Meal pattern correlations: facts and artifacts.

The feeding behavior of 37 albino rats was used to demonstrate that significant postprandial correlations could be obtained from unordered raw data when actual intake was used as the measure of meal size. When meal duration was used instead, substantially lower correlations were obtained. It was further demonstrated that caution must be used when correlating ratio measures of feeding behavior. If the ratios are not independent then significant correlations can be obtained due to statistical bias alone. Implications and suggestions for the analysis of meal patterns were discussed.