In a 207‐country sample, we find that rule of law and corruption are both associated with a country's religious heritage, thereby partially explaining the correlation between religion and economic growth found in previous research. We also show that our results change when we control for some variables lacking data for all countries in the sample but that these differences are attributable to changes in sample composition rather than the effects of the control variables. Our research suggests that researchers doing cross‐country analysis should distinguish between the effects of adding a control variable and the resulting sample composition effects.