Correlation analysis is one of the most popular analytic methods for discovering the relationship between two variables in consumer health data. Nevertheless, the simple sample correlation techniques (e.g., Pearson’s or Spearman’s correlation) assume the independence of the observations, which is often violated when there are repeated measures from each subject (also called nested data). In this paper, we compare three methods for correlation calculation on nested consumer health data: (1) repeated measure correlation (rmcorr), (2) subject mean correlation (smcor), and (3) simple sample correlation (sscor). We contend that rmcorr is a more appropriate technique for the correlation analysis of nested data as it could explicitly capture the within-subject variations. In contrast, the smcor and sscor, which could only capture the cross-subject variations, may produce misleading and faulty results.