Winston Bennett

Conducting Meta-Analysis Using SAS

Contents: Preface. The Theory of Meta-Analysis--Sampling Error and the Law of Small Numbers. Meta-Analysis of Effect Sizes. Meta-Analysis of Correlations. Outliers in Meta-Analytic Data. Summary and Guidelines for Implementing a Meta-Analysis. Appendices: Reference and Information Sources for the Behavioral and Social Sciences. Equation for Computing the Pooled Within-Group Standard Deviation. Conversion and Transformation Equations. Upper Percentage Points for the Chi-Square Distribution.

Dyadic Versus Individual Training Protocols: Loss and Reacquisition of a Complex Skill

Winfred Arthur, Jr., and Eric Anthony DayTexas AM Winston Bennett, Jr., Armstrong Laboratory, HumanResources Directorate, U.S. Air Force, Brooks Air Force Base,Texas.This research was sponsored under contract to Winfred Ar-thur, Jr., from the U.S. Air Force Armstrong Laboratory, HumanResources Directorate, Technical Training Research Division,Brooks Air Force Base, Texas. The views expressed herein arethose of the authors and do not necessarily reflect the officialposition or opinion of their respective organizations.We thank the following individuals for their assistance incollecting the data: Arlette DeCuir, Julie Goodwin, John Lane,Robyn Maldegen, Michael Miller, Laura Thomas, and TravisTubre.Correspondence concerning this article should be addressedto Winfred Arthur, Jr., Department of Psychology, Texas A&MUniversity, College Station, Texas 77843-4235. Electronic mailmay be sent via Internet to wea@psyc.tamu.edu.been lost. Likewise, disaster teams may go years withoutevacuating residents from affected areas, managing evacu-ation routes, and rescuing survivors, and yet they are ex-,pected to perform at high proficiency levels should thesituation arise.One of the most pervasive weaknesses of the skill-lossliterature is the lack of attention given to the phenomenaof skill acquisition (Arthur, Bennett, Stanush, & McNelly,in press). R. A. Schmidt and Bjork (1992), for example,criticized the educational and training literature for treat-ing learning (i.e., skill acquisition) and retention (the op-posite of which is skill loss) as two separate phenomenathat have been studied independently by different scien-tists using different methods in different laboratories. Forinstance, they showed that those protocols that maximizeskill acquisition may not necessarily lead to the best reten-tion compared with protocols that degrade speed of acqui-sition. Thus, these authors argued that acquisition andretention are really inseparable and need to be consideredtogether in investigations of skill acquisition. Conse-quently, the objective of our study was to evaluate andcompare the effectiveness and efficiency of a dyad-basedtraining protocol with that of a standard, individual-basedprotocol in terms of complex skill acquisition, loss, andreacquisition after a nonpractice, posttraining interval.The performance task used in our study was the videogame Space Fortress (Human Engineering Laboratory,Technion-Israel Institute of Technology), which has astrong history as an excellent research tool in the area of

Replication and Extension of Models of Supervisory Job Performance Ratings

This study tested causal models of job knowledge, job proficiency, and supervisor ratings in a sample (N = 838) of U.S. Air Force enlisted personnel. Results indicated that (a) effects of ability and experience were linear, not interactive; (b) different conceptualizations of experience play somewhat different causal roles in the determination of job knowledge, job proficiency, and supervisory ratings; (c) general support for the mediational roles of job knowledge and job proficiency; and (d) supervisory ratings reflect both technical (task-based) and motivational (contextual) aspects of performance.

Conducting Meta-Analysis Using the Proc Means Procedure in Sas

The methodology for using the PROC MEANS procedure in SAS to perform the data computations called for by the Hunter and Schmidt approach to meta-analysis is presented. Using SAS for meta-analytic data analysis offers several potential advantages including ease of use, flexibility, and widespread availability. Specific procedures are presented for calculating summary statistics and then correcting these summary statistics for sampling error and other statistical artifacts such as range restriction and measurement error. One particularly attractive feature of this approach is the ease with which potential moderator variables can be analyzed simply by adding a few additional statements to the SAS program.

Good News: Work Sample Administrators' Global Performance Judgments are (About) as Valid as We've Suspected

Data obtained on over 1,500 first-term U.S. Air Force enlisted personnel indicated that work sample administrators global ratings of work sample performance substantially reflect actual ratee behavior in the work sample, and not potentially biasing factors (e.g., race, gender, amount of recent experience), supporting the folk wisdom that these global performance judgments are, in fact, valid and unbiased measures of performance. Good news!