SYLLABUS
Wed. 1-4pm
Room 41-235

Professor Carol S. Aneshensel
Office: 21-268 CHS
Telephone: 310-825-7479
Email: anshnsl@ucla.edu
Office Hours: Monday 2-4 & by appointment

Special Reader: Susan Albers, alber@ucla.edu

Reading:

Aneshensel, C.S. 2002. Theory-Based Data Analysis for the Social Sciences. (Thousand Oaks, CA: Pine Forge). Available at the HS Bookstore. Required.

Course Reader: Available at Westwood Copies. 1001 Gayley Ave. Suite 104. 310-208-3233. Required.

Resources: Readings that are on the class web site: www.ph.ucla.edu/class/chs/chs219. Optional.

Course Description

This course focuses on the translation of theory into a data analytic plan; the application of this analytic plan to real data; and, the interpretation of the results obtained through multivariate analysis. Quantitative data are analyzed using a range of multivariate techniques, such as linear multiple regression and logistic regression. Each student thoroughly analyzes a theoretical problem using his or her own quantitative data.

Course Objectives

1. To develop an understanding of how to translate theory into a set of research questions or hypotheses that can be tested by various statistical techniques.

2. To become aware of similarities in analytic strategies across different multivariate statistical techniques.

3. To become skilled at articulating the substantive meaning of multivariate statistical results.

4. To put these skills into practice by writing a paper that reports the results of an independent multivariate analysis of data.

Course Requirements

1. Each student will analyze thoroughly a theoretical issue utilizing his or her own quantitative data set.

This is not an exercise in number crunching, but the application of a theory that is relevant to the data. Ideally, the data are very close to the area in which you would like to work, for example, data that might be useful for a dissertation or as preliminary work for a dissertation. Most students conduct secondary analysis of existing data, often a public use data set or one provided by their advisors. You do not have to collect the data yourself. The data set has to be sufficient to sustain multivariate analysis. This issue is usually one of sample size. The distributions of key variables are also important: if some of these are badly skewed, the planned analysis may be compromised. You select the statistical package you want to use for the course, and you may use more than one.

2. At each class session, students will present the results of their analysis and give its substantive interpretation.

So that everyone can follow the discussion, students should bring enough copies of relevant output and study materials. Be selective; do not bring every piece of output. Part of the assignment is selecting what information is important and deciding how to present it to others. However, you should bring your copy of all output so that you can answer questions that arise during discussion.

3. At each class session, the assigned write-up of results (see class listing below) will be discussed in small groups.

Bring enough copies for the people in your group and one copy for me. The goal of this discussion is to practice written communication and to receive feedback about whether others understand it.

4. At each class session, students will comment on in a positive constructive manner the work presented by other students.

Critique means identify the positive and negative aspects of the work. Ignoring things you find problematic in someone's work is not helpful to your colleagues because they are left vulnerable to criticism from others. Oftentimes these problems can be addressed, which strengthens the work. However, it is unnecessary and unacceptable to be nasty or personal in your criticism. Focus on the work and how it can be improved.

5. Students will prepare a manuscript of journal length and quality describing the research they have conducted during the course. In addition to describing study results, this paper requires a literature review with substantial outside reading.

The major sections are:

Statement of the problem and its significance, which entails a review of relevant literature;

Methods, including a description of data collection and analysis techniques;

Results, which must include tables and may include figures as needed;

Discussion, which includes interpretation of the results, and a discussion of strengths and limitations; and,

A reference list of materials cited in the text.

In general, journal articles are limited to 4-5 tables and 1-2 figures, and you should follow this guideline. Use published articles as a guide to format, especially for tables and figures. Put the tables and any figures on separate pages at the end of the text; do not put them into the text. The text is usually about 20 pages double-spaced pages. Citations in the text to the relevant literature should use the (name, year: page if a direct quote) format. The references should give the complete source including author, title, date, journal or publishing house/city, and page numbers.

DUE MONDAY OF FINALS WEEK.

DELIVER TO MY OFFICE OR A-LEVEL MAILBOX.
DO NOT EMAIL OR FAX.

Course Grading

Grades in course are based 50% on class participation and 50% on final paper.

Class Outline

Week TOPICS
Reading
Assignments

1 INTRODUCTION: THEORY, DATA ANALYSIS, AND STATISTICS
Definition of data analysis; statistics and analysis; translation of theory into an analytic strategy; inductive and deductive reasoning; operationalization and the assessment of fit; failure to reject alternatives.

Text: Chapter 1: Introduction to Theory-Based Data Analysis.
Chapter 2: The Logic of Theory-Based Data Analysis.

Reader: Mirowsky, J. (1999). Analyzing associations between mental health and social circumstances. In. C.S. Aneshensel and J.C. Phelan (Eds.) Handbook of the Sociology of Mental Health, Kluwer Academic/Plenum Publishers, N.Y. pp. 105-123.

2 ESTABLISHING CAUSE-AND-EFFECT TYPE RELATIONSHIPS
Univariate statistics as preparation; independent and dependent variables; demonstrating association; eliminating coincident associations; accounting for redundant associations; determining directionality; explicating a causal nexus; internal validity.

Text: Chapter 3: Associations and Relationships
Chapter 4: The Focal Relationship

Resources: Aneshensel, C.S. (2002). The person-variable matrix. Unpublished ms.

Aneshensel, C.S. (2002). Univariate analysis: Central tendency, spread and associations: Unpublished ms.

Aneshensel, C.S. (2002). Measurement: Reliability, validity, and associations. Unpublished ms.

Aneshensel, C.S. (2002). Bivariate analysis: Estimating associations. Unpublished ms.

Assignment 1: Univariate distributions of the focal independent and dependent variables with relevant descriptive statistics. Write-up: Describe the construct and its measurement, the distributions of these two variables, and the substantive meaning of these distributions.

Assignment 2: Estimate the association between the independent and dependent variables using the appropriate bivariate statistical technique. Write-up: Give a theoretical explanation for why you think one construct influences the other (i.e., describe the causal mechanism), and then describe the results of your bivariate analysis. If the independent and dependent variables are not associated with one another, try to figure out why this is the case and what it means for your plan for analysis. You may need a Plan B.

3 RELATIONSHIPS AND SPURIOUSNESS - I
Third variables; correlation, simple regression, multiple linear regression; the interpretation of gross and net associations.

Text: Chapter 5: Ruling Out Alternative Explanations: Spuriousness and Control Variables

Resources: Aneshensel (2002). Multiple linear regression, unpublished ms.

Assignment 3: Estimate the focal relationship net of one control variable (a 3-variable model) using the appropriate multivariate technique; this analysis should include all of the relevant bivariate associations. (Hint: all three variables must be associated with one another to cause spuriousness.) Write-up: Explain why these variables are thought to produce spuriousness, and then describe the results of your analysis, including: the impact of the control variable on the magnitude of the focal relationship; and an explanation of why the control variables did or did not generate spuriousness.

4 RELATIONSHIPS AND SPURIOUSNESS - II
Other methods of estimating relationship; analysis of contingency tables; logistic regression.

Reader: DeMaris, A. (1995). A tutorial in logistic regression. Journal of Marriage and the Family. 57: 956-968.

Aneshensel, C.S. (1983). An application of log-linear models: The stress-buffering function of alcohol use. Journal of Drug Education. 13:287-301.

Resources: Aneshensel, C.S. (2002). Logistic Regression: A Primer. Unpublished ms.

Assignment 4: Estimate the focal relationship net of each control variable in your model (multiple 3-variable models); this analysis should include all of the relevant bivariate associations. Then estimate the focal relationship net of all control variables (i.e., those variables that produce spuriousness in the 3-variable models). Write-up: Explain why you think each these variables should generate spuriousness, and then describe the results of this analysis, including the points in Assignment 3 and an explanation of why some of the control variables generate spuriousness in the 3-variable model, but not the multivariate model (if this is the case for your results).

5 RELATIONSHIPS AND OTHER THEORIES - I
Alternative theories; operationalizing alternative theories; redundancy; "other" independent variables.

Text: Chapter 6: Ruling Out Alternative Theoretical Explanations: Additional Independent Variables

Assignment 5: Estimate the focal relationship net of each additional independent variable in your model (3-variable models). Write-up: Explain why you think these variables operationalize alternative theories and describe the results of this analysis.

6 RELATIONSHIPS AND OTHER THEORIES - II
Multiple "other" independent variables; independent variables and control variables.

Assignment 6: Combine all of "other independent variables" into one multivariate model, and then integrate the analysis of the control and other independent variables into one multivariate model. Examine an all-inclusive model, and then try to simplify it. Write-up: Report the results of this analysis and explain what it means.

7 THE INCLUSIONARY STRATEGY: SPECIFYING PATHWAYS OF INFLUENCE - I
Mediating variables; direct, indirect, and total effects; interpretation of intervening variables.

Text: Chapter 7. Elaborating an Explanation: Antecedent, Intervening, and Consequent Variables.

Assignment 7: Write-up: Identify antecedent, intervening, and consequent constructs for your theory of the focal relationship. These constructs need not be present as variables in your data set. Indeed, it is unusual to find all of these variables in one data set unless you have this model in mind before you collect the data. Explain why you think these constructs ought to play these roles.

8 THE INCLUSIONARY STRATEGY: SPECIFYING PATHWAYS OF INFLUENCE - II
Establishing internal validity with antecedent, intervening, and consequent variables.

Reader: Wheaton, B. (1985). Models for the stress-buffering functions of coping resources. Journal of Health and Social Behavior. 26: 352-364.

Assignment 8: To the extent that these types of variables are present in your data set, add antecedent, intervening, and consequent variables to the model containing control and "other" independent variables. Write-up: Present the results of this analysis and explain their meaning.

9 CONDITIONAL RELATIONSHIPS
Moderating variables: estimation and interpretation of contingencies; effect-modification and subgroup variation.

Text: Chapter 8. Specifying Conditions of Influence: Effect Modification and Subgroup Variation

Reader: Cleary, P.D. and Kessler, R.C. (1982). The estimation and interpretation of modifier effects. Journal of Health and Social Behavior. 23: 159-169.

Finney, J.W., Mitchell, R.E., Cronkite, R.C., and Moos, R.H. (1984). Methodological issues in estimating main and interactive effects: Examples from coping/social support and stress field. Journal of Health and Social Behavior. 24: 85-98.

Baron, R.M. and Kenny, D.A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personal and Social Psychology. 51, 1173-1182.
Continued

Assignment 9: Identify constructs that you think should modify the focal relationship. To the extent that these constructs are present in your data set, estimate conditional relationships using both subgroup analysis and interaction terms. Write-up: Explain why you think effect moderation should occur and present your results.

10 THE ELABORATION MODEL
The integration of inclusive and exclusionary strategies; interpreting associations as relationships revisited.

Text: Chapter 9. A Synthesis and Comment

Assignment 10: Estimate a final model for your data, combining control variables, "other" independent variables; specifying antecedent, intervening, and consequent variables; and including effect-modifiers - to the extent that these variables are present in your data. Estimate both a comprehensive model and a parsimonious model. Write-up: Describe how you developed this model and present the results.