Madrid’s Diversity Masks Persistent Prejudice

Madrid is Spain’s largest and most diverse city, home to a mix of long-established Spaniards, immigrants from across Latin America, Africa, and Asia, and internal migrants from other regions. Its role as the capital and cultural hub creates an image of cosmopolitan openness. Yet the city also mirrors Spain’s long-standing patterns of discrimination historically toward Jews, Muslims, and Roma, and more recently toward immigrants, racial minorities, and even other Spaniards perceived as outsiders due to regional or linguistic differences. In this environment, progressive inclusion efforts exist alongside far-right rhetoric, public protests against racism, and everyday experiences of prejudice.

The Experiment That Measured Racial Preference on Tinder

In May 2024, researchers ran a 2×2×2 field experiment on Tinder, varying profile race (Black vs. White), gender (female vs. male), and search mode (heterosexual vs. same-gender). They used four fake profiles: a Black woman, White woman, Black man, and White man that were all aged 26, with Spanish names, neutral bios, and profile photos drawn from a standardized image set rated for attractiveness. Attractiveness ratings were tightly clustered between 3.55 and 3.86 on a 1–7 scale. Premium subscriptions ensured maximum visibility under Tinder’s own algorithm. All profiles were geolocated to central Madrid, set to search within a 159 km radius, and targeted ages 20–36. A Python script randomly liked 80% of the profiles they encountered, generating 4,160 right swipes over the course of the study.

What the Numbers Showed

The results were consistent across conditions: profiles coded as Black reduced the likelihood of receiving a match by 7.7 percentage points and reduced the likelihood of receiving an unsolicited conversation by 5.5 points. The average match rate for likes sent was around 23%. The strongest racial gap appeared in the lesbian condition, where the White woman received 94 matches compared to 39 for the Black woman. For heterosexual men, differences in match rates were not statistically significant, largely because they received few matches overall. These patterns suggest a persistent preference for White profiles in Madrid’s Tinder environment, even in same-gender searches where social acceptance might be assumed to be higher.

Methodology Was Strong but Missing Key Controls

The study used well-controlled synthetic profiles, standardized bios, and randomized likes to reduce human bias. These design strengths give the experiment credibility. However, several missing elements limit how precisely the results can be interpreted, because each gap introduces uncertainty about whether the observed racial differences reflect actual user preferences or other factors.

1. Lack of Impressions Data

First, Tinder does not provide impression counts, meaning the researchers could not determine how many times each profile was shown to others. Without this denominator, it is impossible to know if the lower match rate for Black profiles was due to active rejection by users or simply because the profiles appeared on fewer screens. This uncertainty directly affects whether the measured 7.7% gap represents genuine bias or an algorithmic visibility effect.

2. Profiles Created in Different Weeks

Second, the heterosexual and same-gender conditions were run in different, non-overlapping weeks. This design choice leaves open the possibility that external factors such as holidays, events, or fluctuations in app usage could have influenced results. If, for example, overall activity was higher in the week that White profiles were tested in a certain condition, part of the observed racial difference could come from timing rather than preference.

3. Clustering Errors Skew Results

Third, the statistical analysis treated each “like” interaction as independent, without clustering adjustments for the fact that many outcomes came from the same experimental profile or the same user seeing multiple profiles. Ignoring this dependence can underestimate standard errors, making small differences appear more statistically certain than they really are. This raises the risk that some subgroup findings may look significant when they are partly due to repeated exposures.

4. Algorithm May Have Changed Visibility

Fourth, the large number of likes sent by the experimental profiles could have triggered Tinder’s algorithm to adjust their visibility dynamically, potentially rewarding certain accounts more quickly. If visibility changes mid-experiment, the timing of exposure may interact with profile race in ways the design could not control. This feedback effect could inflate or dampen racial gaps depending on how the algorithm responded.

Each of these issues matters because they all create alternative explanations for the observed racial disparities. While the direction of the effect — favoring White profiles — is consistent across conditions, the exact size of the gap could be overestimated or underestimated depending on how these uncontrolled factors played out.

Persistent Racism in Madrid Impacts Dating

The Tinder results align with events exposing persistent racism and xenophobia in Madrid. Housing discrimination is widespread, with 99% of real estate agencies in Madrid and Barcelona reportedly accepting rental agreement clauses that exclude foreign renters. Police profiling continues despite a 2014 legal ban, with studies showing Black people are stopped up to 42 times more often than white individuals in certain contexts. Entire neighborhoods, such as the Cañada Real Galiana settlement, illustrate structural marginalization, where racialized communities live with environmental hazards and inadequate infrastructure. Roma communities also face persistent exclusion in dating, housing, education, and employment. These systemic patterns create a social backdrop in which digital discrimination can operate.

Dating Apps Amplify Racial Inequalities

Dating apps like Tinder condense first impressions into seconds, making it easy for offline stereotypes to shape online behavior. In Madrid, where social segregation and racial profiling already influence daily interactions, dating apps can replicate or even amplify these inequalities. The anonymity and low accountability of swiping may also make users more willing to act on biases they might suppress in face-to-face settings. This dynamic means that online dating is not an escape from societal prejudice, but another arena where it plays out.

The Bigger Picture

Madrid’s Tinder racial preference gap fits into a global pattern where online dating both reflects and reinforces offline racial hierarchies. In Spain, historical prejudices, structural inequalities, and modern xenophobia converge to shape personal interactions, whether on the street or on a screen. Addressing these patterns requires both broader social change from housing reform to anti-profiling enforcement and greater transparency from digital platforms about how their algorithms mediate visibility and interaction. Without such efforts, bias in online dating will remain a mirror of offline inequality, even in cities that pride themselves on diversity.

FAQs

Does This Study Prove Madrid Is More Racist Than Other Cities?

The experiment focused on Madrid and did not include comparative cities. The findings are consistent with evidence of bias in Spain but cannot rank Madrid against other locations without similar studies elsewhere.

Could Tinder’s Algorithm Have Influenced the Results?

Yes. Because dating app profile impression counts were unavailable, the study could not separate user choices from algorithmic visibility. The measured gaps could reflect both preference and exposure differences.

Do Same-Gender Dating Dynamics Reduce Racial Bias?

Not in this experiment. The largest White preference appeared among lesbian users, contradicting the assumption that same-gender contexts necessarily reduce racial selectivity.

Can These Findings Be Generalized Beyond Madrid or Tinder?

Cautiously. The study used Madrid-based profiles on Tinder with Premium features. Results may differ in other regions, on other platforms, or with different profile characteristics.

What Would Strengthen Future Studies?

Concurrent testing windows, impression data, clustered or mixed-effects models, and designs that capture respondent demographics would clarify whether effects are due to preference, exposure, or homogamy.

Keep Reading

• Dating App Algorithm Bias Leaves You With Fewer Likes and Matches – How dating app recommendation systems can amplify racial bias by deciding who appears in your swipe queue.
• Why Passport Bros Cross Borders for Love – Global research on interracial dating trends and how location shapes who swipes right on whom.
• Dating App Swipe Fatigue and Bias – How rapid-fire swiping increases reliance on stereotypes and decreases openness to diverse matches.
• Dating App Profiles That Break the Pattern – Examples of real profiles that overcame algorithmic sorting and racial bias to create meaningful matches.

References

Arranz Aldana, A., & Salazar, L. (2024). Racial preferences in dating apps: An experimental approach. The History of the Family, 29(4), 620–640. https://doi.org/10.1080/1081602X.2024.2352547