From Dating to Marriage: Has Online Dating Made a Difference?

Who marries whom, and why? Do people prefer someone like themselves or someone different? How much effort is placed into finding their prince or princess charming? How strong are people’s preferences for certain characteristics? Seeking a mate is a fundamental part of the human experience. Have the dynamics of this search changed with the advent of online dating?

In their 2019 paper, Michael Rosenfeld, Reuben Thomas and Sonia Hausen documented how the percentage of married couples who met online went from only 2% in 1998 to 20% in 2008 and then to nearly 50% in 2017, becoming the dominant form of initial contact for couples who marry. But is it true that online dating has had a big effect on the marriage market and its characteristics? One would think it has, given that dating sites and apps represent a grand technological advance that can potentially decrease the costs of searching for a mate.

To test that idea, Anton Cheremukhin, Antonella Tutino and I used a model of targeted search to analyze the U.S. marriage market—and its evolution over time. In our working paper, For more on our methodology, see the St. Louis Fed working paper “Marriage Market Sorting in the U.S.,” revised September 2024. we estimated people’s preferences and developed new tools to assess the following:

• To what extent people prefer someone like themselves
• How selective (picky) people are when searching for a potential partner
• The degree of complementarity or substitutability of different attributes
• How income inequality has been affected by the degree of selectivity of people
• Whether search costs (such as time and effort) have gone down over time due to the innovation in meeting technologies brought about by the increased popularity of online dating websites and apps

My co-authors and I used the American Community Survey for the 2008-21 period; for longer-term comparisons, we employed data from the U.S. Census for 1960 and 1980. These data contain information on the education, race, income, skill level and age of people that have just gotten married.

Our results indicated that there were strong preferences for marrying someone of the same race and someone with more or less the same education and age, but everyone preferred someone with higher income and skill levels. Furthermore, the paper found minimal changes in these preferences over the 2008-21 period, the years during which online dating grew to dominate marriage selection. But when they include data from 1960 and 1980, the results showed that people increased their preferences for mates who have the same income, education and skill levels.

In terms of selection, our paper found that people were most selective on race, followed by education and age to a much lesser degree. Also, people were more selective when considering income and skill level jointly than when evaluating the two characteristics individually. During the 2008-21 period, females became slightly more selective on age while males became slightly more selective on education. When the results included data from 1960 and 1980, it was also clear that both males and females became less selective on race.

Interestingly, when my co-authors and I studied how marriage outcomes have affected inequality, we found that because people have increasingly been marrying someone more like themselves, that can account for approximately half of the increase in household income inequality between 1980 and 2020. The most important factors contributing to household income inequality are selection on education (35%) and skills (30%), with selection on income (15%) and age (15%) trailing significantly; selection by race (5%) plays a relatively inconsequential role. The collective influence of partner selection on household income inequality is substantial, leading to a 14-point increase in the coefficient of variation, or a 3-point increase in the Gini coefficient.

Finally, our findings of minimal changes in preferences, selectivity and aggregate sorting of couples over the 2008-21 period is surprising. Given the proliferation of online dating, we would expect to observe a substantial improvement in the ability to find and meet potential partners, reflecting reduced search costs brought about by this improved search technology. Instead, the findings show that the data contradict these predictions and suggest that search costs have not changed over time despite the wide use of online dating sites and apps.

The reduction in search costs cannot be the explanation of these medium- to long-term changes. The recent literature documents a similar lack of improvement in matching efficiency in labor and product markets, despite the apparently substantial enhancement in search technologies and associated reduction in physical search costs. A 2020 paper by Paolo Martellini and Guido Menzio proposed that increasing selectivity in searches (people becoming pickier) could compensate for reduced search costs, thereby resolving this conundrum.

However, within the context of the marriage market, my co-authors and I observed no decrease in search costs and no discernible increase in selectivity since 1960. Instead, we argue that effective search costs reflect individuals’ capabilities to process and evaluate information, which subsequently determines their proficiency in mate selection within the marriage market. Despite technological progress and its promise of easier matching, individuals’ capabilities to process information appear to remain unaltered, accounting for the apparent lack of enhancements in matching efficiency and selectivity.

Note

1. For more on our methodology, see the St. Louis Fed working paper “Marriage Market Sorting in the U.S.,” revised September 2024.