Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13

Applied Psychometrics Using SAS

reviewed by Grant B. Morgan & Courtney A. Moore - November 08, 2016

coverTitle: Applied Psychometrics Using SAS
Author(s): W. Holmes Finch, Brian F. French, and Jason C. Immekus
Publisher: Information Age Publishing, Charlotte
ISBN: 1623966949, Pages: 66, Year: 2014
Search for book at Amazon.com

By now, I suspect that many people in the field of educational research and associated quantitative specializations have come to learn, as I have, that the works produced by Holmes Finch and Brian French are of the highest quality as well as both practically and technically sound. In Applied Psychometrics Using SAS, Finch and French, along with added co-author Jason Immekus, have produced a text that any practitioner and/or young scholar learning and applying psychometric analysis using Statistical Analysis System (SAS) must have. In fact, this is the text I had hoped existed a decade ago when I was learning psychometric analysis in graduate school.

The text is divided into nine chapters that are ordered in what I would contend is the way that allows the topics to most effectively build on each other. Chapter One is an introductory chapter to psychometric concepts, including measurement basics, classical test theory, and item response theory. Chapter Two provides readers with a discussion of classical item analysis (i.e., difficulty and discrimination), and Chapter Three naturally follows with a presentation of reliability within the classical theory framework along with estimates for internal consistency, split-halves, and test-retest. Generalizability theory is covered in Chapter Four, which conceptually follows the information presented in Chapter Three via the partitioning of the error term into its various sources. The longest chapter in the book, Chapter Five, focuses on validity, paying particular attention to the types of validity supported by statistical evidence. Supposing acceptable evidence for an instrument’s reliability and validity, one might next consider scoring issues. Chapter Six offers different types of scoring options and some considerations for their use. The last three chapters in the book center on more advanced topics: item response theory (IRT; Chapter Seven), differential item functioning (Chapter Eight), and non-IRT equating (Chapter Nine). The chapter on IRT is nearly as long as the validity chapter, which is not surprising considering the authors introduce several models for both dichotomous and polytomous data. The inclusion of the polytomous models was a welcome surprise as I enjoyed reviewing the PROC NLMIXED code that could produce the models presented.

In short, this text delivers on precisely that which it sets out to: emphasizing applications of psychometric and measurement concepts via SAS® software following brief discussions of theoretical/technical issues related to each topic. As such, the presentation of material is largely focused on topics and issues for which analysis is appropriate and/or necessary. For example, in the chapter on validity, the authors introduce content-, criterion-, and construct-related validity, yet content-related validity is given very little space (i.e., 1–2 paragraphs) compared to criterion- (i.e., more than 16 pages) and construct-related validity (i.e., more than 33 pages). Upon reflecting on the allocation of attention, I believe this to be entirely consistent with the stated purpose/goal of the book. Content-related validity evidence tends not to be statistical in nature; hence, SAS® software would not likely be used. The authors clearly state that the book could be used to supplement a more theoretical text. Given the focus, I paid particular attention to the citations used to support the substantive portions of the book. The citations include both seminal and important current works (as of the time it was published), which provides a comprehensive set of references for readers who have focused interest on the theoretical/technical issues beyond the scope of the book. Furthermore, the content presented by the authors is wholly sufficient for establishing a solid foundation and working knowledge of psychometric concepts for students and/or practitioners who choose to pursue more theoretical/technical sources.

As noted above, this is the text I had hoped for as a graduate student learning psychometrics and technical aspects of measurement using SAS® software as the primary platform. I spent countless hours writing and testing code through trial and error, often experiencing tremendous frustration, to produce many of the types of analyses presented in this book. The SAS code is clearly provided with the accompanying output. The authors chose not to explain the SAS code command-by-command and line-by-line because the book assumes basic working knowledge of SAS® software. Explanations of relevant pieces of output are provided in-text. The explanations are incredibly readable, which will likely empower readers to feel that they can interpret their own output from code they have adapted.

The authors should also be commended for including many SAS® macros (e.g., %ALPHA_CI: Kromrey, Romano, & Hibbard, 2008; %parallel: Kabacoff, 2003) that are not built into the SAS/BASE or SAS/STAT software. These macros provide invaluable information for certain psychometric analyses. On one hand, readers should be incredibly grateful to the authors for compiling these macros and for providing explanations, considerations, and interpretations; on the other hand, many of these macros are very long and quite detailed. Readers reproducing the code will inevitably encounter errors due to typos or some other type of user error. A companion website where users could copy/paste code or download SAS® program files would greatly alleviate the burden of use and may ultimately increase the impact of the book. Given the proliferation of the R software, which (a) is capable of all psychometric analysis, (b) is open source, and (c) has oodles of code available online, availability of SAS® program files or code electronically/online would enhance the book’s usability and competitiveness.

One topic that was not included in the text that would further increase the usability of the book was IRT-based equating. The decision to exclude this topic may have been due to the complexity of SAS® code that would be necessary for conducting IRT-based equating or to the fact that the topic is slightly more advanced than what the authors set out to cover in this text. A second reflection is that the chapter on DIF could be tied back to the chapter on validity a bit more strongly.


(Morgan) As a faculty member who learned these analyses originally using SAS® software, I found this book exceptionally well written and consider it a must read for anyone who is learning or applying psychometrics using SAS® software. The book offers readers technical precision, important considerations based on current research, and numerous examples of functional SAS® code, its associated output, and guidance for interpretation. This book is masterfully written and delivered. It should be on the shelf of any SAS® software user who is learning and/or conducting psychometric analysis.

(Moore) As a graduate student studying quantitative methods, I want to highlight the book’s exceptional readability. While the authors provide clear and simple explanations and recommendations, they also promote a plethora of supplemental resources. Particularly meaningful to me is their intentionality in introducing each new topic with an explanation of the importance and usefulness of the subsequently explained procedure. Though I have little experience using SAS® software, I thoroughly enjoyed, and benefited from, this text.


Kabacoff, R. (2002). Determining the dimensionality of data: A SAS® macro for parallel analysis. Proceedings of the 27th Annual Meeting of the SAS Users Groups International. Cary, NC: SAS Institute.

Kromrey, J. D., Romano, J., & Hibbard, S. T. (2008, March). ALPHA_CI: A SAS macro for computing confidence intervals for coefficient alpha. Paper presented at the SAS Global Forum, San Antonio, TX.

Cite This Article as: Teachers College Record, Date Published: November 08, 2016
https://www.tcrecord.org ID Number: 21721, Date Accessed: 12/2/2021 2:45:07 PM

Purchase Reprint Rights for this article or review
Article Tools
Related Articles

Related Discussion
Post a Comment | Read All

About the Author
  • Grant Morgan
    Baylor University
    E-mail Author
    GRANT B. MORGAN is an associate professor in the Department of Educational Psychology at Baylor University. His research interests include latent variable models, psychometrics, clustering/classification, and nonparametric statistics. He teaches courses in measurement and evaluation, item response theory, psychometric theory, latent variable models, research methods, and experimental design, all at the graduate level.
  • Courtney Moore
    Baylor University
    E-mail Author
    COURTNEY A. MOORE is a student in the quantitative methods master's program in the Department of Educational Psychology at Baylor University. Her applied research interests generally revolve around improving people's livelihood, with particular interest in adoption, foster care, and hunger.
Member Center
In Print
This Month's Issue