Social Media and Online Dating Safety Practices by Adolescent Sexual and Gender Diverse Men: Mixed-Methods Findings from the SMART Cohort Study

Abstract

Purpose:

Sexual and gender diverse youth (SGDY) develop and employ safety strategies on their own to manage risks while using dating apps. This study aimed to describe the online dating safety practices of SGDY and determine the effectiveness of an eHealth HIV-prevention educational intervention with dating safety content to promote future safety behaviors.

Methods:

SGDY aged 13–18 from across the U.S. (N=1087) were assigned to increasingly intensive HIV-prevention educational programs using a sequential multiple-assignment randomized trial design. Data were collected at three time points, each three months apart. Participants were asked multiple choice and open-ended questions about the safety practices they used online and in-person; mixed-methods described the safety behaviors of SGDY. Logistic regressions were used to determine psychosocial predictors of safety behaviors and the effectiveness of the intervention in promoting future safety practices.

Results:

60% (n=662) of participants used dating apps, most of whom reported using online (96.4%, n=638) and in-person (92.9%, n=615) safety strategies, such as limiting the disclosure of personal information or meeting other users in public. Outness and previous victimization were important psychosocial predictors of engaging in safety behaviors. Additionally, participants who received the online safety education were over 50% more likely to employ certain safety behaviors than SGDY who did not.

Discussion:

While most SGDY reported at least one type of safety strategy when dating online, safety practices differed across psychosocial variables, such as outness. This study provides evidence for the effectiveness an eHealth educational intervention tailored to SGDY to promote additional safety behaviors.

Introduction

Teenagers in the U.S. begin dating on average by 13 years, with over half having sought platonic, romantic, or sexual relationships online.¹ Due to a lack of dedicated digital spaces, teenagers turn to social media and adult dating apps (herein dating apps) for romantic and sexual exploration.^2,3 These apps are neither intended nor designed for teenage dating, and therefore, lack the necessary safety features – such as connecting with age-appropriate users or content moderation – to prioritize the wellbeing of minors.⁴ Sexual and gender diverse youth (SGDY) rely on digital spaces more than cis-straight counterparts, with several studies suggesting that between 52.5–77.7% of SGDY have used dating apps.^2,3,5,6

Dating apps offer SGDY a space for identity exploration and to connect with other members of their communities.^7–10 Nevertheless, most dating apps require that users be at least 18 years of age, meaning there are risks for minors who are likely to interact with adult users. Age-discordant relationships increase vulnerability to coerced sexual experiences,¹¹ physical assault, and HIV/STIs resulting from power imbalances between partners.^12–14 Other risks include being deceived by users who fabricate or misrepresent their identity – known as catfishing.¹¹

Like adults, SGDY use various techniques to manage risk when using dating apps. Online safety practices commonly include verifying the identity of other users and limiting the disclosure of personal information.^5–7 While there is less consistency among SGDY regarding in-person safety precautions, SGDY have endorsed employing the following practices: meeting in public, informing a friend, and remaining sober.^5,15

The safety practices of SGDY may differ across demographic and psychosocial factors due to differing perceptions of risk and/or access to resources and supports. Youth begin to rapidly increase personal information disclosure online between 10–13 years of age, which stabilizes in middle-late adolescence, suggesting that younger SGDY may be more likely to limit the disclosure of personal information on dating apps.¹⁶ Out adult SGDM often have greater access to affirming social supports and are more likely to disclose information to friends and family.¹⁷ Accordingly, out SGDY may also be more likely to let friends or family know about online dating experiences and meeting other users in person. Similarly, SGDY who receive online safety education from family may be able to more accurately assess risks while using dating apps and employ strategies to mitigate them.¹⁸ SGDY who have previously been victimized or had negative experiences with dating apps may develop and utilize additional safety practices after the experience. Lastly, adult SGDM living in rural areas have described dating apps as “inherently private” reported less perceived risks with the use of dating apps in their hometowns as compared to urban centres.¹⁹ These perceptions may be shared by ASGDM living in rural areas and thus may employ fewer safety practices when dating online. Other sociodemographic factors that shape the experiences of dating app users include race, gender, and body type, which may also play a role in the safety strategies employed by SGDY.^15,20

Due to a lack of comprehensive online safety and sex education, SGDY across the U.S. determine personal safety risks using different strategies, which they often come up with on their own.^5,6,21 For example, SGDY frequently report identifying “red flags” and relying on intuition to identify risky interactions without clearly defining what constitutes a red flag.⁵ Similarly, trans and gender diverse youth often lack skills and knowledge related to navigating romantic relationships and online dating particularly when they lack previous dating experience.⁶ SGDY are eager to learn about, and would benefit from, online safety and sex education.^5,6 An effective online dating safety educational intervention requires affirming and relevant information about the unique experiences of SGDY. The curriculum may include content regarding the culture of SGD virtual spaces; physical, emotional, and sexual health needs; healthy communication, relationships, and consent; and safer dating practices.

This study builds on research⁵ exploring the safety practices of SGDY by assessing the effectiveness of sexual health education,²² which included content on safer online dating, to promote safety behaviors. We were guided by the following research questions:

1. What are the safety practices of SGDY when interacting with other users online and meeting in-person?

2. How do safety practices differ across demographic and psychosocial factors?

3. Does a sexual health and HIV-prevention educational intervention with information about dating app safety promote future safety behaviors by SGDY at follow up compared to an intervention that does not include online safety content?

Methods

Ethics approval was obtained from Northwestern University Institutional Review Board with waivers of parental permission.²³

Study Design, Recruitment, and Randomization

Data were collected over three timepoints, three months apart, between April 2018–2021 as part of SMART, a pragmatic trial of a suite of HIV prevention interventions for SGDY. SMART participants were assigned to increasingly intensive educational programs through a sequential multiple-assignment randomized trial design (Figure 1).²⁴ Following baseline, all participants received the SMART Sex Ed (SSE) intervention, which included four modules related to sexual and gender identities, sexual health education, and healthy relationships. Response to intervention was measured at t2 and defined as: 1) 100% condom use if sexually active during the assessment period; 2) intention for condom use regardless of sexual activity; and 3) a high degree of perceived self-efficacy for the use of condoms for penetrative sex.²⁴ Responders did not receive another intervention before t3. Non-responders were randomized to SMART Squad (SS; intervention condition) or SSE2.0 (control condition). SS included information on dating, hookups, and online dating safety education; this intervention also included an interactive simulated hookup app, Humpr, which enabled participants to practice behavioral skills related to online dating. SSE and SSE2.0 were information-only interventions that did not include content about online dating.

Figure 1.

Consort diagram describing the SMART cohort and randomization. Intervention response was defined conservatively (e.g., no condomless sex and very high intentions to use condoms in the future) and differs from outcomes that indicate benefit from participating in the intervention program. Responders at t3 were excluded from t3 analyses. Only participants to randomized to the SS intervention condition received app use/safety information. SSE and SSE2.0 were information-only interventions that did not include content about online dating.

*this number includes a participant that was not randomized at T2 and is excluded from the analysis in this publication. The analytic sample is 1087.

NR – non-response; R – response; SS – Smart Squad; SSE – SMART Sex Ed; dau – dating app user at t2.

Participants were recruited between April 2018-June 2020 from across the U.S. and three territories largely by social media. Eligibility included: 1) aged 13–18; 2) assigned male at birth; 3) identified as SGDM; 4) prior sexual experience with another person; 4) HIV-negative or unknown status; 6) speak and read English or Spanish; and 7) consistent internet access.

Prospective participants completed an online screener. Those eligible were directed to a consent form and assessed on their understanding of the research procedures. Eligibility was confirmed by a research team member prior to participants’ enrollment in the cohort. Survey data were collected using REDCap,²⁵ and participants were compensated $25 per survey.

Data were analyzed from baseline, 3 months (t2), and 6 months (t3). Only individuals who completed all three surveys were included in this analysis. The analytic sample was 1087 SGDY (Figure 1). Effects of SSE and the overall SMART package on primary HIV-related outcomes have already been reported, so here we report secondary outcomes related to online safety.²²

Measures

Sociodemographic Characteristics

At baseline participants reported birthdate, from which age was calculated, and zipcode, which provided information on rurality (dichotomized to rural and urban).²⁶ Participants self-reported current gender, sexual orientation, race, living situation, and whether they received any sex education from parents or another family member.

Outness

Sexual orientation disclosure was collected at t2 by asking, “How out are you to people around you?” Responses were on a 4-point Likert scale: not out to anyone; only out to a select few people; out to most people; and out to everyone. Outness was dichotomized into low and high by combining the first two and latter two responses.

LGBT Victimization

Lesbian, gay, bisexual, and transgender (LGBT) related victimization was collected at baseline.²⁷ Participants reported how many times in the past three months they experienced the following because they were LGBT: physical violence; object thrown at them; punched, kicked, or beaten; threatened with a knife, gun, or another weapon; chased or followed; had their property damaged. Responses were measured on a 4-point Likert scale from “never” to “three or more times”. Victimization was dichotomized into none and at least one experience of victimization.

Social Media and Dating App Use

LGBT social media and dating app use characteristics were collected at t2 and t3. The following questions were amended from Winetrobe et al.²⁸ Current duration of use was measured using a 5-point Likert scale from “never/don’t use” to “for more than six months.” Participants also reported lifetime number of “hookups” with someone they met online. Twelve options ranged from zero/never to nine, more than 10 times, and I do not wish to answer. This item was dichotomized into never and having at least one hookup with someone from a dating app.

Participants reported on safety precautions in the last three months when talking with other users online through a multiple-response question with 10 options, as well as for in-person meetings with 13 options; choices were derived from previous qualitative research on safety precautions employed by teens when using dating apps.⁵ Responses were conceptually regrouped into seven strategies and dichotomized into whether participants employed said strategy (Table 2). Two open-ended questions asked participants to describe additional online and in-person safety precautions not listed.

Table 2.

Joint Display of Safety Precautions at t2 among Dating App Users

Quantitative Outcomes (Closed Responses)	% (N)	Qualitative Themes (Free Response)
Online Safety Practices		Verify Their Identity “Ask for a pic and put a paper with the date” Be in Control Conceal Personal Information “I don’t share too much information but enough to keep a good conversation” Trust your Intuition “Trust my gut” Use Blocking Feature “If I feel uncomfortable, I warn them, then block them” Plan Ahead “I make sure I know where were going and I don’t give any sign of weakness” “Think of how to get out or what to do if this turns bad/have a game plan” Cautious First Meetings “Avoid doing anything on the first meeting”
No Online Safety Practices	3.6 (24)
Verify Their Identity
Check out their other social media (e.g., Twitter, Instagram, etc.)	69.5 (460)
Use Snapchat to make sure it’s actually the person	67.7 (448)
Reverse image search their profile pic to make sure it’s actually the person	32.5 (215)
Verify the person through Face Timing/Skyping with them	28.6 (189)
Conceal Personal Information
Don’t reveal too much personal info (keep a basic profile that’s locked)	72.4 (479)
Use an alias/fake name	38.8 (257)
Let Them Know You’re Uncomfortable	38.4 (254)
In-Person Safety Practices
No In-Person Safety Practices	7.1 (47)
Be Alert and/or Sober
Don’t drink alcohol or take any drugs during the meeting so I can keep an even head	54.4 (360)
Look around the room/place you’re meeting	57.7 (382)
Let Someone Know
Make sure someone (e.g., a close friend) knows where you are	58.2 (385)
Use a tracking app with friends to share location	47.1 (312)
Have friends text for updates	35.2 (233)
Send a friend a pic of what you and the other person looks like before meeting	26.9 (178)
Bring along a friend to the meeting	6.0 (40)
Bring Lube and/or Condoms	54.7 (362)
Meet in Public	47.7 (316)
Bring a Weapon	27.2 (180)

Notes. The quantitative data safety responses are derived from individuals who used dating apps (n=662) and qualitative themes were derived from the subset of dating app users who responded to open-ended questions (n=57).

Data Analysis

Open-ended responses from t2 were inductively and deductively analyzed by the first author using reflexive thematic analysis.²⁹ Two responses in Spanish were translated to English. The first author relied on their expertise as a queer health scholar and lived experience using dating apps to guide analysis and interpretation.³⁰ They familiarized themselves with the responses and generated initial codes, then developed themes, which were reviewed by the research team, defined, and named.

Statistical analyses were conducted using SPSS.³¹ Eight hierarchical logistic regressions determined which psychosocial variables predicted safety precautions at t2. Only individuals who reported using dating apps at t2 were included in these analyses. First, sex education by parents, outness, and LGBT victimization were included as predictors. In the second step, living situation, rurality, and age were added as predictors, and was reported/interpreted if there was a significant (p<0.05) decrease in deviance (−2LL). The assumptions of logistic regression (linearity, overdispersion, and multicollinearity) were tested.³²

Another eight hierarchical logistic regressions determined if intervention condition predicted safety precautions at t3. Only participants randomized at t3 and used dating apps were included in the analysis (Figure 1). Intervention condition was added in the first step. The safety precautions used at t2 were added as a covariate in the second step for the respective analysis. Participants who reported having used dating apps at t3 but not t2 were included in the analysis as having not used any safety precautions at t2. The same logistic regression procedures outlined above were employed.

To achieve mixed methods integration, the qualitative themes and quantitative outcomes were similarly named, presented in a joint display (Table 2),³³ and both used to guide interpretation.

Results

Sample Characteristics

Table 1 summarizes the demographics for the total sample (N=1087). Participants represented 50 U.S. states and Puerto Rico (n=30).

Table 1.

Sociodemographic Characteristics and Gay, Bisexual, and Queer Social Media/Dating App Use

Characteristics	Full Sample	Dating App Use (t2)			Intervention Conditiona
		Users		Non-Users	SS	SSE 2.0
	% (N)
Total	100 (1087)	60.9 (662)		39.1 (425)	46.6 (507)	46.6 (506)
Age
13–14	7.2 (78)	2.7 (18)		14.1 (60)	7.1 (36)	7.3 (37)
15–16	36.0 (391)	24.6 (163)		53.7 (228)	34.1 (173)	36.4 (184)
17–18	56.9 (618)	72.7 (481)		32.2 (137)	58.8 (298)	56.3 (285)
Gender Identity
Male	94.3 (1025)	94.1 (623)		94.6 (402)	94.5 (479)	94.1 (476)
Female	0.9 (10)	0.8 (5)		1.2 (5)	1.2 (6)	0.6 (3)
Transgender	0.6 (6)	0.6 (4)		0.5 (2)	0.6 (3)	0.6 (3)
Non-binary/Queer/Non-conforming	3.6 (39)	4.1 (27)		2.8 (12)	3.4 (17)	4.0 (20)
Other	0.6 (7)	0.5 (3)		0.9 (4)	0.4 (2)	0.8 (4)
Sexual Orientation
Gay	67.7 (736)	72.2 (478)		60.7 (258)	62.7 (318)	72.1 (365)
Bisexual	23.8 (259)	21.3 (141)		27.8 (118)	28.4 (144)	20.2 (102)
Queer	1.5 (16)	1.4 (9)		1.7 (7)	1.6 (8)	1.4 (7)
Pansexual	4.0 (43)	3.2 (21)		5.2 (22)	3.9 (20)	3.6 (18)
Other/Unsure	2.9 (31)	1.7 (11)		4.7 (20)	3.4 (17)	2.6 (13)
Race
Latinx	35.9 (390)	37.9 (251)		32.7 (139)	36.4 (186)	35.8 (181)
American Indian or Alaska Native	6.7 (73)	7.3 (48)		5.9 (25)	7.7 (39)	6.1 (31)
Asian	10.5 (114)	12.5 (83)		7.3 (31)	9.7 (49)	11.5 (58)
Black or African American	21.3 (232)	21.5 (142)		21.2 (90)	21.7 (110)	21.2 (107)
Native Hawaiian or Other Pacific Islander	2.7 (29)	2.7 (18)		2.6 (11)	2.8 (14)	2.6 (13)
White	63.8 (693)	59.8 (396)		69.9 (297)	62.5 (317)	64.4 (326)
Other	9.8 (106)	11.3 (75)		7.3 (31)	10.1 (51)	9.3 (47)
Multiracial	14.2 (154)	14.2 (94)		14.1 (60)	14.6 (74)	13.8 (70)
Living Situation
With parents or family	88.4 (961)	82.6 (547)		97.4 (414)	88.4 (448)	87.6 (443)
Alone, with partner, or roommates	10.9 (118)	16.2 (107)		2.6 (11)	11.1 (56)	11.7 (59)
Housing insecure	0.7 (8)	1.2 (8)		-	0.6 (3)	0.8 (4)
Rurality (%urban)	83.2 (904)	83.8 (555)		82.1 (349)	84.0 (426)	81.6 (413)
Received sex education by parents (%yes)	40.6 (441)	41.2 (273)		39.5 (168)	42.4 (215)	39.1 (198)
Outness (%high)b	72.4 (787)	72.2 (478)		72.7 (309)	70.4 (357)	73.5 (372)
At least one experience of victimization (%yes)	37.2 (404)	37.0 (245)		37.4 (159)	36.7 (186)	39.3 (199)
Use Dating Apps (%yes)b	-	-		-	61.0 (309)	611.9 (313)
Duration of Dating App Use b
Less than a month	-	13.6 (90)		-	12.9 (40)c	13.4 (42)d
1–3 months	-	13.6 (90)		-	13.3 (41)c	13.4 (42)d
3–6 months	-	16.0 (106)		-	16.5 (51)c	16.9 (53)d
More than 6 months	-	56.8 (376)		-	57.3 (177)c	56.2 (176)d
Have ever hooked up with someone from a dating app (%yes)b	-	80.1 (530)		-	79.9 (247)c	81.8 (256)d
Online Safety b
Verify their identity (%yes)	-	86.3 (571)		-	87.4 (270)c	85.9 (269)d
Conceal personal information (%yes)	-	76.6 (507)		-	76.1 (235)c	76.34 (239)d
Let them know you’re uncomfortable (%yes)	-	38.4 (254)		-	39.8 (123)c	36.1 (113)d
In-person Safety b
Be alert and/or sober (%yes)	-	71.2 (471)		-	68.3 (211)c	73.2 (229)d
Let someone know (%yes)	-	70.2 (465)		-	69.3 (214)c	69.3 (217)d
Bring lube and/or condoms (%yes)	-	54.7 (362)		-	54.1 (167)c	53.4 (167)d
Meet in public (%yes)	-	47.7 (316)		-	45.6 (141)c	47.9 (150)d
Bring a weapon (%yes)	-	27.2 (180)		-	29.1 (90)c	24.3 (76)d
	M (SE)
Age	17.12 (0.04)	17.58 (0.04)		16.39 (0.06)	17.16 (0.06)	17.11 (0.06)

Notes. Demographic characteristics are collected at baseline unless otherwise specified. Dating app characteristics at t3 are reported as supplemental material.

^aParticipants are randomized to the intervention condition at t3 if they did not reach response to intervention (see methods for criteria to achieve response/non-response). 74 participants from t2 were not randomized to either SMART Squad (SS) or SSE 2.0 and were excluded from subsequent analyses.

^bVariables collected at t2.

^cDenominator is 309, which is the number of individuals who use dating apps and were randomized to SMART Squad (SS).

^dDenominator is 313, which is the number of individuals who use dating apps and were randomized to SSE 2.0.

Dating App Characteristics and Safety Precautions (mixed methods)

Dating app use characteristics at t2 are reported in Table 1. Most participants used dating apps (60.9%, n=662) and of those, 27.0% (n=179) reported having ten or more hookups with someone they met through an app.

Table 2 displays descriptive statistics and qualitative themes of safety strategies employed to visualize comparing/contrasting the findings. Nearly all app users reported at least one online (96.4%, n=638) and in-person safety precaution (92.9%, n=615).

Fifty-seven participants described additional safety practices through 64 open-ended responses (50 online and 14 offline). Two themes described online precautions (verify their identity and be in control) and two offline (plan ahead and cautious first meetings).

Verify Their Identity

Participants verified someone’s identity by searching or asking for specific information. For example, “ask them basic information about their everyday” (17, male, gay), “look them up online and check” (18, male, bisexual), and “ask for their phone number and google it” (17, male, bisexual). To ensure legitimacy, participants “Ask for a face pic and put a paper with the date” (15, male, bisexual) and “I ask them to take pictures of their hands with random fingers up” (18, female, bisexual). Other techniques included asking for a live photo through the app, a voice note reading a specific phrase, or a clear video.

Be In Control

This theme describes strategies participants used to be in control of online interactions through three sub-themes: conceal personal information; trust your intuition; and use the blocking feature.

Conceal Personal Information.

Several participants detailed strategies to limit personal information disclosure. For example, “I don’t share too much information but enough to keep a good conversation” (14, male, gay). The most frequently reported strategy was to “Use a photo that is not on any of my social media accounts as my profile photo” (18, male, gay) and “Use fake body pics” (15, male, bisexual). This would ensure that other users could not reverse image search and/or find participants’ other social media profiles. One participant refused to share pictures, “I told him I can’t send a picture” (17, male, gay).

Trust Your Intuition.

Some participants responded that they “trust my gut” (18, male, gay) and avoided users when uncomfortable. As another participant notes, “If they were acting shady or doing something suspicious I would not bother” (18, male, gay). Notably, participants suggested that they “respond/talk to only the people I want to” (15, male, pansexual).

Use the Blocking Feature.

Several participants blocked users to avoid unwanted interactions. As one participant describes, “I think I’m pretty stringent with guys from Grindr so if I feel uncomfortable, I’ll immediately just block them. If anything looks suspicious as well (such as their photo), then I’ll just block them.” (17, male, gay). These responses suggest that SGDY may be more inclined to block users than tell other users about their discomfort.

Plan Ahead

Some participants described ways they prepared for meeting another user in-person. Two participants reported familiarizing themselves with the meeting location: “I make sure I know where were going and I don’t give any sign of weakness” (18, male, bisexual) and “google map it” (17, male, gay). Participants prepared for negative scenarios by “think[ing] of how to get out or what to do if this turns bad/have a game plan” (18, male, gay), “bringing only basics, keys and phone” (18, male, bisexual), and bringing weapons. One participant noted “[having] my friends know where I am to check on me if something went wrong” (15, male, bisexual).

Cautious First Meetings

Being cautious when first meeting another user was common: “I only agree to meet them in safe situations” (18, male, gay). One participant “wait[ed] a certain period of time before meeting them [in person]” (17, male, gay). Three others described safe first meetings as “well lit” (18, male, gay), “during the day” (18, male, gay), and “in public and at a coffee shop” (16, male, bisexual).

Psychosocial Predictors of Safety Behaviors

The results of logistic regressions predicting safety precautions are summarized in Table 3. The second step with demographic variables was nonsignificant for all models and are not reported.

Table 3.

Hierarchical Logistic Regressions Predicting Safety Precautions at t2

DV	Coefficient	-2LL	𝛥 −2LL	β	OR	95% CI		Pseudo R 2
						Lower	Upper	Cox & Snell	Nagelkerke
Verify Identity	Step 1	525.21	4.83					0.007	0.013
	SexEd by Parents (yes)			−0.16	0.85	0.54	1.33
	Outness (high)			0.46	1.58	0.98	2.55
	LGBT Victimization (yes)			−0.31	0.74	0.47	1.17
	Step 2	520.84	4.37					0.014	0.025
Conceal Personal Information	Step 1	712.03	8.53 *					0.013	0.019
	SexEd by Parents (yes)			−0.17	0.84	0.58	1.21
	Outness (high)			−0.45*	0.64	0.41	0.99
	LGBT Victimization (yes)			−0.30	0.74	0.51	1.08
	Step 2	709.84	2.19					0.016	0.024
Uncomfortable	Step 1	879.08	2.49					0.004	0.005
	SexEd by Parents (yes)			−0.04	0.96	0.70	1.32
	Outness (high)			0.30	1.35	0.94	1.93
	LGBT Victimization (yes)			−0.03	0.97	0.92	1.03
	Step 2	868.07	10.78*					0.02	0.027
Let Someone Know	Step 1	760 ***	46.06 ***					0.067	0.095
	SexEd by Parents (yes)			0.07	1.07	0.75	1.52
	Outness (high)			0.95 **	2.58	1.79	3.71
	LGBT Victimization (yes)			0.72 **	2.06	1.41	3.02
	Step 2	757.67 ***	2.32					0.07	0.10
Bring a Weapon	Step 1	760.65 **	14.08 **					0.021	0.031
	SexEd by Parents (yes)			−0.09	0.91	0.64	1.30
	Outness (high)			0.28	1.32	0.88	1.98
	LGBT Victimization (yes)			0.58 **	1.79	1.26	2.55
	Step 2	759.23*	1.41					0.023	0.034
Meet in Public	Step 1	908.52 *	7.84 *					0.012	0.016
	SexEd by Parents (yes)			0.40 *	1.49	1.09	2.04
	Outness (high)			0.07	1.07	0.76	1.52
	LGBT Victimization (yes)			0.19	1.21	0.88	1.67
	Step 2	900.56 *	7.97					0.024	0.031
Bring Lube/Condoms	Step 1	890.44 ***	21.48 ***					0.032	0.043
	SexEd by Parents (yes)			0.49 **	1.64	1.19	2.25
	Outness (high)			0.56 **	1.75	1.24	2.49
	LGBT Victimization (yes)			0.14	1.15	0.83	1.59
	Step 2	885.34 ***	5.09					0.039	0.053
Be Sober/Alert	Step 1	787.36 *	8.12 *					0.012	0.017
	SexEd by Parents (yes)			0.19	1.21	0.85	1.71
	Outness (high)			0.49 **	1.63	1.13	2.36
	LGBT Victimization (yes)			−0.04	0.96	0.68	1.37
	Step 2	778.36 *	9.00					0.026	0.037

Notes. Step 2 included living condition, rurality, and age as predictors in the model. The second step was non-significant for all outcome variables and is therefore not reported. N=662 for all analyses.

^*p<0.5

^**p<0.01

^***p<0.001.

Participants who were out were significantly less likely than closeted participants conceal personal information when talking to other users online (p<0.05, OR=0.64). Conversely, participants who were more out were significantly more likely to express their discomfort to another user (p<0.05, OR=2.58). Similarly, participants who had experienced victimization were also more likely to express discomfort compared to participants without a recent experience of victimization (p<0.05, OR=2.06).

Regarding in-person safety behaviors, participants who were more out (p<0.01, OR=2.58) and have had a recent experience of victimization (p<0.01, OR=2.06) were more likely to let a friend know about meeting another user. Participants who recently experienced victimization were also more likely to bring a weapon to the meeting (p<0.01, OR=1.79). Higher outness predicted bringing lube and/or condoms to meetings (p<0.01, OR=1.75), as well as being alert and/or sober (p<0.01, OR=1.63). Finally, participants who received sex education by family were significantly more likely to meet users in public (p<0.05, OR=1.49) and bring lube and/or condoms to the meeting (p<0.01, OR=1.64).

SMART Squad Predicting Future Safety Behaviors

Table 4 summarizes the binary logistic regressions predicting the effect of the SMART intervention on future safety behaviors at t3. Participants who received the intervention were significantly more likely than those in the control condition to let a friend know about an in-person meeting (p<0.05, OR=1.58), meet in public (p<0.05, OR=1.49), and be sober and/or alert (p<0.01, OR=1.79). The second step was significant for all models, demonstrating that prior use of safety behaviors predicted continued practice of said behavior.

Table 4.

Hierarchical Regressions Predicting Safety Precautions at t3

DV	Coefficient	-2LL	𝛥 −2LL	β	OR	95% CI		Pseudo R 2
						Lower	Upper	Cox & Snell	Nagelkerke
Verify Identity	Step 1	428.89	0.84					0.002	0.003
	Intervention			0.23	1.26	0.77	2.07
	Step 2	400.38 ***	28.51 ***					0.053	0.1
	Intervention			0.20	1.22	0.73	2.04
	Used at t2			1.40 ***	4.04	2.43	6.72
Conceal Personal Information	Step 1	560.51	1.01					0.002	0.003
	Intervention			0.21	1.24	0.82	1.87
	Step 2	515.08 ***	45.43 ***					0.08	0.13
	Intervention			0.24	1.28	0.83	1.97
	Used at t2			1.45 ***	4.28	2.78	6.58
Uncomfortable	Step 1	699.28	0.57					0.001	0.001
	Intervention			0.14	1.15	0.80	1.63
	Step 2	643.20 ***	56.09 ***					0.1	0.14
	Intervention			0.15	1.16	0.80	1.69
	Used at t2			1.45 ***	4.26	2.89	6.28
Let Someone Know	Step 1	616.74	3.82					0.007	0.01
	Intervention			0.39	1.47	1.00	2.17
	Step 2	556.59 ***	60.16 ***					0.113	0.164
	Intervention			0.46 *	1.58	1.04	2.38
	Used at t2			1.58 ***	4.84	3.20	7.34
Bring a Weapon	Step 1	638.65	2.90					0.005	0.008
	Intervention			0.33	1.39	0.95	2.02
	Step 2	504.04 ***	134.61 ***					0.227	0.325
	Intervention			0.27	1.31	0.84	2.03
	Used at t2			2.64 ***	14.01	8.64	22.71
Meet in Public	Step 1	733.44	3.22					0.006	0.008
	Intervention			0.31	1.37	0.97	1.92
	Step 2	654.10 ***	79.34 ***					0.143	0.191
	Intervention			0.40*	1.49	1.03	2.16
	Used at t2			1.64 ***	5.15	3.54	7.51
Bring Lube/Condoms	Step 1	737.13	1.89					0.004	0.005
	Intervention			0.24	1.27	0.90	1.79
	Step 2	647.40 ***	89.72 ***					0.158	0.21
	Intervention			0.30	1.34	0.92	1.95
	Used at t2			1.73 ***	5.66	3.88	8.25
Be Sober/Alert	Step 1	633.82	5.91*					0.01	0.016
	Intervention			0.47	1.60	1.09	2.35
	Step 2	604.85 ***	28.97 ***					0.063	0.091
	Intervention			0.58 **	1.79	1.20	2.66
	Used at t2			1.06 ***	2.90	1.96	4.29

Notes. N=534 for all analyses. Of the 622 participants at t2 who used a dating app, 67 no longer used a dating app and 21 had yet to complete the survey at t3 at the time of analysis.

^*p<0.5

^**p<0.01

^***p<0.001

Discussion

Most SGDY in our study reported using dating apps, a proportion consistent with existing literature.^5–7,34 Nearly all of whom employed behaviors to increase their safety online and in person. These behaviors were predominantly to verify the identity of users and limit the disclosure of personal information until it was perceived to be safe. Requesting clear and unique pictures from users was a common strategy reported by our sample, and LGBT users in other studies, to establish trust and assess risk.⁵ Adolescents attempted to balance sharing information to maintain positive perceptions while limiting the disclosure of personal information.²⁰

Our study revealed that different psychosocial variables – but not demographic characteristics – were associated with safety precautions among adolescents. SGDY who were mostly or completely out were less likely to conceal personal information online. This finding is consistent with literature demonstrating that increased personal information disclosure is associated with greater LGBT community connectedness,^16,35 which is associated with outness.¹⁷ Outness was also associated letting someone know about meeting a dating app user in person, bringing lube and/or condoms to the interaction, as well as remaining sober and/or alert. Closeted SGDY may be at increased risk for not being able to confide in someone about meeting other users or for not receiving relevant sexual health education. Additionally, closeted SGDY may experience internalized homophobia, which has previously been associated with increased alcohol and substance use.³⁶

Adolescents who recently had an experience of LGBT-related victimization were more likely to let a friend know about meeting another user and bringing a weapon to said meeting. Social support buffers against in-person victimization and sexual harassment for LGBT youth, which may explain future reliance on social supports after experiencing victimization.¹⁰ It is unclear whether the aggressions reported by our participants were enacted by other dating app users. Additional research is needed to understand which dating app experiences result in future safety behaviors or discontinued use of said apps.

Education also predicted safety behaviors. SGDY who received sex education by a family member were more likely to meet other users in public as well as being lube and/or condoms to the meeting. It is unclear what this education contained or whether it was helpful or affirming; future studies should explore if the effectiveness of parental sex-ed varies according to quality and content.

The SMART eHealth intervention, which included online safety education, significantly increased several safety behaviors over three months. SGDY randomized to the treatment condition were more likely to let a friend know about meeting another user, meet in public, and be sober or alert. Our study provides evidence that comprehensive sexual health education including app safety is associated with future safety strategy utilization versus education that does not discuss app use. Curriculum pertaining to online dating, safety, relationships, sexual health, and identity/wellbeing should be made widely available to SGDY across the U.S. Educational interventions should target SGDY who have not previously received sexual health education and be widely accessible so that it may be accessed by closeted youth. One study exploring stakeholder perspectives, including clinicians and parents, determined internet safety education should start in childhood.³⁷ Being able to accurately assess risk and learning new safety behaviors is necessary for SGDY to safely navigate digital and related in-person interactions.³⁸ eHealth interventions can be widely distributed to youth across the U.S; they can also be privately accessed by closeted youth who otherwise may avoid LGBT-tailored sexual health education. Evidence-based sex-ed programs such as SMART are a key tool in the promotion of SGDY’s safety and wellbeing. Future directions should also extend educational materials to cis-straight adolescents.

Strengths and Limitations

To our knowledge, this study was the first to assess the effectiveness of an eHealth intervention to promote safety practices among SGDY. A key strength was the longitudinal randomized control design used to assess causality between receiving education and future safety behaviors. Additionally, the diverse study sample represented teenagers from across the U.S.

This study is not without limitations. Responses were self-reported and susceptible to bias. Our questionnaire did not differentiate between social media and dating apps, nor did we ask participants which apps they use. Different platforms may elicit different safety practices by SGDY due to the affordances of specific apps or perceptions of risk. Future studies should explore if safety behaviors differ across platforms. Some participants discontinued dating app use between t2 and t3. The reasons for discontinuing use were not reported, and it is unclear if this was due to negative/unsafe experiences. While SGDY and adults discontinue dating app use primarily due to beginning a new relationship, other reasons include negative experiences such as receiving discriminatory and objectifying messages.^39,40

Conclusion

Most SGDY employ at least one safety strategy while using dating apps, which many developed on their own without education or guidance. The safety behaviors of SGDY differed across psychosocial variables, like outness and prior LGBT-related victimization, which may be a result of differing perceptions of risks associated with using dating apps. Culturally appropriate and widely accessible sexual health education that includes online dating safety information can provide SGDY with the knowledge and resources to better protect themselves while online dating.

Implications and contributions :

Using a large sample of SGDY from across the U.S., this study is the first of its kind to determine psychosocial variables associated with dating app safety practices and establishes the effectiveness of and eHealth intervention to promote future safety behaviors.

Abbreviations

SGDY

Sexual and gender diverse youth

LGBT

Lesbian, Gay, Bisexual, and Trans

SS

SMART Squad

SSE

SMART Sex Ed

SSE 2.0

SMART Sex Ed 2.0