1), in order to address Hypothesis 3. All model paths were retained, regardless of significance, to allow for hypothesis testing. Model fit was adequate, χ²(110)=370.06; CFI=.94, TLI=.91; RMSEA=.086, SRMR=.054.

For girls and boys, Selfie Appearance Investment was significantly associated with depressive symptoms. For girls, Selfie Peer Feedback Concern was associated with excessive reassurance seeking and poorer body esteem (negative association), but for boys, Selfie Peer Feedback Concern was not associated with any outcomes. For boys and girls, self-reported frequency of posting selfies was associated with higher body esteem, and for boys, self-reported frequency of posting selfies was also associated with lower depressive symptoms. All models also controlled for overall appearance investment². See Table 5. In addition, bivariate associations suggest that observationally-coded selfie-posting frequency was not associated with any outcomes (Table 3). Self-reported selfie-posting frequency was associated with excessive reassurance seeking for both boys and girls.

Table 5.

Standardized results for multiple group by gender structural equation model predicting psychosocial outcomes from selfie practices

	Psychosocial Outcomes
	Body Esteem		Depressive Symptoms		Excessive Reassurance Seeking
	Girls	Boys	Girls	Boys	Girls	Boys
Predictors	β (SE)	β (SE)	β (SE)	β (SE)	β (SE)	β (SE)
Selfie Practices Scale Factors
Selfie Appearance Investment	−0.04 (0.10)	−0.18 (0.23)	0.32 (0.11)**	0.59 (0.23)**	−0.12 (0.11)	0.32 (0.20)
Selfie Peer Feedback Concern	−0.30 (0.9)**	−0.17 (0.21)	−0.04 (0.10)	−0.18 (0.21)	0.44 (0.09)***	0.19 (0.19)
Covariates
Appearance Investment	−0.18 (0.06)**	−0.07 (0.07)	0.04 (0.06)	0.03 (0.07)	0.06 (0.06)	0.14 (0.06)*
Frequency of Selfie Posting	0.13 (0.05)**	0.15 (0.06)*	0.04 (0.05)	−0.15 (0.06)*	0.09 (0.05)	−0.06 (0.06)

^*p<.05,

^**p<.01,

^***p<.001. Note that models were run separately by gender within a multiple group framework. Path analyses were re-conduced controlling for race/ethnicity, coded at four levels, i.e., White (reference group), Black/African American, Hispanic/Latinx, and Other Race/Ethnicity. Model fit remained adequate, χ²(146) = 411.21; CFI = .93, TLI = .90; RMSEA = .075, SRMR = .046. The pattern of all significant and non-significant results remained the same. Race/ethnicity was not significantly associated with any outcomes in the model, with the exception of a significant positive association between Black/African American and body esteem (B = .15, p = .011 for girls; B = .121, p = .036 for boys).

Table 3.

Bivariate correlations among study variables

	1	2	3	4	5	6	7	8	9	10	11
Self-Reported Measures
1. Selfie Appearance Investment	--	.78***	.37***	.42***	−.26***	.32***	.46***	−.11	-.12	.03	−.11
2. Selfie Peer Feedback Concern	.59***	--	.42***	.33***	−.28***	.27***	.49***	−.18	−.22	.15	−.01
3. Frequency of selfie-posting	.17**	.15**	--	.14*	−.02	.01	.16**	−.06	−.04	.17	.25*
4. Appearance Investment	.46***	.36***	.17**	--	−.17**	.19**	.33***	−.23*	−.09	.18	.08
5. Body Esteem	−.26***	−.35***	.05	−.29***	--	−.37***	−.32***	−.01	.03	.05	.19
6. Depressive Symptoms	.28***	.21***	.10	.19***	−.36***	--	.44***	−.18	.06	−.08	−.16
7. Excessive Reassurance Seeking	.21***	.37***	.16**	.19**	−.38***	.36***	--	−.14	−.21	.16	.02
Observationally-Coded Measures
8. Selfies (sum)	.09	.04	.09	−.01	.01	.07	−.07	--	.51***	−.02	.20
9. Selfies (proportion)	.11	.20*	.36***	.01	.10	.10	.02	.25**	--	.02	.11
10. Likes (mean)	−.03	.02	−.09	.21*	−.08	−.13	−.11	−.13	−.02	--	.73***
11. Followers	.04	.07	.03	.06	−.07	−.05	−.12	.20*	.05	.39***	--

Note. Correlations for girls presented below the diagonal, for boys presented above the diagonal. After winsorization for outliers, the association for boys between self-reported frequency of selfie-posting and followers became nonsignificant, r = .19, p = 0.83. Note that total N = 639 (for self-reported measures); n of subset with observationally coded measures = 245 (for observationally-coded followers and sum of selfies variables); n with at least one posted photo = 194 (for observationally-coded proportion of selfies and mean likes variables).

^*p < .05;

^**p < .01;

^***p < .001

Discussion

Concerns have been raised over whether adolescents’ selfie-posting is associated with harmful psychosocial effects. However, studies examining simply whether or how often adolescents post selfies have failed to demonstrate consistent associations with negative outcomes. Thus, it is critical to better explain why and how certain adolescents are susceptible to negative effects of selfie-posting. By examining selfie-related processes in a large, racially and ethnically diverse sample of adolescent boys and girls, this study offers an important extension of prior work. Overall, results of this study suggest that it is not how frequently adolescents post selfies that puts them at risk, but rather the specific behaviors and cognitions underlying their selfie-posting, that is associated with negative outcomes. Adolescent boys and girls who reported greater investment in their appearance in selfies showed higher levels of depressive symptoms, and, for girls, those who were more concerned about peer feedback on selfies reported poorer body esteem and maladaptive social behavior. Results demonstrate important differences between girls’ and boys’ selfie practices, highlighting the need for future work to consider gender in these processes.

Understanding Selfie Practices

Findings suggest that posting selfies is a relatively common behavior among adolescents. Participants self-reported that they, on average, post selfies between once per week and once per month on social media, with 62.9% of all adolescents reporting that they post selfies at least once per month and 20.5% reporting posting selfies a few times per week or more. Observationally-coded Instagram photos from a subset of the sample noted an average of 2.24 selfies across the 3-month coding period. On average, more than one-third of participants’ Instagram photos during this period were selfies. Importantly, neither self-reported or observationally-coded measures of selfie-posting frequency were significantly associated with mental health correlates.

The Selfie Practices Scale was developed to assess adolescents’ investment in and concern about selfies posted on social media. In support of Hypothesis 1, results suggested the emergence of two factors for both boys and girls: Selfie Appearance Investment (i.e., use of strategies to maximize physical attractiveness in selfies) and Peer Feedback Concern (i.e., behaviors and beliefs reflecting concern about peer’s feedback on selfies). Notably, prior work has sometimes conflated these processes. Indeed, high correlations were revealed between the two subscales, particularly among boys (rs = .59 for girls, .79 for boys), indicating overlap between social media practices that prioritize appearing attractive and gaining peers’ approval. However, results also highlight the utility of considering these processes separately, especially for girls, as they may represent distinct underlying risk processes forecasting different outcomes. The emergence of these two selfie-related factors mirrors important developmental features of the adolescent period, including the importance of physical appearance, salience of peer feedback, and presence of peer appearance culture (Dahl et al., 2018; Jones et al., 2004).

Selfie appearance investment behaviors were more common than expressions of peer feedback concern, particularly for girls. Indeed, half of all girls (and one-quarter of boys) reported at least sometimes manipulating photos of themselves through the use of filters or editing apps, and almost 70% of girls (and 32% of boys) reported at least sometimes taking multiple photos of themselves and picking the best one to post. However, certain behaviors reflecting peer feedback concern were also common, with nearly half of girls and one-quarter of boys reporting that at least sometimes, they ask friends before posting a photo of themselves. In terms of attitudes about peer feedback, roughly 30% of girls and one-quarter of boys reported that, at least sometimes, it is important to them that others like or comment on photos of them.

Relatively lower endorsement of peer feedback concern items may reflect social desirability biases. Importantly, qualitative work suggests that adolescents may be more likely to describe peer-related selfie concerns if they are discussing other adolescents’ behaviors, rather than their own. In focus groups with middle school girls, adolescents identified getting compliments from peers as the primary motivator behind posting selfies, but they attributed this goal to other adolescents, not to themselves (Burnette et al., 2017). More broadly, these focus groups suggested that adolescent girls may have “social media etiquette” that frames posting too many selfies as an undesirable behavior, and girls may be reluctant to acknowledge that they themselves post selfies (Burnette et al., 2017).

The frequency with which certain behaviors were endorsed in regard to selfies may reflect the unique features of social media, in which editing one’s photos, seeking out “likes,” and receiving quantifiable feedback from an audience of peers is a common experience (Nesi et al., 2018). Our findings contribute to a growing body of literature highlighting selfie-taking and posting as common adolescent behaviors (Mingoia et al., 2017). Social media may exacerbate the phenomenon of the imaginary audience, in which adolescents feel as if others are watching their every move (Elkind, 1967; Zheng et al., 2019), and these sites may provide the perfect context for the peer “appearance culture” (Jones et al., 2004). Furthermore, social media allows adolescents to engage in self-presentational and impression management strategies (Fox & Vendemia, 2016; Manago et al., 2008). At the same time, adolescents in our sample showed considerable variability in their investment in and concern over selfies, with those reporting higher levels of these practices at greater risk for poorer adjustment.

Gender Differences in Selfie Practices

These findings offer an important extension of prior work examining gender differences in social media behaviors. Consistent with Hypothesis 2, both observationally-coded and self-report data indicated that girls posted selfies at higher frequencies than did boys. Furthermore, girls endorsed nearly all items reflecting Selfie Appearance Investment and Peer Feedback Concern at higher rates, including reporting more frequently using filters and photo editing apps, posing in ways that make them look better in photos, asking friends to like or comment on photos of them, and checking with friends before posting a photo of themselves. Prior qualitative research indicates that adolescent girls may view photo-editing as necessary in order to appear attractive, and that they engage in photo-editing as a self-presentational strategy necessary to adhere to standards of beauty depicted in mainstream media (Chua & Chang, 2016).

Gender differences also emerged in evaluation of the Selfie Practices Scale. Specifically, item intercepts differed by gender for the three items comprising the Selfie Appearance Investment factor, and the Peer Feedback Concern scale item about checking with friends before posting. This suggests that selfie practices, particularly selfie appearance investment, may vary by gender. For example, while girls may be more likely to use filters or photo editing apps, greater use of filters does not necessarily indicate greater selfie appearance investment for girls compared to boys. These findings are not surprising, given prior work suggesting that filtering and editing photos is more normative among girls (Fox & Vendemia, 2016), as is asking for help from friends before posting a photo (Terán et al., 2019). These specific behaviors may reflect broader gender differences in the focus on physical appearance. For example, objectification theory highlights that adolescent girls are socialized to prioritize their physical appearance and to monitor their bodies for compliance with beauty norms (Fredrickson & Roberts, 1997), and these experiences are likely exacerbated through social media (Nesi et al., 2018). Concerns about one’s appearance in selfies may also align with cultural norms of femininity which, according to the gender intensification hypothesis, become particularly salient and prescriptive during adolescence (Hill & Lynch, 1983). However, our findings indicate that some boys also experience appearance-related concerns related to selfies, highlighting this as an important area for future research. More broadly, as researchers identify new constructs capturing adolescents’ social media experiences, it will be critical to consider gender differences in the makeup and validity of these constructs.

Selfie Practices and Adolescent Adjustment

In addressing Hypothesis 3, results indicated that it was not the frequency of selfie-posting that was associated with poorer adjustment, but rather adolescents’ appearance investment and concern over peer feedback in their selfies. For both boys and girls, selfie appearance investment was associated with higher levels of depressive symptoms, controlling for overall appearance investment and self-reported frequency of selfie-posting. Furthermore, for girls, selfie peer feedback concern was associated with higher levels of excessive reassurance-seeking and lower body esteem. Thus, selfie posting may only be associated with poorer outcomes when accompanied by greater effort invested into improving how one looks in photos, and for girls, greater concern over how peers will respond, as reflected in behaviors and cognitions.

The current study adds to a growing body of work indicating the importance of understanding young people’s personal and subjective experiences with social media, particularly with regard to appearance-related behaviors (Cohen et al., 2018; Meier & Gray, 2014). Appearance-focused activities on social media are associated with body image dissatisfaction and disordered eating (Holland & Tiggemann, 2016; Lonergan et al., 2020; McLean et al., 2015; Terán et al., 2019), and internalization of the thin ideal (Mingoia et al., 2017). Further, prior work indicates associations between social media-based social comparison and body dissatisfaction (Fardouly et al., 2015) and depressive symptoms (Nesi & Prinstein, 2015), as well as between investment in selfies and thin-ideal internalization, body dissatisfaction, and disordered eating (Butkowski et al., 2019; Cohen et al., 2018). Our findings highlight potential risks associated with youths’ specific selfie-related practices. Research that focuses solely on overall social media use may miss such individual differences in young people’s subjective experiences online.

Our finding that selfie feedback concern was associated with higher levels of excessive reassurance-seeking only among girls may be related to gender differences in susceptibility to interpersonal factors. During adolescence, girls are socialized toward traits that may predispose them to depressive symptoms, including relying on interpersonal relationships for one’s self-worth, tendency toward rumination and co-rumination, and generally heightened sensitivity to interpersonal stress (Nolen-Hoeksema & Girgus, 1994). The features of social media may encourage girls—especially those highly concerned with peer feedback—to engage in excessive reassurance seeking, which may put them at risk.

Interestingly, higher self-reported frequency of selfie posting was associated with higher body esteem in both girls and boys, and with lower depressive symptoms in boys. This is consistent with prior work showing positive associations between selfie-posting and body esteem (Chang et al., 2019; Cohen et al 2018), and suggests that there may be beneficial aspects of posting selfies online. These results may also be driven by the receipt of positive feedback on selfies (McLean et al., 2019). However, it is also possible that those who post selfies simply have higher body esteem to begin.

Limitations and Future Directions

Future studies should address limitations of the current work. First, in this study it was not possible to determine whether selfie practices led to poorer psychosocial adjustment, or vice versa. Longitudinal and experimental studies are needed to establish directions of causality in these processes. Second, future work should clarify definitions of “selfies,” and consider a broader set of experiences related to selfies. In this study, self-report items asked simply about “photos of the self” (for the Selfie Practices Scale) or “photos of the self alone” (for the selfie frequency item) posted to social media. For purposes of observational coding, any Instagram photos of the participant alone were considered selfies. Other selfie-related behaviors may also be relevant for adolescent psychosocial development, including taking (versus posting) selfies, posting photos of the self with others, viewing others’ selfies, appearance comparisons with others’ selfies, and imaginary audience ideation (Kleemans et al., 2018; McLean et al., 2019; Terán et al., 2019; Zheng et al., 2019). Future research should build on the current findings to examine broader mechanisms likely underlying associations between selfie-posting and psychosocial adjustment, such as self-objectification (Cohen et al., 2018; Terán et al., 2019) and appearance comparisons (Fardouly et al., 2015; Kleemans et al., 2018). In addition, a growing body of work exists linking narcissism to selfie-posting, primarily in adults (McCain & Campbell, 2018); further work should explore the role of narcissism and other personality characteristics in adolescents’ selfie investment and concern. Finally, more research is needed to understand the role of gender, including among adolescents identifying as gender minorities, in selfie-related behaviors and cognitions, and in psychological adjustment outcomes.

Conclusions

Overall, results suggest that frequency of selfie-posting may be less relevant for understanding adolescent adjustment than investment in and concern over the selfie-posting experience. These findings underscore the importance of understanding adolescents’ specific, subjective experiences with social media. During a developmental period characterized by identity exploration, a focus on physical appearance, and heightened importance of peer feedback, selfie-posting reflects a complex behavior at the intersection of self-presentation and feedback-seeking. Future research should use longitudinal and experimental methods to better understand adolescents’ motivations for selfie-posting and the psychological outcomes of these experiences.

Public Significance Statement.

This study suggests that the frequency with which adolescents post selfies on social media may be less important to their well-being than their investment in and concern over the selfie-posting experience. For all adolescents, higher investment in appearance in selfies was associated with depressive symptoms. For girls only, concern about peer feedback on selfies was associated with lower body esteem and reassurance-seeking behavior.