Peer Sexual Harassment, Affect, and Substance Use: Daily Level Associations among Adolescents

Abstract

Introduction:

Peer sexual harassment is associated with adolescent substance use at the global level; however, it is unknown whether substance use occurs proximal in time to the sexual harassment experience. This study used daily reports to examine the proximal relations between sexual harassment victimization and affect and substance use. Based on theories of self-medication, we hypothesized that negative affect and substance use (cigarettes, electronic cigarettes, alcohol, marijuana) would be higher than typical on days when sexual harassment occurred relative to non-victimization days.

Method:

A community sample of 13 – 16-year-old adolescents (N = 204, 55.4% female) from a metropolitan area in the northeastern United States completed 56 days of online reports assessing experiences with peer sexual harassment, substance use (cigarettes, electronic cigarettes, alcohol, and marijuana), and positive and negative affect.

Results:

Multilevel modeling revealed that experiencing sexual harassment on a given day was associated with higher than typical negative affect on that day, relative to non-victimization days. The likelihood of cigarette and alcohol use (but not electronic cigarettes, marijuana, or negative or positive affect) was greater on days when sexual harassment occurred.

Conclusion:

Sexual harassment victimization is proximally associated with negative affect and alcohol and cigarette use, suggesting that adolescents may be using substances to cope with sexual harassment victimization. The co-occurrence of sexual harassment with negative affect and substance use points to the need for prevention efforts that conjointly address sexual harassment victimization, coping, and substance use.

Sexual harassment, defined as unwanted and unwelcome sexual behavior that interferes with one’s life (American Association of University Women; AAUW, 2001), is a form of peer aggression that adolescents frequently experience. Like bullying, peer sexual harassment may be verbal, physical, relational, or cyber; however, with sexual harassment, the aggression targets an individual’s sexuality, gender, or sexual orientation (Espelage et al., 2016; Felix & McMahon, 2006). Examples include spreading sexual rumors, making sexual jokes or comments, touching someone in an unwelcome sexual way, or homophobic name-calling (Espelage et al., 2018; Hill & Kearl, 2011). Whereas bullying is highly prevalent in middle school, in high school, peer aggression often becomes more gendered and sexual, taking the form of sexual harassment (Cunningham et al., 2010; Espelage, Basile, et al., 2012; Pellegrini, 2002). Prevalence rates of sexual harassment are alarmingly high, with 30% to 81% of middle and high school students reporting such experiences (AAUW, 2001; Clear et al., 2014; Hill & Kearl, 2011). Both boys and girls experience sexual harassment, although girls experience it at higher rates and report more negative consequences, including depressive symptoms and substance use (Felix & McMahon, 2006; Gruber & Fineran, 2008).

Sexual harassment can be particularly damaging to adolescents because it attacks an individual’s sexual identity during a period in development characterized by heightened sensitivity to peer approval and rejection (Steinberg, 2014). Indeed, sexual harassment has been associated with negative affect and depressive symptoms (i.e., sadness, anger), self-harm, substance use, and feeling unsafe at school in cross-sectional (Bucchianeri et al., 2014; Clear et al., 2014) and longitudinal studies (Chiodo et al., 2009; Dahlqvist et al., 2016; Goldstein et al., 2007; Wolff et al., 2017). Of concern is that adolescents may turn to substances as a means of coping with emotional distress from sexual harassment, placing them at risk for earlier initiation of substance use, as well as more frequent or heavier use (Kuntsche et al., 2005). Early and risky substance use in turn can trigger a cascade of short- and long-term negative developmental and health outcomes, including depression, substance use disorders, sleep disturbances, high-risk sexual behavior, and involvement in dating and sexual violence (Badour et al., 2020; Dodge et al., 2009; King & Chassin, 2007; Kwon et al., 2020; Marshall, 2014; Temple et al., 2013; Windle et al., 2008). Although the research literature has established robust associations linking peer sexual harassment to depressive symptoms and substance use, whether these associations occur proximal to the victimization experience or develop over time is unknown. If both negative affect and substance use occur close in time to the harassment incident, it would point to the need for early joint intervention to address both coping and substance use. However, if substance use and depressive symptoms do not arise until much later, it would suggest that there are other mechanisms through which sexual harassment contributes to depression and/or substance use.

Proximal Associations between Sexual Harassment, Negative Affect, and Substance Use

One way that sexual harassment victimization may be proximally associated with substance use is via self-medication, whereby individuals use substances to cope with negative affect and emotional distress resulting from sexual harassment experiences (Khantzian & Albanese, 2008). Adolescents must grapple with managing intense emotions arising from victimization at a time when their self-regulatory capacity, including their ability to regulate their emotions, is immature, leaving them vulnerable to the development of maladaptive coping responses (Brooks-Gunn et al., 1994; Mischel et al., 2014). Adolescents who lack effective strategies for regulating negative emotions may use substances to cope. Indeed, coping has been identified as an important motive for the use of alcohol (Gottfredson & Hussong, 2013; Kuntsche et al., 2007; Meisel et al., 2018), marijuana (Vilhena-Churchill & Goldstein, 2014), and cigarettes (McKee et al., 2020; Mischel et al., 2014; Piko et al., 2015) among adolescents and young adults.

There is support for the self-medication theory in the adolescent bullying literature. Cross-sectional studies with large samples of high school students have shown that internalizing symptoms (e.g., depressive symptoms) mediate the association between bullying victimization and substance use, including use of cigarettes, marijuana, and alcohol (Hong et al., 2019; Lambe & Craig, 2017; Luk et al., 2010). Longitudinal studies also have linked experiencing bully victimization in middle school to high school substance use, as mediated via depression and suicidal ideation (Earnshaw et al., 2017; Marschall-Levesque et al., 2017). Other research suggests that coping motives are the mechanism through which peer victimization and substance use are linked (Meisel et al., 2018; Topper et al., 2011). For example, in a sample of late adolescents, chronic peer victimization was linked to alcohol use via internalizing symptoms but only among those adolescents who drank for coping motives; in the absence of coping motives, internalizing was protective against alcohol use (Meisel et al., 2018).

Although the self-medication hypothesis has not been tested with adolescent victims of sexual harassment, findings from studies of sexual harassment and substance use among adults support this hypothesis. In a military sample, Gradus and colleagues (2008) found that depression mediated the relationship between sexual harassment and binge drinking among women but not men. Another study showed that the associations between workplace sexual harassment and drinking outcomes were partially mediated by distress (anxiety, depression, hostility) for both men and women (Richman et al., 2002).

Alternatively, sexual harassment victimization may occur within the context of engaging in delinquent activities. Consistent with problem behavior theory, adolescent problem behaviors such as aggression, substance use, and risky sexual behavior tend to cluster together and occur among the same individuals (Jessor, 1987). Because substance use is illicit for adolescents, it typically occurs in unsupervised contexts in the presence of other delinquent and substance-using peers (Finan & Lipperman-Kreda, 2020; Lipperman-Kreda et al., 2015). There is a robust association between delinquency and risky sexual behavior (Beaver et al., 2016), as well as an expectation for sexual activity to occur in contexts where alcohol is consumed, even among adolescents (Lindgren et al., 2009; Livingston et al., 2013). Several studies have shown that involvement in delinquent activities, including using illegal or age-inappropriate substances and affiliating with delinquent peers, increases the likelihood of sexual harassment victimization (Fineran & Bolen, 2006; Goldstein et al., 2007; Tillyer et al., 2016).

Using Daily Report Studies to Examine Acute Effects of Sexual Harassment

Daily report studies can provide valuable data with which to examine the same-day and next-day associations among sexual harassment, affect, and substance use. Daily report methodology can also allow for the examination of within-person changes in response to sexual harassment on a daily basis (Shorey et al., 2016). That is, it is possible to determine whether changes in affect or substance use patterns differ for an individual depending on whether or not they experience sexual harassment that day or the previous day. A few daily-level studies have examined same-day associations with bully victimization. These studies have established that on days when adolescents are bullied, they are also more likely to report negative affect, including sadness, anger, humiliation, and worry, later that day (Jiang et al., 2019; Morrow et al., 2014; Nishina, 2012). Using the present sample, Livingston et al. (2019) extended this work by considering the proximal associations of bullying with negative affect and substance use. Results showed that on days when adolescents were bullied, they were more likely to report negative affect and cigarette use than on non-victimization days. This finding provides important evidence linking negative affect and substance use close in time to peer victimization, supporting self-medication theory. However, the temporal ordering of victimization and cigarette use within days could not be determined.

A few daily diary studies that have examined the daily level associations between substance use and interpersonal violence in college samples have shown that substance use occurring after victimization incidents can even carry into the next day. For example, among a sample of college women, Parks et al. (2008) found that the odds of alcohol use were three times higher in the 24 hours following verbal (but not physical or sexual) victimization, and that negative affect increased and positive affect decreased the day following verbal or physical aggression. Shorey et al. (2016) found that experiencing physical or sexual violence was associated with next-day cannabis (but not alcohol) use among college students. These findings provide further support for self-medication and point to the importance of considering effects from the following day as well as same-day effects.

The Current Study and Hypotheses

The goal of the current study was to examine the proximal associations among sexual harassment victimization, negative affect, and substance use to determine whether they occur close in time, and if so, whether they are supportive of the self-medication or delinquent context explanations. The current study examined the daily-level associations among sexual harassment victimization, affect, and substance use in a community sample of 13- to 16-year-old adolescents who had previously reported peer victimization (i.e., bullying or sexual harassment), using daily diary report data collected over 56 consecutive days (8 weeks). Consistent with self-medication theory (Khantzian & Albanese, 2008), we first hypothesized that experiencing sexual harassment on a given day would predict greater negative affect for an individual on that day compared with that person’s typical negative affect. Similarly, we expected that positive affect would be lower than typical for that person on sexual harassment days. In contrast, if the relationship between sexual harassment victimization and substance use is due to spending time in high-risk contexts, we would not expect to see daily level associations with negative affect. Second, we hypothesized that individuals would be more likely to report substance use on days when sexual harassment victimization occurred than on days without sexual harassment. We considered use of cigarettes, electronic cigarettes (e-cigarettes), alcohol, and marijuana since these are the substances most often used by adolescents in this age range (Johnston et al., 2020). We also examined the lagged effects to determine whether experiencing sexual harassment was related to negative affect or substance use the following day. Consistent with prior research that showed lingering effects of victimization (Parks et al., 2008; Shorey et al., 2016), we anticipated that negative affect would be higher than typical, positive affect would be lower than typical, and substance use would be more likely to occur on the day following sexual harassment victimization.

Method

Participants and Recruitment

Participants (N = 204, 55.4% female) were a community sample of adolescents between 13 and 16 years of age (M = 14.51, SD = 0.85), residing in a metropolitan county in Western New York State. Approximately 80.8% of participants self-identified as White, 12.8% as Black, 2.5% as multiracial, 1.0% as Native American, and 3.0% as Other. Hispanics/Latinos comprised 8.4% of the sample. This racial/ethnic composition is very similar to the 2019 demographics for the county as a whole (79.3% White, 14.0% Black). Median household income was $40,000 - $79,999 (as reported by mother), consistent with the median income for the county from which the sample was drawn ($58,000; U.S. Census Bureau, n.d.). Most of the participants were in 8^th (28.4%), 9^th (37.7%), or 10^th (25.0%) grades.

Participants in the daily study were drawn from a larger sample of adolescents (N = 800) enrolled in a longitudinal survey study of adolescent health and social relationships. Adolescents were recruited to participate in the longitudinal survey study via address-based sampling. To be eligible for the survey study, adolescents had to be 13 – 15 years of age at recruitment, be enrolled in a public or private school (i.e., not homeschooled), speak, and read English at a minimum 6^th-grade proficiency level, and be living with a mother or female guardian who was also willing to participate in the study. Eligible participants completed a separate web-based baseline survey that included assessments of childhood victimization, bullying, and peer sexual harassment.

Adolescents who self-reported being bullied or sexually harassed by a peer at least once a month in the past six months on the baseline survey were contacted by phone and invited to participate in the daily report study of peer victimization (i.e., bullying and sexual harassment) and substance use. Participants had to be recruited to the daily report study within 90 days of completing their baseline survey to increase the likelihood of capturing current and ongoing victimization. Recruitment initiated in January, 2015 and continued through early May, 2016; because peer victimization is most likely to occur during the school year, participants were enrolled in the daily study between the months of September and May. Those who met eligibility criteria over the summer months were delayed until September. A total of 345 (43.1%) participants from the larger sample were eligible to participate in the daily report study, which was presented as separate but related to the main study. Of these, 110 were not recruited within the 90-day window due to delays in the start-up of the daily report study, 20 refused participation, and 11 agreed to participate but failed to complete any daily reports, leaving a final sample of 204 participants. Verbal parental consent and electronic adolescent assent were obtained and a secure link to access the daily reports was sent to each of the adolescent participants. Procedures for both the longitudinal and the daily report studies were approved by the University Institutional Review Board.

Using baseline data, we compared youth who were eligible for the daily report study to youth who were not eligible. Girls were more likely than boys to experience peer victimization and hence to be eligible for the daily study (63.9% girls vs. 36.1% boys), χ²(1, 800) = 10.05, p = .001. Those who were eligible for the daily report study were more likely than those who were ineligible to report childhood violence (M = 0.40, SD = 0.39 vs. M = 0.22, SD = 0.28), t(798) = −7.76, p < .001, and to have used cigarettes (M = 0.16, SD = 0.81 vs. M = 0.04, SD = 0.29), t(797) = −3.06, p = .002, e-cigarettes (vape pen) (M = 0.42, SD = 1.21 vs. M = 0.26, SD = 0.93), t(796) = −2.16, p = .031, and have been drunk (M = 0.23, SD = 0.75 vs. M = 0.12, SD = 0.47), t(795) = −2.53, p = .012 in the past six months. There were no differences in age or marijuana use during the last six months. Eligible youth who participated did not differ from eligible your who did not participate on sexual harassment victimization, childhood experience of violence, age, cigarettes, e-cigarettes, alcohol, or marijuana use.

Procedure

Participants completed a web-based, five-minute daily report survey for 56 consecutive days between the hours of 3:00 PM and 11:59 PM using a computer, tablet, or smartphone device. Participants received $0.50 each day for each completed report plus a $10 bonus for each week when they completed all seven reports, or a partial bonus ($7) for each week that they completed five or six reports. At the end of the 8-week period, participants received an additional $20 bonus for completing a total of 51, 52, or 53 reports (n = 35/204), or a $25 bonus for having completed 54, 55, or 56 reports (n = 82/204). If participants missed a reporting day, they were able to complete an abbreviated make-up report the following day (e.g., affect was not assessed retrospectively). The maximum possible payment amount, including bonuses for complete data, was $133. Payments were made by checks, which were mailed to participants every two weeks.

Measures

Positive and Negative Affect

Current positive and negative affect was measured on each day using the two higher-order scales of the Positive and Negative Affect Schedule - Expanded Form (PANAS - X; Watson & Clark, 1994; Watson et al., 1988). The Positive Affect scale consisted of five items (i.e., happy, energetic, active, lively, hopeful), as did the Negative Affect Scale (i.e., gloomy, irritable, anxious, sad, angry), measured on 5-point scales from 0 = Not at all through 4 = Extremely. Based on exploratory factor analysis we included positive affect (active, lively, energetic; α = 0.89) and two subscales of negative affect— sadness (sad, gloomy; r = 0.68) and anger (angry, irritable; r = 0.60). Within report, positive and negative affect composites were modestly negatively correlated (positive affect correlated, r = −0.17 with sadness, and r = −0.14 with anger). Sadness and anger were positively correlated (r = 0.68) within report.

Sexual Harassment

Sexual harassment victimization was assessed using a six-item modified version of the American Association of University Women survey (AAUW, 2001; Hill & Kearl, 2011). Participants indicated whether they experienced each of the following behaviors today (0 = No, 1 = Yes): someone made unwelcome sexual jokes, comments, or gestures (e.g., calling you names like “slut,” “ho”); spread sexual gossip or rumors; touched you in an unwelcome sexual way; showed you sexual pictures you did not want to see; sent sexual pictures of you that you did not want others to see; or physically intimidated or threatened you in a sexual way. Given the relatively low frequency of these experiences over the 56-days, the six victimization items were summed and then dichotomized to create a summary sexual harassment victimization score (0 = No sexual harassment victimization; 1 = Sexual harassment victimization) for each day for each participant.

Substance use

In each report, participants indicated whether they had used each of the following substances today: cigarettes, e-cigarettes, alcohol, and marijuana (0 = No, 1 = Yes). If a substance was used, participants were also asked to report on the amount or frequency of usage that day (i.e., number of cigarettes smoked, number of times e-cigarette used, number of drinks, number of times smoked marijuana), rated on scales from 1 to more than 10.

Childhood experiences of violence

As part of the baseline survey, adolescents’ exposure to childhood victimization was measured using the Childhood Experiences of Violence Questionnaire (CEVQ; Walsh et al., 2008). Two items assessed witnessing adult aggression toward other adults (i.e., interparental violence), eight items assessed adult aggression toward the child (e.g., slapped; grabbed or shoved; kicked, bit, or punched), and six items assessed child sexual abuse (e.g., showed private parts; threatened to have sex). All items were rated on a 5-point scale from 0 (Never) through 4 (More than 10 times). An average score was computed over these sixteen items to create a childhood victimization variable for each participant. Internal consistency for the CEVQ scale was acceptable (α = 0.77). Childhood victimization (e.g., interparental violence, childhood maltreatment) was included as a covariate because of its potential impact on the key variables of interest (e.g., Espelage, Low, et al., 2012; Kristman-Valente et al., 2013; Shin et al., 2009).

Demographics

Demographic information was assessed as part of the baseline survey and included age, year in school, gender, race, and ethnicity. Mothers reported family income.

Data Analyses

We conducted mixed-effects multilevel modeling using Bayesian estimation within Mplus Version 8.2 (Gelman et al., 2014; Muthén & Muthén, 2017). The Bayes estimator in Mplus uses the probit link function and is appropriate for use with relatively rare events. Bayesian estimation neither requires nor assumes that parameters are normally distributed (Muthén et al., 2016; van de Schoot et al., 2014). Bayesian estimation uses a full-information approach to missing data, like full information likelihood estimation (FIML). In the current study, data included 11,424 rows for analysis (204 individuals x 56 days). At the daily level (Level 1), we included as predictors each participant’s report of sexual harassment victimization on that day (i.e., today’s sexual harassment victimization) and on the prior day (i.e., yesterday’s sexual harassment victimization). These categorical (0/1) victimization variables were left uncentered. We also included weekend (uncentered, coded as 1 for Saturday – Sunday and 0 for Monday – Friday) and day of the study (1 – 56, grand mean centered) at the daily level to control for potential unmeasured temporal confounds.

At the individual level (Level 2), we controlled for each participant’s number of sexual harassment victimization days and mean affect (or number of substance use occasions) over the study days, allowing us to account for between-person effects and to distinguish within-person effects from between-person effects (Enders & Tofighi, 2007). At the individual level, we also included gender (0 = Male; 1 = Female), age, and childhood victimization. All individual level variables (except for gender, uncentered) were grand mean centered. In addition, we examined whether gender altered any of the associations by testing the cross-level Gender X Daily Sexual Harassment interaction and also the individual level Gender X Total Number of Sexual Harassment Reports interaction, but we offered no a priori hypotheses for direction.

Results

Descriptive Data

Compliance with the daily reports was very good; out of a total possible 11,424 reports (204 individuals x 56 days), 9,600 reports (80.4%) were completed (M = 49.82, SD = 8.52, range = 8 – 56). Of these, 7,825/9,600 (81.5%) were completed on time, and 1,775/9,600 (18.5%) were make-ups. On average, reports were completed at 19.57 (SD = 2.97; Median = 19.00) hours on a 1 – 24-hour scale, or roughly 8:00 PM. Time completed and completion rates were virtually identical for male and female participants. Within each individual’s report, average positive affect (M = 1.78, SD = 1.14, range: 0 – 4), sadness (M = 0.72, SD = 0.92, range: 0 – 4), and anger (M = 0.81, SD = 0.95, range: 0 – 4) were low. However, girls reported lower self-perceived current positive affect and higher negative affect compared to boys (see Table 1).

Table 1.

Sexual Harassment, Positive and Negative Affect, and Substance Use

Variable	Males (N = 4,283)	Females (N = 5,317)	Gender difference, t-test or χ2
Time completed (1–24 Hour), M (SD)	19.54 (2.85)	19.64 (3.07)	t(9,598) = −1.623
Gender, N (%)	91 (44.6)	113 (55.4)	--
Completed daily reports, M (SD)	49.85 (8.56)	49.79 (8.50)	t(9,598) = 0.375
Sexual harassment victimization episodes, N (%)	28 (0.7)	103 (1.9)	χ2(1, 9,562) = 29.081***
Positive affect a, M (SD)	1.85 (1.08)	1.73 (1.19)	t(7,806) = 4.644***
Sadness b, M (SD)	0.60 (0.80)	0.82 (0.99)	t(7,806) = −10.470***
Anger c, M (SD)	0.70 (0.87)	0.89 (1.00)	t(7,807) = −8.920***
Childhood victimization, M (SD)	0.36 (0.30)	0.47 (0.47)	t(202) = −2.031
Age, M (SD)	14.48 (0.88)	14.54 (0.83)	t(202) = −0.471
Cigarette use days, N (%)	72 (1.7)	31 (0.6)	χ2(1, 9,523) = 26.775***
Number of cigarettes smoked, M (SD)	3.24 (2.59)	2.84 (1.92)	t(101) = 0.864
E-cigarette use days, N (%)	72 (1.7)	139 (2.6)	χ2(1, 9,521) = 9.786**
Number of times of e-cigarette used, M (SD)	5.42 (3.73)	4.32 (3.52)	t(209) = 2.094*
Alcohol use days, N (%)	35 (0.8)	90 (1.7)	χ2(1, 9,519) = 14.256***
Number of drinks, M (SD)	2.06 (1.39)	2.70 (2.47)	t(123) = −1.834
Marijuana use days, N (%)	60 (1.4)	52 (1.0)	χ2(1, 9,521) = 3.607
Number of occasions marijuana smoked, M (SD)	2.29 (1.82)	1.52 (1.13)	t(109) = 2.632**

Note.

^aAverage of active, lively, and energetic.

^bAverage of sad and gloomy.

^cAverage of angry and irritable.

^*p < .05.

^**p < .01.

^***p < .001.

Numbers were computed on 9,600 rows of data.

Across the 56 days of reporting by 204 individuals (11,424 total reporting days), adolescent participants reported a total of 103 cigarette smoking days (M = 0.50; SD = 3.51), 211 e-cigarette use days (M = 1.03; SD = 4.13), 125 drinking days (M = 0.61; SD = 1.78), and 112 marijuana use days (M = 0.55; SD = 2.13). Sexual harassment victimization was reported on 131 days (M = 0.64; SD = 1.62). On average, male participants smoked cigarettes on more days than female participants (see Table 1). However, female participants used e-cigarettes and alcohol on more days than male participants, although male participants reported higher frequency of e-cigarette use on a given use day than female participants. There was no difference in marijuana use days by gender, but male participants smoked marijuana on more occasions than female participants. Overall, rates of substance use and sexual harassment were low, with a substantial proportion (95%) reporting no sexual harassment or substance use during the study.

Table 2 provides comparisons between participants with and without sexual harassment experiences over the course of the study. Compared with those who did not report sexual harassment (n = 151), those who experienced at least one experience of sexual harassment over the study period (n = 53) were more likely to be girls, more likely to report childhood victimization, had higher negative affect (sadness), and were more likely to report e-cigarette use. There were no other differences observed.

Table 2.

Comparison of Participants With and Without Sexual Harassment (N = 204)

Variable	No sexual harassment (N = 151)	Sexual harassment (N = 53)	t-test or χ2
Gender, female participant N (%)	73 (48.3)	40 (75.3)	χ2(1, 204) = 11.684***
Childhood victimization, M (SD)	0.38 (0.35)	0.52 (0.52)	t(202) = −2.126*
Age, M (SD)	14.57 (0.83)	14.34 (0.88)	t(202) = 1.650
Positive affect a, M (SD)	1.81 (0.80)	1.73 (0.79)	t(202) = 0.673
Sadness b, M (SD)	0.68 (0.57)	0.93 (0.64)	t(202) = −2.640**
Anger c, M (SD)	0.79 (0.62)	0.96 (0.66)	t(202) = −1.698
Cigarette use days, M (SD)	0.37 (2.57)	0.89 (5.39)	t(202) = −0.671
Number of cigarettes smoked, M (SD)	3.15 (2.52)	2.93 (1.59)	t(101) = 0.433
E-cigarette use days, M (SD)	0.68 (2.75)	2.04 (6.74)	t(202) = −2.041*
Number of times of e-cigarette used, M (SD)	4.77 (3.68)	3.71 (2.70)	t(209) = 1.370
Alcohol use days, M (SD)	0.48 (1.45)	1.00 (2.49)	t(202) = −1.447
Number of drinks, M (SD)	2.44 (2.18)	3.40 (2.72)	t(123) = −1.084
Marijuana use days, M (SD)	0.58 (2.35)	0.45 (1.38)	t(202) = 0.482
Number of occasions marijuana smoked, M (SD)	1.94 (1.59)	1.50 (1.00)	t(109) = 0.848

Note.

^aAverage of active, lively, and energetic.

^bAverage of sad and gloomy.

^cAverage of angry and irritable.

^*p < .05.

^**p < .01.

^***p < .001.

Does Sexual Harassment Victimization Predict Positive or Negative Affect?

We considered the effects of sexual harassment victimization occurring earlier that day or on the prior day on positive affect, sadness, and anger experienced at the time of the report. As shown in Table 3, there were significant effects of same-day (i.e., earlier that day) sexual harassment victimization on negative affect (sadness, but not anger) reported later that day. We did not observe significant effects of sexual harassment on positive affect. Prior day sexual harassment was unrelated to today’s positive and negative affect. Participants experienced higher levels of negative affect on weekdays relative to weekends.

Table 3.

Positive and Negative Affect as a Function of Daily Sexual Harassment

Variable	Positive affect		Sadness		Anger
	Estimate (S.D.)	95% CI	Estimate (S.D.)	95% CI	Estimate (S.D.)	95% CI
Intercept	1.781 (0.014)***	[1.751, 1.806]	0.753 (0.013)***	[0.727, 0.776]	0.845 (0.013)***	[0.818, 0.868]
Level 1
Sexual harassment	−0.099 (0.083)	[−0.263, 0.063]	0.165 (0.074)*	[0.020, 0.310]	0.146 (0.075)	[−0.001, 0.294]
Prior day sexual harassment	−0.133 (0.081)	[−0.293, 0.027]	0.043 (0.072)	[−0.099, 0.186]	0.098 (0.074)	[−0.047, 0.244]
Weekend vs. weekday a	0.035 (0.020)	[−0.002, 0.074]	−0.037 (0.018)*	[−0.071, −0.002]	−0.051 (0.018)**	[−0.085, −0.016]
Day of the study b	−0.003 (0.001)***	[−0.004, −0.002]	0.002 (0.001)***	[0.001, 0.003]	0.002 (0.001)***	[0.001, 0.003]
Level 2
Gender	−0.005 (0.019)	[−0.042, 0.030]	−0.006 (0.017)	[−0.040, 0.026]	−0.004 (0.018)	[−0.038, 0.029]
Age	0.003 (0.010)	[−0.017, 0.022]	−0.001 (0.009)	[−0.019, 0.017]	0.002 (0.009)	[−0.016, 0.019]
Childhood victimization	0.002 (0.026)	[−0.047, 0.057]	0.001 (0.024)	[−0.045, 0.052]	0.007 (0.024)	[−0.036, 0.056]
Total sexual harassment over 56 days	0.004 (0.006)	[−0.007, 0.017]	−0.004 (0.005)	[−0.014, 0.007]	−0.005 (0.005)	[−0.015, 0.006]
Mean of DV over 56 days	1.006 (0.011)***	[0.985, 1.027]	1.011 (0.015)***	[0.981, 1.040]	1.011 (0.014)***	[0.982, 1.038]

Note.

^aWeekend (0 = Monday to Friday; 1 = Saturday to Sunday).

^bDay of the study (1–56).

^*p < .05.

^**p < .01.

^***p < .001.

At the individual level (Level 2), there were robust effects of mean positive and negative affect on daily affect, as expected. We also considered the between-person (Level 2) effects of number of sexual harassment victimization days but found no significant effects on positive affect, anger, or sadness. To examine gender differences, we entered the cross-level Gender X Sexual Harassment Victimization interaction and the individual level Gender X Total Number of Sexual Harassment Reports interaction and re-ran the models in Table 3. The pattern of results was the same, and no interaction effects were significant (hence not displayed), indicating that effects of sexual harassment on affect were comparable for male and female participants.

Is Sexual Harassment Victimization Proximally Related to Substance Use?

To examine whether sexual harassment victimization was positively associated with same day substance use, we first combined the substance use items (cigarette, e-cigarette, alcohol, and marijuana use) to create a dichotomous, any substance use (n = 419; M = 2.05; SD = 5.82; versus no use) variable. As displayed in Table 4, there were positive within-day associations between sexual harassment victimization and any substance use, indicating that participants were more likely to report substance use on days when sexual harassment victimization occurred. Prior-day sexual harassment was not a significant predictor of current-day substance use. Adolescents were more likely to use substances on weekends relative to weekdays. Older adolescents were more likely to use substances than younger adolescents. The interaction term effects between sexual harassment and gender were not significant (not displayed).

Table 4.

Substance Use as a Function of Daily Sexual Harassment

Variable	Any use
	Estimate (S.D.)	95% CI
Intercept	2.792 (0.135)***	[2.565, 3.092]
Level 1
Sexual harassment	0.691 (0.209)**	[0.277, 1.098]
Prior day sexual harassment	0.176 (0.206)	[−0.237, 0.572]
Weekend vs. weekday a	0.265 (0.066)***	[0.137, 0.397]
Day of the study b	−0.006 (0.002)**	[−0.010, −0.002]
Level 2
Gender	0.163 (0.120)	[−0.064, 0.410]
Age	0.202 (0.075)**	[0.058, 0.351]
Childhood victimization	−0.042 (0.136)	[−0.313, 0.221]
Total sexual harassment over 56 days	−0.086 (0.037)*	[−0.163, −0.016]
Total of any substance use over 56 days	0.126 (0.010)***	[0.109, 0.149]

Note.

^aWeekend (0 = Monday to Friday; 1 = Saturday to Sunday).

^bDay of the study (1–56).

^*p < .05.

^**p < .01.

^***p < .001.

Next, we examined the within-day associations between sexual harassment and each type of substance use to determine whether effects were specific to a particular substance or substances. We used the same daily level and individual variables as in Table 4. As shown in Table 5, there were positive within-day associations between sexual harassment victimization and cigarette and alcohol use (but not e-cigarette use or marijuana use), indicating that adolescents were more likely to use cigarettes and alcohol on days when sexual harassment occurred than on non-victimization days. Prior day sexual harassment was a significant predictor of e-cigarette use (but not cigarette use, alcohol use, or marijuana use). Participants were more likely to use alcohol and marijuana on weekends relative to weekdays.

Table 5.

Use of Different Substances as a Function of Daily Sexual Harassment

Variable	Cigarette use		Electronic cigarette use		Alcohol use		Marijuana use
	Estimate (S.D.)	95% CI	Estimate (S.D.)	95% CI	Estimate (S.D.)	95% CI	Estimate (S.D.)	95% CI
Intercept	4.560 (0.452)***	[3.761, 5.477]	3.087 (0.189)***	[2.787, 3.536]	3.100 (0.159)***	[2.842, 3.449]	3.477 (0.226)***	[3.114, 4.058]
Level 1
Sexual harassment	0.896 (0.360)*	[0.211, 1.601]	0.325 (0.244)	[−0.160, 0.792]	0.760 (0.232)**	[0.291, 1.202]	0.064 (0.337)	[−0.633, 0.671]
Prior day sexual harassment	−0.144 (0.347)	[−0.841, 0.529]	0.469 (0.229)*	[0.008, 0.907]	−0.012 (0.282)	[−0.601, 0.507]	0.360 (0.281)	[−0.215, 0.882]
Weekend vs. weekday a	−0.056 (0.165)	[−0.384, 0.256]	0.097 (0.095)	[−0.090, 0.281]	0.449 (0.091)***	[0.277, 0.632]	0.257 (0.111)*	[0.039, 0.468]
Day of the study b	−0.007 (0.004)	[−0.016, 0.002]	−0.005 (0.003)	[−0.010, 0.000]	−0.002 (0.003)	[−0.008, 0.004]	−0.015 (0.003)***	[−0.022, −0.008]
Level 2
Gender	0.023 (0.398)	[−0.757, 0.828]	−0.206 (0.179)	[−0.556, 0.148]	0.053 (0.132)	[−0.190, 0.333]	0.057 (0.189)	[−0.329, 0.414]
Age	0.606 (0.315)*	[0.006, 1.307]	0.153 (0.108)	[−0.054, 0.371]	0.055 (0.080)	[−0.105, 0.206]	0.130 (0.130)	[−0.116, 0.400]
Childhood victimization	0.206 (0.360)	[−0.554, 0.878]	0.307 (0.183)	[−0.045, 0.674]	0.146 (0.127)	[−0.104, 0.401]	0.118 (0.184)	[−0.252, 0.476]
Total sexual harassment over 56 days	−0.025 (0.100)	[−0.231, 0.167]	0.007 (0.045)	[−0.081, 0.098]	−0.009 (0.030)	[−0.071, 0.047]	0.063 (0.041)	[−0.015, 0.148]
Total of DV over 56 days	0.160 (0.034)***	[0.101, 0.233]	0.145 (0.015)***	[0.120, 0.178]	0.267 (0.030)***	[0.217, 0.335]	0.237 (0.032)***	[0.189, 0.315]

Note.

^aWeekend (0 = Monday to Friday; 1 = Saturday to Sunday).

^bDay of the study (1–56).

^*p < .05.

^**p < .01.

^***p < .001.

As expected, there were robust individual effects of each substance: participants with more occasions of use were more likely to use on a given day. The between-person (Level 2) effect of number of sexual harassment victimization days was not significant in any analysis. Older adolescents were more likely to smoke cigarettes than younger adolescents. We also tested the interaction effects between sexual harassment and gender in each model, as displayed in Table 5. We did not observe any significant interaction effects, with one exception: the interaction between gender and number of sexual harassment victimization days was significantly associated with cigarette use, b = 0.426 (0.161), p = .006. Specifically, number of sexual harassment victimization days was negatively associated with the likelihood of cigarette use for female participants, b = −0.247 (0.126), p = .032; whereas number of sexual harassment victimization days was not associated with the likelihood of cigarette use for male participants, b = 0.166 (0.108), p = .136.

Discussion

Substance use and negative affect have been identified as outcomes of sexual harassment victimization (e.g., Chiodo et al., 2009; Wolff et al., 2017), raising the concern that sexual harassment victimization could be a risk factor for the development of substance use problems, particularly if substances are being used to manage negative affect associated with victimization. In this study, we sought to determine whether sexual harassment victimization was proximally linked to affect and substance use in an adolescent sample. Consistent with theories of self-medication (Khantzian & Albanese, 2008), we hypothesized that adolescent participants would report higher than typical negative affect and lower than typical positive affect on days on which sexual harassment occurred compared to days when they were not victimized. This hypothesis was partially supported; sadness (but not anger or positive affect) was related to sexual harassment victimization at the daily level. We further hypothesized that the likelihood of reporting substance use would be higher on victimization days relative to days when sexual harassment did not occur. We did find support for this hypothesis. Specifically, there was a greater likelihood of cigarette and alcohol use (but not e-cigarettes or marijuana) on the days when sexual harassment occurred relative to non-victimization days. In addition, there was an increased likelihood of e-cigarette use on the day following sexual harassment victimization, suggesting that there may be carry over effects to the next day.

The finding that sexual harassment victimization is associated with same day sadness as well as cigarette and alcohol use provides some support for the self-medication hypothesis. While substance use is normative for older adolescents, use among younger adolescents, such as those included in this sample, is a risk factor for future substance use problems and other negative outcomes (Mason et al., 2009). Although these findings are consistent with self-medication theories, we cannot rule out the possibility the daily level associations between sexual harassment victimization and alcohol and cigarette use observed in this study also may reflect involvement in activities and common contexts in which both behaviors occur. Alcohol and cigarette use are illicit for adolescents in this age group; thus, are likely to be used in unsupervised settings with other substance-using peers (Espelage et al., 2008; Finan & Lipperman-Kreda, 2020; Lipperman-Kreda et al., 2015). Alcohol consumption is particularly likely to occur in mixed-gender social contexts (e.g., parties) (Lipperman-Kreda et al., 2018), unlike marijuana which tends to be used when alone or with a few friends (Lipperman-Kreda et al., 2017). Social contexts that involve drinking are often imbued with expectations for sexual activity, even among adolescents, with alcohol serving as a social lubricant (Livingston et al., 2013). The combination of sexual expectancies along with the presence of alcohol-using peers make the situation ripe for sexual harassment (Kaltiala-Heino et al., 2018), which could in turn result in negative affect.

It is important to note that the temporal ordering of sexual harassment victimization and substance use within a given day could not be established. Future studies should consider use of ecological momentary assessment, (EMA) which would permit researchers to capture detailed in-the-moment information on when victimization and substance use occurred and the affective experiences at those time points. Prior research conducted with this sample also showed same day associations between bullying victimization and negative affect and cigarette use (Livingston et al., 2019), yet unlike bullying, in the current study sexual harassment was also related to same day alcohol use and next day e-cigarette use. This suggests that different types of victimization may be differentially associated with substance use. Whether different types of peer victimization contribute to different outcomes should be examined prospectively, so that the temporal ordering of victimization, distress, and substance use can be established. The current study contributes to current understandings of sexual harassment by showing that experiencing sexual harassment increases the likelihood of feeling sad on the same day. Negative affect may intensify over time in response to repeated exposure, concerns for safety, or stress from the pressure to conform to heteronormative ideals, beyond the acute negative response to a single incident as assessed in the current study. Studies that have examined repeated victimization over time have revealed associations with negative psychological outcomes. For example, in a longitudinal study of Swedish adolescents, sexual harassment victimization was stable over three waves, especially for girls, and there were reciprocal relations between sexual harassment victimization and depression for both boys and girls over time (Dahlqvist et al., 2016). Similarly, Wolff and colleagues (2017) found reciprocal relationships between sexual harassment and anger, depression, and alcohol problems (but not binge drinking) over two waves among male and female college students. Taken together, results from the current study and others show that sexual harassment is distressing to adolescents in both the short and long-term and should not be dismissed as harmless or normative.

Currently, there are a dearth of interventions targeting reduction of sexual harassment among high school students. Intervention approaches that promote prosocial norms and bystander intervention may be helpful in changing school culture to reduce sexual harassment and increase victim support among high school students. For example, Nickerson and colleagues (2022a, 2022b) have found that personal and peer anti-harassment norms were related to reduced perpetration and greater bystander intervention behaviors, suggesting that personal and peer norms should be targeted for intervention, along with bystander intervention. In addition, teaching adolescents strategies for coping with emotional distress and emotional regulation may help to manage emotional distress and subsequently reduce coping-related substance use. Counselors and pediatric health care providers who work with sexually harassed youth may also screen for depressive symptoms and substance use and provide resources to youth and their families.

Limitations and Strengths

This study adds to current understandings of peer sexual harassment by showing that peer sexual harassment can have adverse effects on adolescent well-being almost immediately. The study also demonstrates that sexual harassment is proximally linked to substance use, although causality could not be established. The study included a relatively strong completion rate over 56-days of daily diary data collection and the use of established measures of sexual harassment. Findings of the study are limited by use of a mostly White sample; however, the sample demographics for race/ethnicity and income are consistent with those for the county from which participants were drawn. Because the qualities of the sexual harassment experience and outcomes associated with sexual harassment may differ by race, future research should be carried out with a more diverse sample (Espelage et al., 2016). For example, Espelage and colleagues (2016) found that Black students were more likely than White students to experience more invasive forms of sexual harassment, such as being forced to kiss or to do something sexual other than kissing. In addition, sexual orientation and gender identity were not examined in this study. Sexual and gender minorities experience more sexual harassment and tend to be more adversely affected by it than heterosexual youth (Espelage et al., 2008; Kaltiala-Heino et al., 2019; Williams et al., 2003). Future studies should consider the unique sexual harassment experiences of gender and sexual minorities and the mechanisms through which harassment affects the health outcomes of this vulnerable population.

Due to limitations in the data, we were unable to ascertain the exact temporal ordering of sexual harassment and substance use within a given day. We assessed substance use, affect, and sexual harassment as it occurred today. On average the daily reports were completed at approximately 8 PM in the evening; substance use occurring after the daily report was completed was not assessed. This may partially explain the relatively low rates of substance use observed in the study. Further, we did not assess the contexts in which victimization and substance use occurred. Future research should examine whether adolescents are being sexually harassed in the contexts where substance use occurs or whether sexual harassment proceeds or follows from substance use. Qualitative or mixed methods studies that provide detailed information about the nuances of the experience and the settings in which they occur may be especially enlightening. Due to the relatively low incidence of sexual harassment incidents over the eight-week assessment period, we created a dichotomous sexual harassment variable that did not account for severity. It may be that some types of sexually harassing experiences are more distressing than others. Severity, chronicity, and frequency of victimization should be considered in future research examining the proximal and long-term impact of sexual harassment on emotional and psychological well-being. Finally, other current forms of stress (e.g., family conflict) that may be associated with substance use at the daily level were not assessed.

We note that episodes of both sexual harassment and substance use were low in this sample of young adolescents. Although we emphasized the confidential nature of the study, it is possible that some adolescents did not report their sexual harassment or substance use due to social desirability or concerns about getting caught. It is also possible that the strength of the hypothesized associations differs by developmental time periods. Younger adolescents, such as those in this sample, have inconsistent access to substances, thus the relationship between victimization and substance use may be more visible in older samples. Future research should examine these relationships among older adolescents using longitudinal survey data. This study provides a snapshot of victimization and substance use occurring over a 56-day period; thus, it may not have captured a period in which substance use or victimization occurred. Additionally, participants’ rates of reported victimization and substance use decreased slightly over the course of the 56-days, suggesting that participants may have been fatigued and were less likely to report incidents. Use of burst measurement designs where daily level data can be collected over multiple shorter-interval periods (e.g., 2 weeks) may provide a more nuanced and comprehensive picture of how sexual harassment and substance use are related proximally over time and across developmental stages. Future research should also consider potential moderators to better understand the conditions under which sexual harassment may be related to substance use. For example, Shadur et al. (2015) found that adolescents with strong friend support were less likely to use substances to cope with feelings of worry, compared to those adolescents who had high daily fluctuations in worry but less friend support.

Despite these limitations, this study contributes to the understanding of peer sexual harassment by establishing that sexual harassment and some types of substance use (i.e., alcohol and cigarettes) are proximally associated among adolescents. Sexual harassment is a unique and largely understudied form of peer victimization. Given the sexual nature of the victimization and its occurrence during a sensitive period in the development of sexual identity, sexual harassment victimization may be particularly damaging.

Conclusion

The current study provides new evidence that sexual harassment victimization has a same-day adverse impact the psychological well-being of adolescents. Further, the study shows proximal associations between sexual harassment and negative affect and substance use, providing support for the self-medication hypothesis. More longitudinal and developmental research is needed to establish the timing and conditions under which sexual harassment leads to psychological distress and problematic substance use, including information about the contexts in which victimization and substance use occur. Given the small number of adolescents reporting sexual harassment and substance use and the young age of the sample, it may be that these relationships occur more often among older adolescents or those involved in delinquent activities. The close, proximal associations between sexual harassment victimization and substance use at this early stage of development raise the concern that adolescents who experience sexual harassment may be at risk for developing substance use problems. Adolescents who have been targets of sexual harassment should also be screened for substance use. In addition, intervention to provide support and adaptive coping strategies may also serve to reduce substance use.