| Abstract: | There is a large body of literature devoted to critiquing the adequacies of various approaches to analyzing data from quasi-experiments. It has slowly become clear that any discussion of approaches to analysis of quasiexperiments is incomplete unless certain assumptions are made about data resulting from growth in the absence of treatment effects. One popular assumption, and for which there exists appropriate analyses, is the fan spread hypothesis. The utility of the fan spread hypothesis for educational data is explored through relationships to a large class of continuous growth models. It is shown that the fan spread hypothesis is both more and less restrictive than past literature would suggest. Implications for data analysis are also discussed. |
