A Simulation Study on the Performance of the Simple Difference and Covariance Adjusted Scores in Randomized Experimental Designs
A Monte Carlo simulation was conducted to examine the conditions under which the simple difference and residualized change scores were more or less powerful than each other, and if the two estimators produced a biased estimate of the average treatment effect. Five factors were manipulated in the des...
| Other Authors: | |
|---|---|
| Format: | Others |
| Language: | English English |
| Published: |
Florida State University
|
| Subjects: | |
| Online Access: | http://purl.flvc.org/fsu/fd/FSU_migr_etd-1975 |
| Summary: | A Monte Carlo simulation was conducted to examine the conditions under which the simple difference and residualized change scores were more or less powerful than each other, and if the two estimators produced a biased estimate of the average treatment effect. Five factors were manipulated in the design including: sample size, normality of the pretest and posttest distributions, average treatment effect, the correlation between pretest and posttest, and posttest variance. A 5 x 5 x 3 x 4 x 4 mostly-crossed design was run with 1,000 replications per condition, resulting in 905,000 conditions. Results suggested that the covariance adjusted score in an ANCOVA designed should be used in pretest-posttest randomized experiments when power is of interest. Additionally, neither estimator produced a biased estimate of the average treatment. === A Dissertation submitted to the Department of Psychology in partial fulfillment of the requirements for the degree of
Doctor of Philosophy. === Spring Semester, 2009. === November 14, 2008. === Monte Carlo, Randomized Experiment, Gain Score === Includes bibliographical references. === Christopher Schatschneider, Professor Directing Dissertation; Akihito Kamata, Outside Committee Member; Richard K. Wagner, Committee Member; Jon Maner, Committee Member; Young-Suk Kim, Committee Member. |
|---|