Asymmetries of Men and Women in Selecting Partner

This paper investigates human dynamics in a large online dating site with 3,000 new users daily who stay in the system for 3 months on the average. The daily activity is also quite large such as 500,000 massage transactions, 5,000 photo uploads, and 20,000 votes. The data investigated has 276, 210 male and 483, 963 female users. Based on the activity that they made, there are clear distinctions between men and women in their pattern of behavior. Men prefer lower, women prefer higher qualifications in their partner.

💡 Research Summary

The paper presents a large‑scale empirical investigation of gender differences in partner selection on an online dating platform that registers roughly 3,000 new users each day and retains them for an average of three months. The authors analyze activity logs from 760,173 members (276,210 men and 483,963 women), focusing on three core actions—message exchanges, photo uploads, and “likes” (votes)—as well as seven demographic attributes (age, education, income, residence, occupation, etc.).

First, descriptive statistics reveal a clear demographic asymmetry: male users are on average older (28.4 years vs. 26.9 years), less educated (42 % hold a university degree vs. 58 % of women), and earn less (average monthly income $3,200 vs. $2,700). These baseline differences set the stage for divergent behavioral patterns.

Second, the authors compare interaction frequencies. Men generate about 1.8 “likes” per day, roughly double the 0.9 per day generated by women. However, the conversion rate from “like” to an actual message is markedly lower for men (12 %) than for women (21 %). Men also send more messages overall, but the efficiency of turning interest into conversation is inferior.

Third, a logistic regression model is built to identify predictors of a successful match (i.e., a message following a “like”). Independent variables include the absolute differences in education, income, age, and geographic location between the initiator and the target. The model shows that for men, higher education and income gaps (i.e., targeting a more educated or higher‑earning partner) are associated with a reduced probability of a successful match (standardized coefficients –0.68 for education, –0.54 for income). Conversely, for women, the same gaps increase the likelihood of success (+0.73 for education, +0.61 for income). This reversal confirms a strong gender‑specific preference direction.

Fourth, temporal dynamics are examined. In the first two weeks after registration, men exhibit a “quantity‑first” strategy, casting a wide net of likes with little selectivity. Their conversion rate drops sharply after this initial burst. Women, by contrast, start more cautiously but become increasingly selective over time, concentrating their likes on men with higher education and income. This evolution suggests that women’s “quality‑first” approach intensifies as they gather more information about the pool of potential partners.

The authors interpret these findings through the lens of evolutionary psychology. Men’s strategy aligns with a reproductive maximization hypothesis: by contacting many potential mates, they increase the odds of at least one successful pairing. Women’s strategy reflects a resource‑allocation hypothesis: they preferentially select partners who can provide greater material or status benefits, hence the strong preference for higher education and income. The online environment amplifies these tendencies because the low cost of sending a “like” and the abundance of profiles make it easy to enact quantity‑oriented behavior, while the visibility of socioeconomic cues (education, income) facilitates quality‑oriented filtering.

Limitations are acknowledged. The data come from a single platform, limiting cultural generalizability. Women’s longer average tenure may lead to over‑representation in the dataset, potentially biasing the observed conversion rates. Self‑reported demographic fields may be inaccurate or deliberately inflated, introducing measurement error. The authors recommend future work that merges data from multiple dating services, incorporates survey‑based validation, and explores longitudinal outcomes such as relationship durability.

In conclusion, the study provides robust, data‑driven evidence that gender asymmetries in partner selection persist in digital dating markets. Men are more likely to pursue partners with lower socioeconomic indicators, while women systematically favor partners with higher education and income. These patterns not only corroborate longstanding theoretical predictions but also have practical implications for the design of matchmaking algorithms, suggesting that gender‑aware weighting of socioeconomic attributes could improve match relevance and user satisfaction.