PS296 Chapter Notes - Chapter 18: Repeated Measures Design, Multiple Comparisons Problem, Observational Error
Document Summary
Comparisons are made of scores within the same subjects thus are within-subjects designs. Similar to a t test for related samples but not restricted to only two measurement conditions. When using the protected t procedure to make multiple comparisons among data, we would use the mserror or ssw in our denominator. The t would have dferror degrees of freedom because mserror was used in plae of the pooled variance. Using the sd of the pre-test data as our denominator for dhat results in a lower value/effect size. When using the square root of mserror instead, we get a higher effect size because we are factoring out the correlation among scores (ex using depression scores, we would be factoring out individual differences in depression) The general suggestion is to use the sd of a control condition (ex pre test scores) for a more accurate effect size.