Swipe Right?: Experimental Analyses of App-Based Dating in the Age of Criminal Stigma

ABSTRACT

Stigma associated with prior incarceration could make dating difficult. To test this in online dating, an experimental audit used constructed dating app profiles of Black, Latinx, and White males and females. Experimental and control profiles were identical with the exception of a parole disclosure statement in bios of experimental profiles. Surprisingly, the White female profile disclosing parole was the only one to match with significantly fewer users. We analyzed the racial congruence of matches and found that White parole disclosing profiles were more likely to attract White users and Black and Latinx profiles attracted more heterogeneous users regardless of parole disclosure. The interaction between racial and criminal stigma and implications for relationship prospects among the formerly incarcerated are considered. Although the timing of criminal history disclosure matters, criminal stigma is unique in app-based dating, having more negative effects for Whites, which differs from other social domains.

KEYWORDS:

• Criminal stigma
• experiment
• online dating
• app-based dating
• parole
• race

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. “Googling” someone’s name, if unique enough, could uncover online articles of crimes that received media attention, or it could expose anyone on a state sex offender registry. For common names, internet searches would only yield criminal information in highly publicized cases.

2. Evans (2019) found that a single picture for profiles resulted in some post-match messaged accusations of the profile being fake or a “catfish.” In the current study there were negligible accusations of profiles being fake.

3. Prior research has defined “dating” as “a prolonged period of polygamous learning that eventually leads to a long term relationship such as marriage” (Jung et al., 2019, p. 54), as distinguished from more casual experiences, such as “hooking up.” However, such distinctions are fraught. Characterizing the date experience itself is complicated; people bring varying interests and motives even to the same date, possibly competing within the same individual, and often not completely in line with the other person. This is much more so for the collection of activities colloquially grouped into “dating” in general (Seffrin et al., 2009). This complexity is perhaps exacerbated when such meetings are facilitated through dating apps, which host users with far flung motivations, and those changing from moment to moment or depending on the individual on the other side of the interaction. Our definition is designed to encapsulate this bevy of potentialities in a broadly applicable category that roughly describes what takes place on these apps.

4. The researchers defined a match as a series of at least four message communications exchanged between two users (Jung et al., 2019).

5. It is worth noting that there are no blanket restrictions against cell phone ownership and dating app use for parolees, although convicted sex offenders on parole are generally prohibited from cell phone and internet use (Evans, 2012)

6. In those studies, the researchers used the following disclosure after several lines of positive description: “Not to get too deep, but in the past I made some poor choices (we all a work in progress), but I’m passionate about doing good in the world now. I speak to and counsel troubled youth because, up front, I’m coming off of parole (trying to live like the Joneses). Life is all about forward motion though, using your past to propel a brighter future” (Evans, 2019; Evans & Vega, 2020). The biography followed with several subsequent, welcoming statements inviting a match.

7. While swipe-based apps reveal one user at a time and require a decision (reject or attempt to match) be made before additional users are revealed, other apps require comments on other users’ pictures or answers to question prompts before a match can be made.

8. The IRB that approved this study asked the research team to contact app developers to inform them about the study and seek their approval to conduct it. Because no app developers responded, the IRB indicated that the research team could not disclose app names.

9. One app included in the study connected users worldwide. This app had the lowest overall match rate across all profiles (2.3%), by far, compared to the average across all apps (17.1%).

10. The average age of users who matched with study profiles was 27.4 years old.

11. Previous research has shown that younger adults, particularly in their twenties, utilize dating apps at somewhat higher rates than do older adults. A Pew Research Center survey revealed that nearly half of those aged 18–29 had used dating apps, the highest proportion of all age groups (Vogels, 2020). Both women and men tend to desire partners within three years of their own age, though in opposite directions (i.e. women seek slightly older men, with the reverse trend in men; DiDonato, 2019). The range of ages assigned to our study profiles fell within a range most active on dating apps and within three years of likely app users.

12. Swiping right indicates that one user “likes” another user but does not necessarily lead to an in-person date or even communication between matched users.

13. Common training would lead some readers to assume error here, i.e. that the dichotomous nature of the outcome required the use of logistic regression models. We sought to test the statistical significance of an interaction in a model with a categorical dependent variable – the precise situation in which the American Sociological Association, in its flagship publication, has advised against the use of nonlinear (e.g., logistic) models (Mustillo et al., 2018). The association cites work (e.g., Mood, 2010) that advise caution in using logistic models, especially when seeking to make comparisons across samples.

14. In the current study, we standardized most common demographic characteristics that would ordinarily be used as control variables, manipulating only race and parole disclosure. This standardization was designed to serve as an experimental control, as opposed to the statistical controls used more typically in survey strategies. We considered whether differences in believability or attractiveness may have influenced match rate, despite our attempt to minimize such differences through our profile selection process. Models including these variables did not substantially increase the explained variance of our outcome variable. That, and because the two ratings reflected the perceptions of our classroom raters and not of individual app users, we declined to include believability or attractiveness in our final model.

15. Of males who matched with female study profiles, 40% were White, 25% Latino, and 22% Black. The race/ethnicity of the remaining 13% of matches was not able to be confidently categorized because of differences in perceptions of at least two members of the research team.

16. Of males who matched with female study profiles, 40% were White, 25% Latino, and 22% Black.