Logo image
Problems in the estimation of autocorrelation in brief time series and some implications for behavioral data
Journal article   Peer reviewed

Problems in the estimation of autocorrelation in brief time series and some implications for behavioral data

T A Matyas and Ken Greenwood
Behavioral Assessment, Vol.13(2), pp.137-157
1991

Abstract

Reexamined the lag 1 autocorrelations of 182 time series collected from the initial baseline phases of single-case designs by T. A. Matyas and K. M. Greenwood (1985). The 182 values of r₁ were tested for significance against the values provided by the simulation. Analyses revealed that 42 (23.1%) time series showed statistically significant autocorrelation; 32 were positive autocorrelations and 10 negative. Application of these findings, within the meta-analytic approach of B. E. Huitema, provided a convincing demonstration against his conclusion that autocorrelation in behavioral data is a myth.

Details

Metrics

333 Record Views
Logo image