Conceptualizing the Role of Racial–Ethnic Identity in U.S. Adolescent Social Technology Use and Well-Being

Abstract

Adolescent development and wellbeing now involve how the use of social technologies (e.g., social media and other online spaces) impact daily life. Especially during crises such as COVID-19 and persistent injustices, adolescents rely on online spaces for social connectedness and informational knowledge. Psychosocial impacts, both positive and negative, have been found among racial-ethnic minority adolescents. However, the role of racial-ethnic identity on social media use and wellbeing has been understudied. The current study addresses differential associations on social media experiences and mental health (i.e., depressive, online anxiety symptoms) among a diverse group of adolescents (n = 668; ages 10-17; 45.7% non-White). Furthermore, the roles of self-identified racial-ethnic groups, identity importance, exposure to hate messaging, and gender are investigated. Our study found significant moderating effects of racial-ethnic importance, gender, and online hate messaging. Additionally, the moderating role of race-ethnicity reveals a stronger association between greater social media frequency and heightened depressive symptoms among Asian adolescents. Black adolescents showed a significant association between greater social media frequency and decreased online social anxiety. Significant effects of online hate messaging exposure also reveal associations between online behaviors and depression and online social anxiety across adolescents. As social media adoption coincides with identity exploration, this study highlights how racial-ethnic identity and its formation in the digital age is important to understand its association with online interactions that may help or hinder adolescent wellbeing. Future work should continue examining trajectories of identity formation in relation to social media content and differential mental health impacts.

Background

In the United States (US), we live in a unique time of history where racial-ethnic and cultural diversity continues to thrive, and the historically racial-ethnic minorities of this country now comprise over half of the youth population under the age of 15 years old (U.S. Census Bureau, 2019). Racial-ethnic (RE) identity development is dynamic among adolescents, which now co-occurs alongside the exploration and integration into digital online spaces. The rapid adoption and ubiquity of social technologies (e.g., social media, interactive online forums) includes increased access to seek out and exposure to information and more opportunities for civic engagement online related to persistent and current social injustices (Tao & Fischer, 2022; Vogels et al., 2022). The “digital divide” among adolescents has more recently deviated away from being focused on the point of access to technologies based on socioeconomical, educational, regional, or racial backgrounds and pivoted towards understanding the division of how diverse adolescents are using these technologies in adaptive or maladaptive ways (George et al., 2020). While RE minority youth are among the largest consumers of digital media (Vogels et al., 2022), the research has a limited focus on the sociocultural and contextual factors related to this identity formation, particularly in an online milieu.

Despite most ongoing research focusing on the negative impacts of social media use (e.g., cyberbullying), emerging research evidence has suggested these online platforms may have mechanisms which provide counternarratives for how these tools support positive development (Odgers & Robb, 2020; Tao & Fischer, 2022), especially for youth from marginalized communities (Bravo et al., 2019; Brough et al., 2020). In this article, we explored the intersections of adolescent RE identity development, online experiences, and mental health applying aspects of traditional psychosocial developmental (Erikson, 1968; Phinney, 1990) and critical race theories (Crenshaw et al., 1995) to the digital age. As rising concerns for adolescent mental health and persistent inequities among communities of color (Brown, 2003; Ladson-Billings & Tate, 1995) continue, it is important to understand how contexts of identity shapes the way that social technologies may be helping or hindering adolescent wellbeing.

Adolescent Racial-Ethnic Identity and Mental Health

RE identity formation becomes more salient in development as heightened social cognition leads to self-awareness around individual identity (Phinney, 1990). Adolescents experience a transition towards increased self-awareness and exploring the balance between family and external influences (e.g., peers and community members, media and online content) in alignment with the fifth stage of Erikson’s stage of psychosocial development (Erikson, 1968). Coinciding with shifting priorities of social impact, connectedness, and autonomy among adolescents (Williams et al., 2014), RE identity, and in particular, RE centrality (i.e., how important one’s race or ethnicity is to an individual’s identity), has been described as a point of strength within minority youth resiliency (Charmaraman & Grossman, 2010; Masten & Reed, 2002). Differences in how one’s RE identity influences psychosocial development may be partially explained by variability in the importance of one’s identity (Charmaraman & Grossman, 2010). Prior research has shown a stronger sense of RE identity among Black and Latine adolescents and higher levels of identification among females compared to males (Charmaraman & Grossman, 2010). Those who tend to face greater racial adversity are also more likely to attribute greater importance of RE aspects to their social identities (Charmaraman & Grossman, 2010). In contrast, low levels of RE importance among White adolescents have been tied to discomfort with White privilege and acceptance of Whiteness as the “norm” (Frankenberg, 1993; McIntosh, 2003), as well as post-cultural color evasive (more commonly recognized as “colorblind”) ideologies which reject the notion that one’s identity is related to one’s physical appearance (Grossman & Charmaraman, 2009).

Parallel to the implications of RE identity formation is an increased awareness of marginalization and deep histories of injustice and obstacles in the US which RE minorities continue to endure. It is worth noting the significant histories (e.g., slavery, incarceration, colonization, segregation) which have been passed down through the generations in the US and in many ways have been amplified in the era of social media. From a Critical Race lens, these oppressive histories are perpetuated today promoting inequitable systems that stem from the superior complex of Whiteness which infiltrate the systems of justice (Crenshaw, 1989; Crenshaw et al., 1995), education (Ladson-Billings & Tate, 1995), family outcomes (Burton et al., 2010), and mental health (Brown, 2003). Continued marginalization of many ethnic communities across the US has led to persistent and disproportionate poverty, violence, and social inequities. The effects of these histories are often lost by minimizing the salience of RE identity and cultural influences on adolescent development. These effects remain integral to child outcomes (Williams et al., 2012), especially in a moment of heightened oppression in the era of pervasive digital media consumption.

Considering the historical and cultural contexts, there is emerging scholarship that links perceived discrimination as a strong predictor of increased stress and anxiety throughout adolescence with lasting effects into young adulthood (Benner et al., 2018; Tynes et al., 2020). Among Black and Latine teens, increased early experiences of online racial discrimination was linked to heightened depressive and anxiety symptoms in later adolescence (Tynes et al., 2020; Umaña-Taylor et al., 2015). RE minority youth are often exposed to discrimination at an early age, which can lead to lasting impacts of internalizing and externalizing behaviors (Benner et al., 2018; Priest et al., 2013; Rivas-Drake et al., 2014). The implications of discrimination on mental health (Tao & Fischer, 2022; Tynes et al., 2020; Umaña-Taylor et al., 2015) should be interpreted through the lens that youth of color who are also less likely to have access to mental health care due to systemic barriers (Nestor et al., 2016). Individuals from marginalized groups often encounter significant distrust towards the healthcare systems which often cascades down through a family system from the caregivers to youth (Aguilar-Gaxiola et al., 2012; Rodgers et al., 2022). Furthermore, these barriers are often compounded with racial-ethnic based cultural perceptions of mental health such as a male-centric perception of machismo in Latine cultures (Barrera & Longoria, 2018), or the pressures of model minority status among Asian Americans leading to increased risk for internalizing behaviors (Sue & Sue, 1973; Yoo et al, 2010).

However, prior research has also demonstrated the possible protective characteristics of heightened RE identity among Filipino Americans (Mossakowski, 2003), which indicated that this sense of identity formation may act as a buffer for harm incurred through discrimination. For American Indian adolescents, self-esteem significantly mediated the relation between the centrality, or strength, of ethnic identity and levels of anxiety also showing varying pathways based on identity on mental health (Charmaraman & Grossman, 2010; Smokowski et al., 2014). Therefore, the mental health implication of RE identity during adolescence in remains complex, yet critical to understanding development in an increasingly diverse digital age.

Online Experiences of Racially and Ethnically Diverse Adolescents

Research evidence has exemplified distinct RE differences with regard to experiences and impacts using online spaces. With 95% of youth being connected online by their teenage years (Vogels et al., 2022), there is increased recognitions by key adolescent stakeholders (e.g., parents, educators, researchers, policymakers) around the ubiquity of the online ecosystem. African American adolescents are more likely to report that they really enjoy using social media compared to their White peers (Rideout et al., 2022), and Hispanic/Latine youth are more likely (54%) than their White counterparts (41%) to be online constantly (Vogels et al., 2022). Recent reports suggests that Black youth (ages 8-18) are using social media for a significantly greater amount of time each day (2 hours 50 mins.) than their Hispanic/Latine (2 hours 19 mins.) and White counterparts (2 hours 05 mins.), respectively (Rideout et al., 2022). Beyond the assessment screen time, prior research has outlined both positive and negative influences of online spaces for diverse groups of adolescents (Charmaraman, Hernandez, & Hodes, 2022; Ito et al., 2020; Odgers & Jensen, 2020; Odgers & Robb, 2020). Jackson and colleagues (2009) find that African American adolescent girls are more likely to spend greater time online, which is a positive predictor of child outcomes such as academic achievement. Furthermore, Latine and Black youth from lower-income neighborhoods found that despite being embedded in a digital ecosystem that exposed them to more content around substance use, violence, and sexual content, most do not actively engage with such content and are self-aware of the risks even online (Stevens et al., 2019). Online contexts may also provide the opportunity for youth of diverse backgrounds to expand social capital online and beyond their offline environment despite the oppressive systems that exist (Brough et al., 2020).

Online connections among peers with similar background or cultural reliability may offer a protective and positive impact on RE identity for adolescents and may also be a signal for expansion of bi- or multiculturalism to support wellbeing. Larson (2002) shows the early potential for the internet to foster a place for ethnic minority youth to connect with others like themselves which inherently promotes a sense of ethnic identity. In a mixed-method study of diverse adolescents and young adults aged 12-25, Charmaraman and colleagues (2015) found that girls and young women of color participated in more online blogs and were more likely to report revealing their stress on social media compared to both White and male participants. Additionally, the qualitative interview data revealed the ways in which young women of color are joining or following websites such as Racialicious.com and Colorlines.com, writing and re-blogging social commentary on racial identity issues, and joining culture-related online groups as a part of school-related activities (Charmaraman et al., 2015). As children adopt the internet and other social technologies at a younger age today, we must consider the potential that this particular form of identity formation may be occurring at an earlier time in development than previously studied and that the exposure to various stereotype narratives portrayed in online spaces may impact long term wellbeing.

Current Study

Currently, many studies conducted on RE identity development and impacts of social media use have focused on the latter half of adolescence or young adulthood. The current cross-sectional survey study consists of an ethnically and racially diverse cohort of early to mid-adolescents (ages 10-17) during the COVID-19 pandemic when hybrid remote learning and physical distancing became the norm across the US. This dataset has also uniquely positioned us to explore online reactions to social and racial injustices of recent times and how social media might play a role in online behaviors and mental health of adolescents.

The primary aim of this study was to examine the association between online behavior indicators and adolescent internalizing symptoms (e.g., depression, online social anxiety) among a racially-ethnically diverse sample during the COVID-19. We hypothesized significant effects of the RE identity levels of importance on internalizing behavioral outcomes on online experiences. Next, we analyzed the moderating role of (1) the RE importance at this given moment of development, (2) self-identified RE, and (3) online hate messaging exposure. Due to the salient nature of identity development during adolescence, we believe that RE identity and importance, and related implications of race and ethnicity such as hate messaging, are important constructs to test as potential moderators on mental health (i.e., depressive symptoms and online social anxiety) of diverse online experiences (i.e., frequency of social media use, online harassment, and prosocial social media use). Lastly, we explored additional moderating roles of gender, as prior literature also indicates gender differences in online experiences and internalizing behaviors (Twenge, 2020).

Method

Participants and Study Design

This study used data collected as part of a larger longitudinal study on early adolescent social technology use since 2018 (Charmaraman et al., 2022a; Charmaraman et al., 2022b). Students from three urban and suburban middle schools and one high school in the Northeastern US were recruited based on varying school enrollment size, internet accessibility (e.g., Chromebooks provided for students), and diverse RE composition (e.g., 32 to 47% of the students depending on the school identified as a RE minority). The percentage of students from lower socioeconomic backgrounds, measured through the participation of free or reduced lunch programs, ranged from 32 to 45% depending on the school. English was not the primary language spoken in the home for 23 to 48% of students, also dependent on the school. Upon obtaining Institutional Review Board (IRB) approval from our institution and school district-level permissions, we worked with school and Parent-Teacher Organization (PTO) liaisons to distribute informed consent/opt-out forms (in English, Spanish, and Portuguese) to parents through paper flyers, parent email listservs, school electronic newsletters, and direct emails. For the Fall 2020 data collection, trained members of the research team virtually proctored the online Qualtrics survey emailed to students using Google Meets during a pre-scheduled advisory period in two schools lasting up to 60 minutes. For the other schools that participated through the PTO liaisons, survey links were distributed through parents who then shared with their own children to participate at their convenience. The total sample size of the larger longitudinal study consisted of 968 students. Survey participation rates averaged 76%, which ranged from 40% through PTO outreach for home completion to 90% during a designated in-school advisory period to complete the survey. Less than 3.5% of parents opted out, 4.0% of students opted out or did not provide assent, and 7.2% were absent on the day the survey was administered in-person. For the variables related to this study, 69% (n = 668) of adolescents enrolled in the study completed all questions of the survey (see Statistical Analysis Section and Supplementary Table S1 for further information).

Measures

Adolescent Identity and Characteristics

Demographics and Background Characteristics.

Participants reported their age (10-17 years old; M = 13.2, SD = 1.53) at the time of survey completion, along with the gender they identify with (i.e., “Are you a…” 1 = Male; 2 = Female; 3 = Other), and self-identified race-ethnicity. For the race-ethnicity item, participants could check all that apply from the following: a) Asian/South Asian/Pacific Islander, b) Black/African American/West Indies, c) Latin American/Central or South American/Brazilian, d) Native American, e) White/Caucasian/European, f) Middle Eastern, or g) Other (fill-in). A Multiracial category was created for participants who checked more than one of the RE groups listed. They were also asked about participating in the free or reduced lunch programs as an indicator of family economic standing.

Racial-Ethnic Identity Importance.

We adapted Yip’s (2005) Ethnic Salience measure that included 8 domains for social identities originally developed for older adolescents. To be more inclusive of domains of relevance to younger adolescents, we added 10 additional items to the measure. Adolescent participants were asked to think about the identities or parts of them that make up who they are with the following question: “Which of the following top 5 are important to you in your life right now?” Participants selected their 5 most important identities from the following 18 options: age, student at my school, daughter/son, sibling, friend, gender, race/ethnicity, neighborhood or town where I live, religion, significant other, clubs/organizations I belong to, athlete, American, immigrant family, sexual orientation, hobbies/interests, physical appearance, and having a pet. For the purposes of this study, we extracted the single-item dichotomous variable response for “race/ethnicity” as a proxy measure to centrality to identify those who rank this social identity with greater importance or lower relative to other possible social identities.

COVID-19 Peer Relationships.

Given that this wave of the study was collected during the COVID-19 pandemic (Fall 2020), the item “Since the beginning of social distancing due to COVID-19, has the quality of relationships with your friends changed?” has also been included to account for relational impacts of the pandemic. This item was measured on a 5-point Likert scale (1 = A lot worse to 5 = A lot better).

Online Experiences

Frequency of Social Media Use.

We asked participants to indicate how often they checked their social media on a 6-point Likert scale (1 = Never/Does not apply to me to 6 = More than every hour).

Prosocial Social Media Use.

Adapted from a digital citizenship scale (Jones & Mitchell, 2016) and Facebook Relationship Maintenance Behavior Scale (Ellison et al., 2014), 8 items evaluated participants’ use of social media for educational purposes, providing social support to friends, joining groups that make them feel less lonely, organizing civic engagement activities or raising awareness of issues the participant cares about. Adolescents reported on a 5-point Likert scale (1 = Never to 5 = Always). Internal validity of this measure was acceptable (α = 0.77).

Online Harassment.

Items were adapted from validated measures of Internet Harassment Victimization and Internet Sexual Solicitation Victimization (Ybarra et al., 2007). Participants were asked about negative online experiences such as how often someone else made rude or mean comments, someone else spread rumors, the participant was hurt by someone excluding them, or receiving unwanted attention via inappropriate sexual messaging. Responses for this five-item measure were reported on a 4-point Likert scale (1 = Never to 4 = Often). The internal validity of this measure was acceptable (α = 0.82).

Exposure to Online Hate Messages.

Participants were asked whether, in the past 12 months (e.g., December 2019-December 2020), they had seen websites or social media posts where other people discussed the following: “Hate messages that attack certain groups or individuals” and was coded as a dichotomous variable (1 = Yes, 0 = No). If participants selected “Yes” for this item, they were invited to describe what groups they saw were being targeted. Responses from those who reporting seeing online hate messages included “race”, “racism”, “POC”, “going back to their country”, “minorities”, “Black Lives Matter”, “dark-skinned”, LGBTQ communities, political affiliations.

Internalizing Behaviors

Depressive Symptoms.

We measured depressive symptoms using the 10-item Depression Scale Revised (CESDR-10), which was validated on diverse adolescent samples (Haroz et al., 2014). Participants reported the number of days in the past week they felt the presence of 10 depressive symptoms, such as “I was bothered by things that usually don’t bother me” and “I had trouble keeping my mind on what I was doing.” Response choices included “not at all or less than 1 day,” “1-2 days,” “3-4 days”, or “5-7 days.” A score of 10 or higher indicates the presence of significant depressive symptoms. The internal validity of this measure is α = 0.84 for this sample.

Online Social Anxiety Scale.

This measure was adapted for an online context using the Fear of Negative Evaluation Subscale from the Social Anxiety Scales for Adolescents Short Form (SAS-A, Nelemans, 2017). The three-item measure included: I worry about what others think of me on social media; I worry about what others say about me on social media; I’m afraid that others won’t like me on social media. This scale used a 4-point Likert response format (1 = Mostly Disagree to 4 = Mostly Agree) and the Cronbach’s alpha in this sample was α = 0.93.

Statistical Analysis

Analyses were conducted in R Version 4.1.0 (R Core Team, 2022) using RStudio (R Studio Team, 2020). Complete case analyses were conducted in the current study to allow for the handling and interpretability of a stepwise regression modeling approach using the ‘stats’ (R Core Team, 2022) and ‘car’ packages (Fox & Weisberg, 2019). Although we recognize that multiple imputations are preferable to listwise deletion, majority of the data remains retained in the current analyses (69%; n = 668) and preliminary analyses indicates minimal biases through score differences across the outcome variables (i.e., depression, online social anxiety) between full and complete case sample (see Table S1 in Supplemental Materials for full comparison). Descriptive statistics were calculated using bivariate correlations and an analysis of covariance (ANCOVA) omnibus test to test mean differences across adolescents stratified into RE groups they identified with. The results of the ANCOVA reflect group differences between the specified RE group compared to the rest of the sample outside of the specified subgroup. Demographic differences across RE groups and online hate message exposure are reported in Table 1.

Table 1.

Participant Demographics

	N	%	High Racial-Ethnic Importance	Online Hate Message Exposure
Gender			N (% within group)	N (% within group)
Male	278	41.6	36 (13.0)	26 (9.35)
Female	390	58.4	68 (17.4)	54 (13.8)
Age (in years)
10-13 (Early Adolescence)	393	58.8	56 (14.2)	40 (10.2)
14-17 (Mid Adolescence)	275	41.2	48 (17.5)	40 (14.5)
Race-Ethnicity
White/Caucasian/European	363	54.3	12 (3.31)	36 (9.92)
African American/Black	62	9.3	28 (45.2)	13 (21.0)
Asian/South Asian/Pacific Islander	42	6.3	20 (47.6)	9 (21.4)
Latine	69	10.3	17 (24.6)	7 (10.1)
Other**	64	9.6	9 (14.1)	8 (12.5)
Multiracial***	68	10.2	18 (26.5)	7 (10.3)
Free or Reduced Lunch *
Yes	258	38.6	44 (17.1)	29 (11.2)
No	238	35.6	28 (11.8)	28 (11.8)
I don’t know	172	25.8	32 (18.6)	23 (13.4)

^*Free or Reduced lunch is used as an indicator of family economic status reported by youth.

^**Other race-ethnicities includes individuals of Middle Eastern (n=17) and Native American (n=7) descent and others identified by youth

^***Multiracial includes adolescents who selected more than one racial-ethnic identity.

To further address the aims of our study, understanding the moderating effects of indicators related to RE identity on online and internalizing behaviors of adolescents, we conducted three multivariate moderated multiple regressions (MMMR). Baseline models included the independent variables RE identity, RE importance, and the online experience variables social media use frequency, prosocial social media use, hate message exposure, and online harassment. These variables were also selected based on prior research of online indicators of online harassment and bullying (Przybylski & Bowes, 2017), and prosocial social media behaviors (Ellison et al., 2014; Jones & Mitchell, 2016). The dependent variables included as multivariate internalizing behavior outcomes were depressive symptoms and online social anxiety. Covariates for the analyses included age, gender, free or reduced lunch participation (i.e., socioeconomic indicator), and peer relationship quality during COVID-19 to control for changes related to the pandemic. In a stepwise manner, we then tested the unique moderating roles of (1) levels of RE importance, (2) self-identified RE, and (3) online hate message exposure by sequentially testing whether, for example, RE importance significantly moderated the effect of social media use frequency, prosocial social media use, hate message exposure, and finally online harassment in the association with anxiety and depression.

Then, we also tested the stepwise addition of exploratory three-way interactions with gender as a second moderator in order to further expand on existing literature on gender differences between online experiences and psychological outcomes during adolescence (Twenge, 2020). For each of the three moderator model sets, univariate effects from the final model were interpreted. This stepwise, nested model approach allowed us to test omnibus effects of indicators related to RE identity on outcomes of internalizing behaviors previously shown to be associated with adolescent patterns of online and social media experiences (e.g., prosocial behavior, online harassment). This also maximized power by removing non-significant interactions, which was crucial given the limitation of a relatively small sample size of several racial-ethnic subgroups in this study.

Results

Descriptive Statistics

The demographic breakdown of this study sample (N = 668) can be found in Table 1. Bivariate correlations were conducted as a preliminary analysis to understand the relations between internalizing and online behaviors and social technology engagement indicators (see Supplemental Table S3 for full results). The ANCOVA stratified by RE groups revealed that Black adolescents reported on average higher levels of depressive symptoms (M_Black = 8.75, SD_Black = 6.40) compared to the rest of the sample (M_non-Black = 7.26, SD_non-Black = 5.85) and no significant differences across subgroups for online social anxiety. Frequency of social media use was significantly lower for Asians (M_Asian = 3.79, SD_Asian = 1.82; M_non-Asian = 4.33, SD_non-Asian = 1.67) and multiracial adolescents (M_Multi = 3.88, SD_Multi = 1.84; M_non-Multi = 4.34, SD_non-Multi = 1.66) compared to the rest of the sample. Eighty adolescents reported witnessing hate messaging online in the past 12 months, which comprised 12.0% of the total analytical sample. Of the responses, 21.0% of Black youth and 21.4% of Asian youth reported higher-than-average levels of witnessing hate messaging online (Table 1).

Multivariate Moderated Multiple Regression (MMMR) Models

Three sets of nested MMMR models determined the moderation effects of (1) RE importance, (2) self-identified RE, and (3) online hate message exposure on the internalizing symptoms (i.e., depression and social anxiety). Univariate effects from the three final models were interpreted to examine the effects of our moderators of interest (see Supplemental Table S3 for nested model results).

For the moderation effects of levels of RE importance, omnibus multivariate effects are shown in Table 2. A significant two-way interaction between SM frequency and RE importance reveals adolescents with high RE importance show a stronger link between greater SM frequency and heightened depressive symptoms (ß = 0.60, p = 0.04, Adj. R² = 0.33), compared to those who report lower RE importance (Table 2). Univariate effects of this model also revealed a significant three-way interaction (ß = 0.68, p = 0.02, Adj. R² = 0.33) between RE importance, gender, and prosocial SM use associated with depressive symptoms (Figure 1). Adolescent females in this sample with lower levels of RE importance showed a significant link between greater prosocial SM use and heightened depressive symptoms (Figure 1a) compared to females with high RE importance and their male counterparts (Figure 1b).

Table 2.

Racial-Ethnic Importance and Gender Three-Way Multivariate Moderated Multiple Regression for Internalizing Behaviors

	MV	Depression			Online Social Anxiety
	F	ß	t	p	ß	t	p
Free or Reduced Lunch	0.77	−0.06	−0.28	0.78	0.04	1.14	0.26
COVID-19 Peer Relationship Quality	3.81*	−0.09	−2.76	0.01	0.00	−0.49	0.62
Gender	17.6***	−0.16	−4.42	<0.001	−0.03	−4.74	<0.001
Age	0.08	0.08	0.40	0.69	0.00	0.14	0.89
Race-Ethnicity	1.72		-	-
Black/African American	-	0.04	1.31	0.19	−0.01	−1.13	0.26
Asian/Asian American	-	0.21	0.96	0.34	−0.01	−0.84	0.40
Latine	-	−0.04	−1.12	0.27	−0.03	0.89	0.38
Other	-	−0.27	−1.32	0.19	0.00	−0.07	0.94
Multiracial	-	0.03	0.78	0.44	−0.01	−2.10	0.04
RE Importance	0.11	0.27	0.91	0.37	−0.02	−0.36	0.72
Hate Message Exposure	3.54*	0.53	2.57	0.01	−0.01	−0.19	0.85
SM Frequency	6.58**	−0.09	−1.75	0.08	0.03	3.79	<0.001
Online Harassment	73.4***	2.29	10.6	<0.001	0.29	7.84	<0.001
Prosocial SM use	19.1***	0.35	6.29	<0.001	0.03	3.37	<0.001
SM Frequency x RE Importance	2.36	0.60	2.09	0.04	−0.02	−0.34	0.73
SM Frequency x Gender	3.50*	0.03	0.59	0.56	−0.02	−2.22	0.03
RE Importance x Gender	3.35*	−0.35	−1.28	0.20	0.09	2.01	0.05
Prosocial SM Use x RE Importance	2.16	−0.13	−2.37	0.02	−0.02	−1.92	0.06
Prosocial SM Use x Gender	3.21*	−1.04	−3.22	0.001	−0.03	−0.53	0.60
SM Frequency x RE Importance x Gender	0.52	−0.05	−0.99	0.32	0.00	−0.39	0.70
Prosocial SM Use x RE Importance x Gender	2.49	0.68	2.20	0.02	0.04	0.77	0.44
		Adjusted R2 = 0.33			Adjusted R2 = 0.25

Note:

^*p<0.05;

^**p<0.01;

^***p<0.001.

Significant interactions of the final model are indicated in bold.

MV = Multivariate Omnibus Test; RE= Racial-Ethnic; SM= social media; Exp = Exposure; Free or Reduced Lunch = Family economic indicator

Figure 1.

Prosocial SM Use and Depression Association Moderated by Racial-Ethnic Importance and Gender

For the moderation effect of self-identified RE, only the RE by SM frequency interaction significantly contributed to the model (Table 3). Univariate effects revealed significant moderation effects of race-ethnicity by SM frequency on reported depressive symptoms (ß = 0.10, p = 0.01, Adj. R² = 0.33). Asian adolescents showed a greater SM frequency was associated with heightened depressive symptoms, compared to the non-Asian adolescent counterparts of the sample who showed a non-significant relation between SM frequency and depression (Figure 2a). In addition, a significant negative association between SM frequency and online social anxiety was found for Black adolescents (ß = −0.11, p = 0.031, Adj. R² = 0.24), whereas other RE subgroups had non-significant associations between SM frequency and online social anxiety (Figure 2b).

Table 3.

Race-Ethnicity Membership Multivariate Moderated Multiple Regression for Internalizing Behaviors

	MV	Depression			Online Social Anxiety
	F	ß	t	p	ß	t	p
Free or Reduced Lunch	1.08	−0.07	−0.33	0.74	0.04	1.35	0.18
COVID-19 Peer Relationship Quality	3.07*	−0.08	−2.48	0.01	0.00	−0.49	0.62
Gender	17.2***	−1.01	−4.96	<0.001	−0.14	−4.01	<0.001
Age	0.24	0.02	0.63	0.53	0.00	−0.16	0.87
Race-Ethnicity	1.65		-
Black/African American	-	0.05	1.50	0.13	−0.01	−1.12	0.26
Asian/Asian American	-	0.37	1.69	0.09	−0.03	−0.81	0.42
Latine	-	−0.02	−0.65	0.51	0.01	0.94	0.35
Other	-	−0.24	−1.13	0.26	0.00	−0.04	0.97
Multiracial	-	0.01	0.43	0.67	−0.01	−2.26	0.02
RE Importance	0.17	−0.04	−0.19	0.85	0.02	0.50	0.62
Hate Message Exposure	5.09**	0.11	3.18	0.002	000	0.30	0.76
SM Frequency	7.12***	−0.18	−0.60	0.55	0.19	3.59	<0.001
SM Online Harassment	73.8***	2.25	10.4	<0.001	0.31	8.14	<0.001
Prosocial SM use	16.7***	0.20	5.37	<0.001	0.02	3.13	0.002
SM Frequency x Race-Ethnicity	2.35**
SM Frequency x Black/African American	-	−0.09	−0.40	0.69	−0.11	−2.92	0.003
SM Frequency x Asian/Asian American	-	0.10	2.77	0.01	0.00	−0.69	0.49
SM Frequency x Latine	-	−0.25	−1.17	0.24	0.00	0.09	0.93
SM Frequency x Other	-	−0.01	−0.34	0.74	0.00	−0.42	0.67
SM Frequency x Multiracial	-	−0.33	−1.47	0.14	−0.03	−0.82	0.41
		Adjusted R2 = 0.33			Adjusted R2 = 0.24

Note:

^*p<0.05;

^**p<0.01;

^***p<0.001.

Significant interactions of the final model are indicated in bold.

MV = Multivariate Omnibus Test; RE= Racial-Ethnic; SM= social media; Exp = Exposure; Free or Reduced Lunch = Family economic indicator

Figure 2.

Social Media Frequency Associated with Internalizing Behaviors Moderated by Race-Ethnicity

The final model tested the moderation effects of online exposure to hate messaging. Two three-way interactions significantly contributed to the model. The first was between hate message exposure, gender, and online harassment on internalizing symptoms (Table 4). Univariate effects showed that adolescents with no exposure to online hate messaging reported a stronger association between greater online harassment and elevated depressive symptoms (ß = −0.70, p = 0.02, Adj. R² = 0.34), compared to those who report exposure to online hate messaging. The second model was between hate message exposure, gender, and prosocial SM use (Table 4). Results of this model also revealed that adolescents who do report exposure to online hate messaging have a stronger association between greater prosocial SM use and heightened online social anxiety (ß = 0.02, p = 0.02, Adj. R² = 0.24) (Figure 3).

Table 4.

Hate Message Exposure and Gender Three-Way Multivariate Moderated Multiple Regression for Internalizing Behaviors

	MV	Depression			Online Social Anxiety
	F	ß	t	p	ß	t	p
SES	0.97	−0.09	−0.47	0.64	0.04	1.21	0.23
COVID-19 Peer Relationship Quality	4.56*	−0.10	−3.01	0.002	−0.01	−0.82	0.42
Gender	17.3***	−0.92	−4.25	<0.001	−0.13	−3.52	<0.001
Age	0.05	−0.06	0.31	0.76	0.00	−0.06	0.96
Race-Ethnicity	1.78	-	-	-	-	-	-
Black/African American	-	0.05	1.43	0.15	−0.01	−1.28	0.20
Asian/Asian American	-	0.18	0.86	0.39	−0.02	−0.63	0.53
Latine	-	−0.03	−0.88	0.38	0.00	0.75	0.46
Other	-	−0.30	−1.42	0.16	0.00	−0.11	0.92
Multiracial	-	0.01	0.35	0.73	−0.02	−2.53	0.01
RE Importance	0.18	−0.06	−0.28	0.78	0.02	0.47	0.64
Hate Message Exposure	4.43*	0.14	2.81	0.01	−0.01	−0.81	0.42
SM Frequency	7.22***	−0.04	−1.32	0.19	−0.02	3.28	0.001
SM online harassment	78.4***	2.78	9.48	<0.001	0.36	6.94	<0.001
Prosocial SM use	19.5***	0.23	4.38	<0.001	0.02	1.54	0.12
Hate Message Exposure x Gender	1.06	−0.28	−0.97	0.33	0.00	0.07	0.94
SM Online Harassment X Hate Message Exp.	4.74 **	−0.70	−2.31	0.02	0.05	0.96	0.34
SM Online Harassment X Gender	2.96	−0.06	−1.23	0.21	−0.01	−1.54	0.12
Prosocial SM Use x Hate Message Exp.	2.42	0.09	1.79	0.07	0.02	2.39	0.02
Prosocial SM Use x Gender	1.68	−0.37	−1.18	0.23	0.04	0.65	0.52
SM Online Harassment x Hate Message x Gender	0.90	0.00	−0.03	0.97	−0.01	−1.33	0.18
Prosocial SM use x Hate Message x Gender	1.54	−0.37	−1.41	0.16	−0.06	−1.28	0.20
		Adjusted R2 = 0.34			Adjusted R2 = 0.24

Note:

^*p<0.05;

^**p<0.01;

^***p<0.001.

Significant interactions of the final model are indicated in bold.

MV = Multivariate Omnibus Test; RE= Racial-Ethnic; SM= social media; Exp = Exposure; Free or Reduced Lunch = Family economic indicator

Figure 3.

Associations with Internalizing Behaviors Moderated by Online Hate Message Exposure

Discussion

The role of RE identity in the digital age remains important to our understanding of adolescent development, alongside its implications for practitioner and educational awareness to support digital youth wellbeing. Our study was well-positioned to explore RE-based differential impacts on internalizing behaviors given unique social and environmental settings during the time of this study. Overall, we found significant descriptive differences in internalizing behaviors and online experiences across certain RE subgroups. Black adolescents reported higher levels of depression compared to the rest of the adolescent cohort when assessing group differences of internalizing behaviors. Asian and Multiracial adolescents report a significantly lower frequency of checking social media compared to their respective subgroup counterparts in the assessment of social technology use group differences.

Furthermore, the analysis revealed that adolescent girls with lower levels RE importance showed links between greater prosocial SM use (e.g., for school, activism, social connectedness) and heightened depressive symptoms. While this finding reflected greater psychological strain on prosocial behaviors online, prior work has shown similar results such that elevated prosocial behaviors are associated to heighten depressive symptoms among adolescent girls (Alarcón & Forbes, 2017) and adults more generally in the context of COVID-19 (McGuire et al., 2020). Gender differences in mental health and prosocial behaviors have also demonstrated that due to girls’ typical relational style (i.e., focus on maintaining harmonious relationships, caring about social evaluations), and there is often an emotional cost when engaging in interpersonal competence behaviors (Rudolph & Conley, 2005). Moreover, since girls are more likely to assign greater RE importance compared to boys (Charmaraman & Grossman, 2010), our finding that girls with lower RE importance struggle more with depression may be partly supported by the added complexities of race and gender (Crenshaw, 1989).

Additionally, we found that Asian adolescents, compared to their non-Asian counterparts, showed a robust association between greater SM frequency and heightened depressive symptoms. This aligns with prior research in the present time of COVID-19 and anti-Asian sentiments that have risen in social technology ecosystems, causing strain on wellbeing (Abidin & Zeng, 2020; Charmaraman et al., 2018). Furthermore, results also showed a significant effect for Black adolescents in this sample, such that an increase in time spent using social media was associated with a decrease in online social anxiety. Although opposing trends can be seen across other RE groups, finding aligned with prior work indicating that Black youth reported more enjoyment from social media as compared to their non-Black counterparts (Rideout et al., 2022). Additionally, the intersectional findings of race and gender in the current study is also supported by works of Crenshaw (1989) which contextualizes the Black female experiences and could be argued is mirrored into online spaces for underrepresented youth of today. Our findings related to low RE importance as a potential signal for heightened depressive symptoms and also increased prosocial behaviors among girls has implications for practitioners who may probe into possible indicators regarding their diverse clients’ depressive symptoms. This can range from deep-seated gender expectations for relational styles to family contexts that support or do not RE socialization and importance.

The absence of exposure to online hate messaging revealed an association between greater experiences to online harassment and heightened depressive symptoms compared to adolescents who had been more exposed hate messaging online. In contrast, youth exposed to online hate messaging showed greater positive associations between prosocial social media use and online social anxiety, which may be indicative of increased stress related to civic engagement and empowerment that social technologies afford young people today (see Charmaraman, Hernandez, & Hodes, 2022 for overview). The result of this particular finding warrants further exploration of why such exposure to, or lack thereof, online hate messaging could help or hinder adolescents’ emotion regulation based on one’s identity.

Limitations of Current Study

While met with many strengths, this study is not without limitations. The cross-sectional nature of this exploratory analysis limits our interpretations for cause and effect. Sample size remains a factor of limited power and interpretability of results particularly for racial-ethnic groups with traditionally lower representation in the field. Addressing RE identity more explicitly in the future can help to understand how it is translated online. The analyses conducted were exploratory by nature, but analyses can be expanded upon by measurements of offline environment, cultural influences in the home, and school experiences in future directions. Lastly, this dataset was also collected during the COVID-19 pandemic, during which we have found a rise in depression and social anxiety in our longitudinal sample relative to pre-COVID-19 waves (Charmaraman et al., 2022b). Therefore, there may be other confounds beyond the indicators included in this study that affect mental health outcomes and social technology use during the pandemic (i.e., shelter in place, hybrid virtual learning, etc.).

Implications for Racial-Ethnic Identity in the Digital Age

Despite the risks that the digital ecosystem has posed on RE identity formation of adolescents, there are growing opportunities to support this identity formation related to empowerment of youth perspectives which can address the ultimate goal of dismantling race-based disparities in the US. As the predominant users of social media, RE minority adolescents have the opportunity to gain online social capital and find ways to support wellbeing which can further support positive development (Kawachi et al., 2004; Raymond-Flesch et al., 2017).

We believe the implications of this work can increase the visibility of this topic for RE minority youth and their unique and diverse experiences. For future studies, oversampling and asking questions about RE identity development as they are rapidly adopting social technology may support the visibility and application of racial-ethnic and intersecting identities to contemporary and persistent concerns (Burton et al., 2010; Brown, 2003, Crenshaw, 1989). This will also support a better understanding and application for practitioners and key stakeholders of adolescent experiences to provide continued support for mental health. Additionally, examining the widespread adoption and acceptance of these social technologies in youth of color without a deficit lens is a step towards dismantling historic and ongoing oppression of, for example, Black youth by creating a more equitable and less discriminatory space for youth of color (Crenshaw et al., 1995). As adolescents explore their race and ethnicity and other identities such as gender and sexuality, we urge researchers and practitioners to implement the intersectionality of these social identities when understanding the wellbeing of minority youth on social media (Benner et al., 2018; Burton et al., 2010; Crenshaw, 1989). The mechanisms of engagement on social technologies among adolescents will not only allow us to further identify potential hinderances to psychosocial outcomes, but also leverage opportunities of growth and support for identity formation which may serve as long-term protective factors (Phinney et al., 1997).

Conclusion

We live in an increasingly diverse society and social technologies play an important role in how adolescents connect to their RE identity. Investigating RE identity in the digital age has an important role in bridging the movement towards the norm of a multicultural youth society as they become increasingly diverse and the future majority of the population. Prioritizing one’s RE identity has been shown as a critical indicator of adolescent mental health and resilience (Kiang et al., 2006; Phinney et al., 1997). Sociocultural and environmental contexts, too, play a vital role in development which is often missed in studies around social technology use.

Prior research has indicated that adolescents of color who strongly identify with RE groups show positive psychosocial outcomes, but this does not erase the obstacles of discrimination and hate that many are exposed to in the digital age. We showed that social technologies have in many ways extended discriminatory content onto another ecosystem but also afforded an opportunity for many adolescents to be exposed to, process, and then to resist the negativity perpetuated by society likely driven by their heighten self-awareness during this period of development. This study recognizes both risks and protective roles that racial-ethnic identity plays in an increasingly psychologically taxing period for adolescents of color. Social technologies can provide opportunities for youth of marginalized communities to increase their social capital, empowerment, and connectedness, and support their wellbeing in the online discourse of race-ethnicity, thereby reducing the power held by oppressors (e.g., online harassers, systemic barriers). However, we cannot ignore the negative indicators of online behaviors such as harassment, but future approaches utilizing mixed methodologies (e.g., quantity of time vs. quality of content) and analyzing sociocultural and environmental factors may help us better understand how the use of social technologies can be further leveraged to support the development of a diverse adolescent population.

Public Significance Statement.

The present study suggests that group differential associations among racial-ethnic groups and identity importance may have a significant role in social media experiences and mental health outcomes among diverse groups of adolescents. Additionally, the findings of this study highlight both opportunities and hinderances that social media spaces afford racially and ethnically diverse groups during a critical period in psychosocial and identity development.