ANALYZING PSUEDO-TIME SERIES INTERVENTIONS

A time series evaluation design of use in analyzing certain kinds of "case-flow "programs is described. Because it combines features of both conventional time-series designs and nonequivalent comparison group quasi-experiments, we refer to it as a "pseudo-time series " or PTS design. Statistically, PTS designs resemble replicated surveys that independently sample some population at successive time points. The substantive analogy is a program that services a rapidly changing clientele the members of which usually enter and exit the program on only one occasion. This article describes the distinguishing features of PTS designs, outlines a methodological strategy for analyzing PTS data, and shows how this strategy can be used to overcome design weaknesses in certain case-flow evaluation studies. ime series intervention analysis has grown steadily in popularityTCampbell and Stanley (1963) first called attention to thesince comparative advantages of the interrupted time series (ITS) quasi-experiment over two decades ago. Increasingly sophisticated modeling techniques (Glass et al., 1975; McCleary and Hay, 1980) have helped bolster this popular image (at least in technical circles) as have persisting doubts about the feasibility of experimental methods in field settings (Mark and Cook, 1984). ITS analysis has also proven to be a versatile yet unobtrusive impact-assessment tool, well suited to analyzing the growing volume of sequential performance data (weekly payroll figures, monthly attrition rates, annual expenditures, and so on) routinely collected by government agencies and private corporations. AUTHOR’S NOTE: This article was first presented in revised form at the 1985 annual meeting of the Evaluation Research Society. For helpful comments on an earlier draft o