A SIMPLIFIED TIME‐SERIES ANALYSIS FOR EVALUATING TREATMENT INTERVENTIONS

TLDR

Time‑series analysis of behavioral data is increasingly supported, yet experts recommend 50–100 points per phase and complex models to extract serial dependence before assessing treatment effects. The authors propose a simple method for evaluating intervention effects using as few as eight data points per phase. The method relies on a straightforward calculation that can be performed with only eight observations per phase. The method’s calculations are simple enough to perform manually.

Abstract

Time‐series analysis procedures for analyzing behavioral data are receiving increasing support. However, several authorities strongly recommend using at least 50–100 points per experimental phase. A complex mathematical model must then be empirically developed using computer programs to extract serial dependency from the data before the effects of treatment interventions can be evaluated. The present discussion provides a simple method of evaluating intervention effects that can be used with as few as 8 points per experimental phase. The calculations are easy enough to do by hand.