Bias in Dating Apps

Bias in dating apps refers to the way the algorithm that operates these apps filters through profiles of users based on certain characteristics, such as race, gender, sexual orientation, age, ability and other factors. This bias can manifest in different ways, including through the use of data gathered from the user’s profile which is usually entered by the user upon creation of the profile, as well as information from social media and other apps that the user chooses to link to the profile, such as Instagram and Spotify . The algorithm also uses a technique known as “collaborative filtering”, where the potential profiles shown to users are based on opinions from other users on these apps, such as on apps like Tinder, Bumble, and Hinge .

History

How the Algorithm Works

Tinder

The algorithm implemented by Tinder is the Elo rating system, which resembles the same method to assess and calculate the relative skill levels of players in zero-sum games such as chess . Tinder implements the same concept by coining the term “desirability” instead . The app uses this scoring system to compare and match users of those in the same or similar ranking . Users who are on the same ratio of right and left swipes are understood by Tinder’s algorithms to be at eye-to-eye desirability .

Bumble

Bumble employs filters to showcase the most compatible matches to users based on a match percentage. These filters encompass a range of factors such as age, location, religious affiliations, hobbies, and others, allowing the app to connect you with someone who shares similar characteristics . When an individual is joining Bumble as a new user, one may encounter numerous visually appealing profiles in their feed. The app utilizes an algorithm that takes into account the positive or negative swipes on a profile and adjusts the match score accordingly. This helps to match users with profiles of a similar level of physical attractiveness. Bumble is said to utilize the ELO rating system, which is a method for determining the relative skill levels of participants in games with winners and losers. The app's algorithm takes into consideration various factors, including the number of profiles a user has swiped left or right on, as well as the number of profiles that have swiped left or right on them. Bumble arranges users by determining the profile with the highest rating, taking into account the number of positive swipes on a profile and the level of activity on the platform.

Hinge

Hinge employs the Gale-Shapley algorithm, which matches individuals who are believed to have a mutual liking towards each other . This algorithm takes into consideration factors such as user engagement and engagement received, as well as aligning users with individuals who have comparable preferences.

The compatibility on Hinge is determined based on various factors such as profile information, answers to prompts, likes, dislikes, and "deal breaker" options. There is also a “most compatible” feature that shows on top of all the other profile cards. The app uses this information to match individuals with similar preferences and interests, reducing the chances of incompatible matches . The more an individual uses Hinge, the more the app will learn about their preferences and make better matches. The founder of Hinge, Justin McLeod, stated that the app becomes more accurate the more the user reveals their preferences . It is like recognizing a pattern in the types of people you have dated or been attracted to in the past, and using that information to make more informed decisions in the future.

Grindr

Grindr uses machine learning algorithms to personalize the user's experience . The company takes into account the personal information provided by the user, such as their profile, along with information collected from the usage of the app to develop algorithms that are based on the user's interests, preferences, and behavior . This allows Grindr to use automated decision-making to make the app more personalized for the user, such as suggesting other profiles for them to view. It's important to note that the company does not share information about the user's HIV status or last tested date with their machine learning partners, nor does it process all of the user's profile information (such as ethnicity) for personalization purposes .

Types of Bias

Race

Many dating apps had ethnicity filters such as Grindr, Tinder and Bumble, but were pressured to eliminate them after the Black Lives Matter (BLM) movement . Currently, dating apps like Hinge and OkCupid allow users to identify their own race and their racial preference in a potential partner . When a user decides to share this information, the algorithm would group other individuals by race and learn their preferences from the data it has access to . However, even when users refuse to state any race or racial preference, the collaborative data still allows the algorithm to make predictions and recommendations based on similar racial lines by using assumptions about an individual’s race based on the data it has about the preferences of other users who are similar to them in other ways .

Research has shown that dating apps that allow users to implement racial preferences by filters, or rely on the algorithm to show potential users of the same race, perpetuates racial biases and divisions . For example, men and women who identify as black on the app are 10 times more likely to message those who identify as white or caucasian than those who were white to message those who are black .

Gender and Identity

A study was done on the dating app Tinder, which suggested that there is the existence of gender bias on the platform, where men had a higher success rate of getting matches than women users . The reason behind this was due to the relationship of the amount of swipes and the amount of matches made as a result, where they both increased in volume . In addition, men users had a higher tendency to make the first move by initiating the conversation after matching . The purpose of men using the app as a platform for casual hookups had also contributed to the higher success rate of receiving matches. The study also found that women are more likely to be judged on their physical appearance and how they present themselves, whereas men are judged based on their occupation and education level .

Sexual Orientation

Queer women reported having to face certain challenges on dating apps, such as exclusion for their sexual orientation, other than being a lesbian . HER, a dating app for queer women remains as one of the platforms for those seeking the same dating social space. A claim was made by Exton where “lesbian users are much more valuable”, dismissing the different sexual orientations and are on a different spectrum such as users who are bicurious or bisexual . Scarcity was prevalent when lesbian users on all-inclusive dating apps like Tinder, would often be left with no other profiles to swipe on who were in the similar spectrum of sexual orientation. Women of color found it difficult to connect with others of the same race, as these apps were predominantly white .

There were also risks imposed by the infiltration of users who did not identify with being a queer women, including those who were heterosexual men, heterosexual women, and couples. To an extent where in China, there were users seeking surrogacy services .

Ableism

Ableism takes on many forms such as not having enough representation for those with disabilities, accessibility options for users with visual or hearing impairments, underrepresentation of individuals with disabilities in advertisements or profiles of users using the app, as well as negative connotations .

Dating apps heavily focus on visual representation, for example, pictures of the users’ profiles, biography and instances where a conversation occurs through text. These examples present challenges for users who have visual impairments. A study done by Arizona State University’s Center for Cognitive Ubiquitous Computing found that dating apps are hard to visualize, even with the assistance of a screen reader. However, some apps such as Tinder, Hinge and OKCupid implemented features like labeled buttons, clear navigation, and a screen reader to improve accessibility .