Proposing Necessary but Not Sufficient Conditions Analysis as a Complement of Traditional Effect Size Measures with an Illustrative Example

Even though classic effect size measures (e.g., Pearson's r, Cohen's d) are widely applied in social sciences, the threshold used to interpret them is somewhat arbitrary. This study proposes necessary condition analysis (NCA) to complement traditional methods. We explain NCA in light of the current limitations of classical techniques, highlighting the advantages in terms of interpretation and translation into practical terms and recognizing its weaknesses. To do so, we provide an example by testing the link between three independent variables with a relevant outcome in a sample of 235 subjects. The traditional Pearson's coefficient was obtained, and NCA was used to test if any of the predictors were necessary but not sufficient conditions. Our study also obtains outcome and condition inefficiency as well as NCA bottlenecks. Comparison and interpretation of the traditional and NCA results were made considering recommendations. We suggest that NCA can complement correlation analyses by adding valuable and applicable information, such as if a variable is needed to achieve a certain outcome level and to what degree.

Automated summary

This study proposes the necessary condition analysis (NCA) as a complement to classic effect size measures in interpreting data. By testing the link between independent variables and a relevant outcome in a sample, the study compares and interprets the traditional and NCA results, and suggests that NCA can provide valuable and applicable information in correlation analyses.