top of page

Racial Bias in Online Dating Algorithms

Research Plan

Notes on online dating and racial bias:
  • Users created an intentional bias and train the algorithms the algorithms to be bias

    • Apps that have race filter:

      • Coffee meets bagel

      • OkCupid

      • Hinge (recent feature, at least not the in early 2012's)

  • There are lack of research scope within the area

    • is it because these online dating each of the online dating apps use differently matching and predictive algorithms?

  • Does the user location play a role in this algorithms bias?

    • i.e. if the user located in White dominant areas, does the algorithms only show them mainly White users or a good mixture of users?

  • Does the algorithms focus more straight relationship or also on same-sex relationship?

    • does the racial bias also effect on same-sex matching?


Photocredit: Photofeeler

  • Prefer to implementing a mix-method (quantitative and qualitative) for the finding of the research

    • Based on the existing research studies: the new direction of the study will be focus sorely on quantitative – testing out the theory of racial bias matching algorithm have a role in online dating environment and research hypotheses.

  • The participant will perform task based in altering their account filter. If the participant does not a Hinge account, they can make one under their alias.

Hinge Features

The application implemented Gale-Sharply

  • Algorithms: Gayle-Shapley/Marriage Matching Algorithm

  • The app also measure based on the user engagement and who the user engage with, as well as anyone have a similar preferences.

  • How User Get Match

    • Answer 3 questions of your choices that other user will see.

    • Upload 6 picture of yourself.

    • The user have the option to select their prefer ethnicity.​

  • How Hinge Algorithm work

  • Gayle–Shapley Algorithm

    • The algorithm solves through a series of iteration in which user A is being propose to their highest rank user B.

  • Read on case studies that associated with algorithms and racial bias

  • Assign more task base activities for the participants to investigate on the algorithms of Hinge.

    • Should I be recruiting more participants to join the research? If so, what demographic would be more benefit to the study – Black users? Latino/Hispanic users? Asian users? Or Caucasian/White users?

      • It would be interesting for the research studies if more non-BIPOC users participate in the project – why?

        • Get a different perspective from a majority group

        • At best, can come up with a predictive algorithms basing on their likes

  • Try to come up a mock matching predictive algorithms that work similarly as Hinge

    • Python programming

    • Read more about Machine Learning algorithms and how it matching predictive algorithms work in a large scale data.

bottom of page