Scholarship on the consequences of racial/ethnic diversity often claims that diversity undermines trust, participation and cooperation. This work has been criticized for its inability to discern the causal effects of diversity. We draw attention to a more elementary issue: most studies are unable to interpret associations between their outcomes of interest and diversity, as these may be due to associations with non-White or immigrant shares. We make the practical and theoretical case for preserving the distinction between diversity—i.e., mixture—and non-White or immigrant shares—e.g., percentage Black or percentage foreign-born—, and we warn scholars about the dangers of using language and measures associated with diversity, especially in contexts, like North America and Europe, where diversity strongly overlaps with non-White or immigrant shares. On a practical front, the policy recommendations that follow from the claim “greater diversity is associated with less prosociality” are different from those that follow from the claim “greater non-White share is associated with less prosociality among Whites.” On a theoretical front, most studies of diversity rely on theories that are predicated on ingroup/outgroup shares—most commonly intergroup conflict/threat theory—rather than on diversity; the predictions implied by popular theories, however, contradict those implied by claims about diversity. Importantly, two empirical obstacles undermine our capacity to disentangle associations with diversity from associations with non-White or immigrant shares. The first stems from the underrepresentation of predominately non-White (or predominately immigrant) communities in the real world. The second concerns our ability to infer that individual attitudes and behavior are correlated with diversity from correlations between macro-level outcomes and diversity. Finally, we spell out the kinds of data required to draw empirically sound conclusions about associations between diversity and social outcomes. With an appendix by Daniel Lacker.