An examination of methods used to generate daily group scores from single-item-per-subject data collected in intensive time-series designs

Authors


Abstract

This study examines three methods which can be used to pool single-item-per-subject data collected in intensive time-series studies, to determine if method of pooling has an effect on subsequent data analysis. The methods examined were based on simple averaging, difficulty weightings of averages, and the application of the Rasch logistic model. Analyses were conducted which examined regression results obtained when the pooled scores of groups of students were regressed by day. Results indicate that the three methods of pooling do not significantly alter subsequent analysis, though the case is made that a pooling procedure based on the Rasch logistic model is the most heuristically sound.

Ancillary