How it works

PhillyMatch replaces the swipe deck with a questionnaire and a similarity score. This page explains what we measure, how we compare people, and the research the approach rests on.

1. The questionnaire

After you create an account, you answer 172 short statements on a seven-point scale from Strongly disagree to Strongly agree. The statements were written to cover the territory that actually comes up when two people build a life together: politics and social attitudes, religion and spirituality, money and ambition, family and children, sex and relationships, humor, risk-taking, routines and habits, cognitive style, and a range of personality traits. Many items are adapted from established, validated scales; others were written specifically for PhillyMatch.

Most people finish in 15 to 20 minutes. There are no right answers and no score to "pass." You can revisit and change your answers whenever you like, and you can hide your answers from other members; hidden answers are still used to match you.

2. Why similarity

Decades of research on attraction and mate choice point in the same direction: people reliably pair off with partners who share their attitudes, values, and outlook, and actual similarity is a robust driver of attraction, particularly early on, before two people know each other well [1, 2]. Similarity is not the whole story of a good relationship, but it is the part that a well-built questionnaire can measure accurately in advance. That is why PhillyMatch starts there.

It also happens to be the part that photo-first apps ignore. Ranking people by how many others swipe on them tells you who is popular. Ranking people by how much they agree with you tells you who you are likely to get along with.

3. The similarity score

Once you have answered the questionnaire, PhillyMatch compares your full pattern of answers with every other eligible member's pattern and produces a similarity score from 0 to 100. Two members who answered every statement identically would score 100; two members who took opposite positions on everything would score 0. The score is shown on profiles and is the main factor in the order you see members, so the people whose outlook is closest to yours come first.

A note on what the score is. The score summarizes how close two people's answers are across the statements they both answered, and it is only computed once there is enough overlap to be meaningful. The specifics of how individual statements are weighted and combined are proprietary and continue to be refined as the community grows. The score is a starting point for a conversation, not a verdict; chemistry, timing, and kindness still do the rest.

You can combine the similarity ranking with the usual practical filters (age, distance, and whether a member has verified their identity), and you can always browse freely.

4. The science behind it

PhillyMatch was built by a professional psychometrician, and the questionnaire and scoring draw directly on methods developed and tested in peer-reviewed measurement research. In particular:

  • Deciding what a long questionnaire actually measures. When an instrument covers many topics, it matters whether responses reflect one broad dimension, several distinct ones, or both. Work on bifactor models and rotation methods provides the tools for answering that question rigorously [3], and iterative target-rotation methods make it possible to recover a clean structure from a large, messy pool of items [4]. These methods informed how PhillyMatch's 172 statements were selected and organized.
  • Measuring more with fewer questions. Nobody wants a 600-item dating questionnaire. Research on computerized adaptive testing [5] and on principled ways to shorten test batteries when adaptive testing is not an option [6] is about exactly this problem: keeping the statements that carry the most information about a person and dropping the ones that are redundant. PhillyMatch's item set reflects that discipline.
  • Estimating a general factor across many different measures. Combining heterogeneous measures into one interpretable overall estimate, and validating that estimate against independent outcomes in a large cohort, is a well-studied problem in the founder's research [7]. The same logic underlies turning 172 separate answers into a single, interpretable similarity score.

The founder's broader body of work on psychological measurement, factor analysis, and adaptive testing is listed on MooreMetrics.

5. Your answers stay yours

  • Your questionnaire answers are used to compute similarity scores and nothing else. They are not sold, shared with advertisers, or used to build marketing profiles.
  • By default other members can see how you answered; you can switch this off in your profile at any time. Hidden answers are still used to match you.
  • You can delete your account, and with it your answers, at any time. See our Privacy Policy for the details.

References

  1. Montoya, R. M., Horton, R. S., & Kirchner, J. (2008). Is actual similarity necessary for attraction? A meta-analysis of actual and perceived similarity. Journal of Social and Personal Relationships, 25(6), 889–922. doi:10.1177/0265407508096700
  2. Luo, S., & Klohnen, E. C. (2005). Assortative mating and marital quality in newlyweds: A couple-centered approach. Journal of Personality and Social Psychology, 88(2), 304–326. doi:10.1037/0022-3514.88.2.304
  3. Reise, S. P., Moore, T. M., & Haviland, M. G. (2010). Bifactor models and rotations: Exploring the extent to which multidimensional data yield univocal scale scores. Journal of Personality Assessment, 92(6), 544–559. doi:10.1080/00223891.2010.496477
  4. Moore, T. M., Reise, S. P., Depaoli, S., & Haviland, M. G. (2015). Iteration of partially specified target matrices: Applications in exploratory and Bayesian confirmatory factor analysis. Multivariate Behavioral Research, 50(2), 149–161. doi:10.1080/00273171.2014.973990
  5. Moore, T. M., Calkins, M. E., Reise, S. P., Gur, R. C., & Gur, R. E. (2018). Development and public release of a computerized adaptive (CAT) version of the Schizotypal Personality Questionnaire. Psychiatry Research, 263, 250–256. doi:10.1016/j.psychres.2018.02.022
  6. Moore, T. M., Butler, E. R., Scott, J. C., Port, A. M., Ruparel, K., Njokweni, L. J., … Gur, R. C. (2021). When CAT is not an option: Complementary methods of test abbreviation for neurocognitive batteries. Cognitive Neuropsychiatry, 26(1), 35–54. doi:10.1080/13546805.2020.1859360
  7. Moore, T. M., Calkins, M. E., Wolf, D. H., Satterthwaite, T. D., Barzilay, R., Scott, J. C., … Gur, R. C. (2025). Estimation and validation of the "c" factor for overall cerebral functioning in the Philadelphia Neurodevelopmental Cohort. Applied Sciences, 15(4), 1697. doi:10.3390/app15041697
Take the questionnaire — it's free About PhillyMatch