Let SAS® do the work: Correlation crossroads
Gilligan, T., Coon, C., Nelson, L., McLeod, L., & Yang, Y. (2010). Let SAS=65533 do the work: Correlation crossroads. In Paper 256-2010, SAS Global Forum, Seattle, WA, April 11-14, 2010,.
Patient-reported outcome (PRO) studies often require correlation analyses using multiple variable types. Producing tables that contain Pearson, polychoric, and polyserial correlations for ordinal and categorical variables is challenging in Base SAS®. This paper introduces a SAS macro called %PROCORR that routes variables to the appropriate correlation computation. The macro also identifies the appropriate output statistics for each type of correlation and creates a report-ready table. The analyses and output discussed were created with Base SAS Version 9.2 software and apply to analysts with macro experience.