Adding some rigor to Necessary Condition Analysis

Necessary condition analysis (NCA) is a statistical method where a so-called necessity effect is estimated as the size of the (semi-) empty space in the upper-left corner in a XY-plot. A (semi-) empty space in the upper-left corner is taken to indicate that a low value on X precludes a high value on Y. However, previous studies have shown that analyses with NCA are susceptible to spurious findings. Here, we add to this body of research and show that necessity effects in NCA may be due to correlations between X and Y rather than due to a genuine necessity effect of X on Y. Moreover, we present a method for estimating “ranges of spuriousness” and argue that empirically estimated necessity effects should be above this range if claiming that X is necessary for Y. If the empirical necessity effect falls within the range of spuriousness, it is not significantly stronger than can be expected due to the correlation between X and Y. In an empirical application, we found that 20 of 25 scrutinized necessity effects reported in the literature fell within the range of spuriousness. In order to add some much-needed rigor to analyses with NCA, we recommend researchers using the method to scrutinize their findings by estimating a range of spuriousness.