creator Collins, Michael W. advisor Morris, Scott B. Meta-analysis is a statistical method for aggregating summary results from archival research. Traditional meta-analysis procedures that involve repeated measures assume homogeneous variances in pretests and posttests, an assumption which is often violated in training evaluation contexts due to variance inflation in posttests. A new meta-analytic procedure that allows for heterogeneous variance in pretests and posttests was tested in the context of leadership and managerial training programs. Application of the new method revealed greater sampling variance and less between-study variance than traditional meta-analysis with little effect on estimation of the mean effect size. However, the procedure screened out a large number of studies due to the intensive data requirements, which limited the number of usable studies for the meta-analysis. Submitted by Lance Garrison (garrlanc@my.dom.edu) on 2012-03-19T18:21:03Z No. of bitstreams: 2 Collins_Signed_Title_Page.pdf: 87311 bytes, checksum: d1c97d3b92b99332b5ff1ea876df0636 (MD5) Dissertation_Collins-Michael_Final.pdf: 740965 bytes, checksum: f3fec574bcc3711d56685be7482f8477 (MD5) Made available in DSpace on 2012-03-19T18:21:03Z (GMT). No. of bitstreams: 2 Collins_Signed_Title_Page.pdf: 87311 bytes, checksum: d1c97d3b92b99332b5ff1ea876df0636 (MD5) Dissertation_Collins-Michael_Final.pdf: 740965 bytes, checksum: f3fec574bcc3711d56685be7482f8477 (MD5) Previous issue date: 2011-12 Ph.D. in Psychology, December 2011 2011-11 2011-12 http://hdl.handle.net/10560/2648 en Dissertation born digital application/pdf In Copyright http://rightsstatements.org/page/InC/1.0/ Restricted Access PSYCH / Institute of Psychology Illinois Institute of Technology Affiliated department