188-31: Characterizing Patterns of Longitudinal Data Completeness through Successive Refinement

The analysis of longitudinal data with multiple assessment periods provides many challenges with respect to completeness and usability of data. Patterns of data completeness over time can vary greatly, and the number of possible longitudinal patterns increases exponentially with the number of assessment periods. With a great number of possible completeness patterns, questions arise as to which patterns have enough complete data to be usable. Some patterns occur much more commonly than others, and through the process of successive refinement, rules can be developed to accommodate these more common patterns of data completeness. Common patterns of complete data include “pure dropout” cases, “a little missing” cases, and “barely any data” cases. These cases will be explored and examples provided. Possible rules will be described to help determine where to draw the line with respect to the completeness of data and its subsequent usability.