Fall 2007                                 Multivariate Statistics Seminar                                  81691-001

 

Instructor:                   Kristin D. Mickelson, Ph.D.

Office:                          Kent Hall Addition 332

Office Phone:              672-2253

Office Hours:              Mondays & Wednesdays 1:30-4:00pm and by appointment

Email:                          kmickels@kent.edu

 

Required Texts: 

 

Kline – Principles and Practices of Structural Equation Modeling, 2nd edition (2005).  Guilford Press.

 

Singer & Willett, Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence (2003).  Oxford University Press.

 

Additional articles will be assigned as supplemental reading for various techniques (available at the UPS Store).

 

Course goals: This seminar is intended to cover two major multivariate statistical techniques currently utilized in psychology: structural equation modeling (SEM) and hierarchical linear modeling (HLM) – some areas of psychology will make more use of these techniques, but all areas require a basic understanding of them. The goal for students is to achieve a conceptual understanding of each statistical technique, be able to apply the correct analysis to any data set, properly conduct and interpret the output of statistical computer packages, and understand and critique scientific papers that utilize such techniques.

 

The term conceptual understanding means that the student gains knowledge of the functional nature of each technique – in other words, what it is used for, how it is used, the assumptions behind it, and the information it provides about the data.  The course will not cover the derivations of the formulas, but rather will use the formulas to assist in understanding the way each technique works.

 

It is important that you read the chapters/articles before and after each class period – this sounds redundant, but reading before the class period will provide a base of knowledge about the terms while re-reading after the class period will help to reiterate the details.

 

Statistical Computer Packages:  Several computer packages will be used in this course, including EQS, HLM and SPSS.  Each package will be available for student use on the computers in the grad student lab throughout the semester.  In addition, several class periods will be held in the computer lab to help students learn to use the programs – students will be notified the prior class period.

 

Grading:  As the goal of the course is to provide an understanding that students can transfer to their own research and reading, no in-class exams will be given.  Instead, grading will reflect your understanding and application of the techniques.  First, each student will pose a multivariate research question, conduct the analyses on an existing data set, and interpret the results.  This project will be written up as a 10-page paper (40%).  In addition, there will be one take-home exam (50%) at the end of the semester; this exam will be more conceptual in nature focusing on the functional nature of the techniques as defined above.  Finally, several small homework assignments will be given - usually in the context of the computer lab sessions.  These will be graded pass/fail and worth 10% of the course grade.

 


 

Analysis Project (40% of final grade):

 

With assistance from the instructor, choose a multivariate research question and analyze using an existing data set (either your own or one that I have access to).  For the 10-page paper, briefly state the question and hypothesis (the Introduction should be no more than 2 pages and should conclude with a statement of your hypothesis).  Next, your Methods section should be the same as you would write for a manuscript – detail the sample, procedures, measures (including how they were scored and alphas), and overview of analysis.  The Results section will be a more detailed version of what you would include in a manuscript; walk through the steps taken to conduct the most appropriate analysis to answer the question and test the hypothesis – this includes what you did with missing data and outliers, how you tested the relevant assumptions and corrected any violations, as well as presenting the results of the proposed analyses.  Interpret the analyses and results briefly in the Discussion section (no more than one page).  Also, in the Discussion state what can be concluded from the analyses you conducted and what cannot be concluded based on your specific dataset (i.e., the pros/cons of your analysis).  Finally, briefly discuss the pros/cons of one alternative analysis that you could have conducted to answer the hypothesis.

 

Paper should be typed, double-spaced, APA format, 12-point font, and 1" margins.  Tables and/or figures should be included in the paper, but are not counted toward the page limit.  In addition, attach a copy of the syntax and output from the computer program used for the project as an appendix.

 

Final Exam (50% of final grade):

The exam will be a take-home cumulative exam that covers the conceptual nature of the techniques discussed in class.  The exam will be an essay format, open-note, open-book.  Students are not allowed to work on the exams together and must sign-out the exam and return it 36-hours later.  The final answers should be typed, not hand-written.

 

Homework Assignments (10% of final grade):

 

Throughout the semester, I will give several short homework assignments, usually in conjunction with the computer lab sessions.  These assignments will help in making sure that you understand the concepts and computer programs.  Grading will be pass/fail.

 

This syllabus schedule is tentative.

 



Background and Review

 

Chapter:

 

Licht, M. H.  (1995).  Multiple regression and correlation. In L. G. Grimm & P. R. Yarnold (Eds.), Reading and Understanding Multivariate Statistics (pp. 19-64).  Washington, D. C.: APA.

 

Wright, R. E.  (1995).  Logistic regression. In L. G. Grimm & P. R. Yarnold (Eds.), Reading and Understanding Multivariate Statistics (pp. 217-244).  Washington, D. C.: APA.

 

Articles:

 

Coulton, C., & Chow, J.  (1992).  Interaction effects in multiple regression.  Journal of Social Service Research, 16, 179-199.

 

Frazier, P. A., Tix, A. P., & Barron, K. E.  (2004).  Testing moderator and mediator effects in counseling psychology research.  Journal of Counseling Psychology, 51, 115-134.

 

Venter, A., & Maxwell, S. E.  (2000).  Issues in the use and application of multiple regression analysis.  In H. E. A. Tinsley & S. D. Brown (Eds.), Handbook of applied multivariate statistics and mathematical modeling (pp. 151-182).  San Diego, CA: Academic Press.

 

 


Structural Equation Modeling

 

Overview and Basic Concepts of SEM

 

Kline – Chapter 1 (pp. 9-16 – Section 1.5 Family Values)

Kline – Chapter 2 (pp. 20-44)

 

Hoyle, R. H., & Smith, G. T.  (1994).  Formulating clinical research hypotheses as structural equation models: A conceptual overview.  Journal of Consulting and Clinical Psychology, 62, 429-440.

 

Musil, C. M., Jones, S. J., & Warner, C. D.  (1998).  Structural equation modeling and its relationship to multiple regression and factor analysis.  Research in Nursing and Health, 21, 271-281.

 

Applications of SEM in Psychological Research

 

Dilalla, L. F. (2000).  Structural equation modeling: Uses and issues.  In H. E. A. Tinsley & S. D. Brown (Eds.), Handbook of Applied Multivariate Statistics and Mathematical Modeling (pp. 439-463).  San Diego, CA: Academic Press. (pp. 439-463).

 

MacCallum, R. C., & Austin, J. T.  (2000).  Applications of structural equation modeling in psychological research.  Annual Review of Psychology, 51, 201-226.

 

Russell, D. W., Kahn, J. H., Spoth, R., & Altmaier, E. M. (1998).  Analyzing data from experimental studies: A latent variable structural equation modeling approach. Journal of Counseling Psychology, 45, 18-29.

 

Data and Methodology Issues for SEM

 

Kline – Chapters 3 & 4 (pp. 45-90)

 

Jackson, D. L.  (2003).  Revisiting sample size and number of parameter estimates: Some support for the n:q hypothesis.  Structural Equation Modeling, 10, 128-141.

Path Analysis

 

Kline – Chapters 5 & 6 (pp. 93-164)

 

Klem, L.  (1995).  Path analysis.  In L. G. Grimm & P. R. Yarnold (Eds.), Reading and Understanding Multivariate Statistics (pp. 65-98).  Washington, D. C.: APA.

 

MacKinon, D. P., Krull, J. L., & Lockwood, C. M.  (2000).  Equivalence of the mediation, confounding and suppression effect.  Prevention Science, 1, 173-181.

 

Shrout, P. E., & Bolger, N.  (2002).  Mediation in experimental and nonexperimental studies: New procedures and recommendations.  Psychological Methods, 7, 422-445.

 

Measurement Models & CFA

 

Kline – Chapter 7 (pp. 165-208)

 

Kenny, D. A., & McCoach, B. (2003).  Effect of the number of variables on measures of fit in structural equation modeling.  Structural Equation Modeling, 10, 333-351.

 

Structural Models

 

Kline – Chapter 8 (pp. 209-234)

 

MacCallum, R. C.  (1995). Model specification: Procedures, strategies, and related issues.  In R. H. Hoyle (Ed.), Structural Equation Modeling: Concepts, Issues, and Applications (pp. 16-36).  Thousand Oaks, CA: Sage.

 

Tomarken, A. J., & Waller, N. G.  (2003).  Potential problems with “well fitting” models. Journal of Abnormal Psychology, 112, 578-598.

 

Nonrecursive Models and Multi-Sample SEM

 

Kline – Chapters 9 & 11 (pp. 237-262; 289-312)

 

Benbenishty, R., Astor, R. A., Zeira, A., & Vinokur, A. D.  (2002).  Perceptions of violence and fear of school attendance among junior high school students in Israel.  Social Work Research, 26, 71-87.

 

Farrell, A. D.  (1994).  Structural equation modeling with longitudinal data: Strategies for examining group differences and reciprocal relationships.  Journal of Consulting and Clinical Psychology, 62, 477-487.

 

Fincham, F. D., Beach, S. R. H., Harold, G. T., & Osborne, L. N.  (1997).  Marital satisfaction and depression: Different causal relationships for men and women?  Psychological Science, 8, 351-357.

 

Vinokur, A. D., & Schul, Y.  (2002).  The web of coping resources and pathways to reemployment following a job loss.  Journal of Occupational Health Psychology, 7, 68-83.

 

Writing Up SEM Analyses

           

Kline – Chapter 12 (pp. 313-324)

 

Hoyle, R. H., & Panter, A. T.  (1995).  Writing about structural equation models.  In R. H. Hoyle (Ed.), Structural Equation Modeling: Concepts, Issues, and Applications (pp. 158-176).  Thousand Oaks, CA: Sage.

 

McDonald, R. P., & Ho, M. R.  (2002).  Principles and practice in reporting structural equation analyses.  Psychological Methods, 7, 64-82.

 

Raykov, T., Tomer, A., & Nesselroade, J. R. (1991).  Reporting structural equation modeling results in Psychology and Aging: Some proposed guidelines.  Psychology and Aging, 6, 499-503.

 

Thompson, B.  (2000).  Ten commandments of structural equation modeling.  In L. G. Grimm & P. R. Yarnold (Eds.), Reading and Understanding More Multivariate Statistics (pp. 261-284).  Washington, D. C.: APA.

 

 


Hierarchical Linear Modeling

 

Introduction to Longitudinal Data Analysis

 

Singer & Willet – Chapters 1 & 2 (pp. 3-44)

 

Weinfurt, K. P.  (2000).  Repeated measures analyses: ANOVA, MANOVA, and HLM.  In L. G. Grimm & P. R. Yarnold (Eds.), Reading and Understanding More Multivariate Statistics (pp. 317-362).  Washington, D. C.: APA.

 

Wendorf, C. A. (2002).  Comparisons of structural equation modeling and hierarchical linear modeling approaches to couples’ data.  Structural Equation Modeling, 9, 126-140.

 

 

Basic Concepts & Principles

 

Singer & Willett – Chapter 3 (pp. 45-74)

 

Bryk, A. S., & Raudenbush, S. W.  (1987).  Application of hierarchical linear models to assessing change.  Psychological Bulletin, 101, 147-158.

 

Nezlek, J. B. (2003).  Using multilevel random coefficient modeling to analyze social interaction diary data.  Journal of Social and Personal Relationships, 20, 437-469.

 

Nezlek, J. B., & Zyznuewski, L. E.  (1998).  Using hierarchical linear modeling to analyze grouped data.  Group Dynamics, 2, 313-320.

 

 

Illustrations of HLM

 

Singer & Willett – Chapters 4 & 5 (pp. 75-188)

 

DeLongis, A., Capreol, M., Holtzman, S., O’Brien, T., & Campbell, J.  (2004).  Social support and social strain among husbands and wives: A multilevel analysis.  Journal of Family Psychology, 18, 470-479.

 

Helson, R., Jones, C., & Kwan, V. S. Y.  (2002).  Personality change over 40 years of adulthood: Hierarchical linear modeling analyses of two longitudinal samples.  Journal of Personality and Social Psychology, 83, 752-766.

 

Raudenbush, S. W., Brennan, R. T., & Barnett, R. C.  (1995).  A multivariate hierarchical model for studying psychological change within married couples.  Journal of Family Psychology, 9, 161-174.

 

 

FINAL EXAM TO BE TAKEN DURING ANY 36-HOUR PERIOD ON MONDAY, TUESDAY, OR WEDNESDAY OF FINALS WEEK