Author Question: Explain autocorrelation and why its presence is an issue for the use of parametric statistical ... (Read 60 times)

corkyiscool3328 · **on:** Jun 20, 2018

Explain autocorrelation and why its presence is an issue for the use of parametric statistical analyses.

What will be an ideal response?

Question 2

Explain how change in level and trend from baseline to intervention phases are immediate indicators of a potentially successful intervention.

What will be an ideal response?

billybob123 · **Reply #1 on:** Jun 20, 2018

Answer to Question 1

Autocorrelation refers to the tendency of observations to correlate with another (rather than being truly independent observations of the dependent variable). In single subject research, one often expects to find trends in data which would indicate the presence of autocorrelation. A lag or time series analysis can definitively establish the presence or absence of autocorrelation. Its presence then precludes the use of parametric statistics because of the assumption that data points are independent of one another.

Answer to Question 2

First, change in level (particularly if there is little or no overlap of data from baseline to intervention phase) indicates the dependent variable is indicating a rapid and substantial change in responding with the introduction of the intervention. This suggests at least a rapid effect.
Second, change in trend in the dependent variable also can be an indicator of rapid effect. An ascending baseline that reverts to a descending trend (assuming reduction in behavior is desired) or vice versa (descending or flat baseline changes to ascending trend in a program to increase the target behavior) is an indicator of a rapid and positive effect.
Repeated measures and replication effects are necessary to fully establish the functional relationship.