Speed-dating studies have yielded little evidence of a relationship between stated preferences and revealed preferences, casting doubt on the assumption that stated ideal partner preferences guide human mate selection. We conducted a high-powered speed-dating study (n=1145) and found only one out of the nine assessed traits provided evidence of a correspondence between stated and revealed preferences. In Study 2, we tested an alternative explanation for such null results. We conducted an agent-based modelling simulation according to the constraints of Study 1’s speed-dating design. We demonstrated that if speed-daters use a substantial number of traits to evaluate a potential partner, stated preferences for any one trait can be only minimally related to revealed preference for that trait. As more traits were used to evaluate mates and as preference measures became more noisy, the maximum association of a stated and revealed preference for a trait decreased rapidly. Therefore, with realistic assumptions about measurement error and number of traits used to evaluate mates, past studies like ours would have been underpowered to detect these associations, calling into question the previous interpretations of null findings. Altogether, our findings suggest that stated preferences may meaningfully drive revealed preferences while remaining difficult to statistically detect.