Canonical Correlation Analysis of Survey Data: The SurveyCC R Package

Classic Canonical Correlation Analysis (CCA) is a popular statistical method that allows for the analysis of the associations between two sets of variables. However, currently it cannot be applied following the published methodological documentation to data sets collected using complex survey design (CSD), which includes factors, such as replicate weights, clusters, and strata, that are critical for the accurate calculation of the statistical significance of any correlation. To close this gap, we have developed the Survey CC algorithm and implemented it in an R package. We describe the theoretical foundations of our algorithm and provide a detailed report of the options of the function that performs it. Moreover, the application of our newly developed method to several national survey data sets shows the differences in conclusions that can be reached if the CSD elements are not taken into consideration during the calculation of the statistical significance of the canonical correlations.