Paradoxical results in and a possible extension of Necessary Condition Analysis (NCA)

Necessary Condition Analysis (NCA) is a quite new statistical method that has been used, for example, to analyze the association between intelligence and creativity. However, a significant result in NCA only indicates that the association between two variables, X and Y, is characterized by some unspecified type of non-randomness, and not necessarily that X is necessary for Y. The present simulation showed that the significance of necessity effects, as calculated through permutation in NCA, tend to increase with increased degree of sufficiency, even when the increase in sufficiency is achieved through decreasing the sample size. This is paradoxical for a method whose stated objective is to help researchers identify necessary but not sufficient conditions for some outcome of interest. One possible way to increase the specificity and usefulness of results obtained through NCA is to calculate not only the degree of necessity but also the degree of sufficiency and the difference between these two. The significance of the difference between degree of necessity and sufficiency can be estimated through bootstrapping. The simulation indicated desirable statistical characteristics for this extended version of NCA and it was applied on empirical data on intelligence, creative ability, and creative achievement.