The University of North Carolina at Chapel Hill School of Social Work SOWO 918: Applied Regression Analysis and Generalized Linear Models Spring Semester, 2014 Instructor Shenyang Guo, Ph.D., Room 524j, Tate Turner Kuralt Building CB #3550, School of Social Work, Chapel Hill, NC 27599-3550 Phone: (919) 843-2455 Email: sguo@email.unc.edu Class Meeting Times & Office Hours Class meets on Wednesdays 9:00-11:50 AM (Room 102 TTK) Office hours are Tuesdays 8:30 10:30 (Room 524j TTK) Course Description: This course introduces statistical frameworks, analytical tools, and social behavioral applications of OLS regression model, weighted least-square regression, logistic regression models, and generalized linear models. Course Objectives: At the completion of the course, students will be able to: 1. Understand the type and nature of research questions and data that are suitable for regression analysis; 2. Use Stata computing software package to manage and analyze data with the OLS regression model; 3. Understand the Gauss-Markov theorem and the BLUE property of OLS, especially conditions under which BLUE does not hold; 4. Have a solid understanding of the five assumptions embedded in the OLS regression; 5. Know how to conduct statistical tests detecting violations of OLS assumptions (i.e., multicollinearity, heteroskedasticity, influential data and outliers, etc.); 6. Know how to take remedial measures if harmful violations exist (i.e., weighted leastsquares regression, etc.); 7. Understand the type and nature of research questions and data that are suitable for the generalized linear models; 8. Have a solid understanding of basic concepts of categorical data (i.e., odds ratio, relative risk, marginal probability, and conditional probability); 9. Use Stata computing software package to manage and analyze data with the binary, ordered, and multinomial logistic regressions; 10. Know how to interpret results of regression analysis and logistic regression analysis, and communicate findings to general audiences clearly and effectively in writing; 11. Understand limitations of the regression and logistic regression models, and common SOWO 918: Applied Regression Analysis and Generalized Linear Models Spring 2014 Page 1
pitfalls in using these models; 12. Understand the basics of conducting a Monte Carlo study. Pre-requirement: Students are assumed to be familiar with descriptive and inferential statistics. They should have statistical and statistical software background at least equivalent to that provided by SOWO 911. Students without such prerequisites should contact the instructor to determine their eligibility to take the course. Statistical Software Package: This course will use Stata as the main software package. Required Textbook: Kutner, M. H., Nachtsheim, C.J., and Neter, J. (2004). Applied Linear Regression Models, 4 th Edition, McGraw Hills/Irwin. Required Articles for Reading All required journal articles are available on E-journals. Required book chapters will be distributed in class. Recommended Textbooks: Kennedy, P. (2003). A Guide to Econometrics, 5th ed. Cambridge, MA: The MIT Press. Berk, R.A. (2004). Regression Analysis: A Constructive Critique, Thousand Oaks, CA: Sage Publications Assignments Grade Percentage Assignment 1 8% Assignment 2 8% Assignment 3 8% Assignment 4 8% Assignment 5 8% Midterm Exam (take home) 20% Final Exam Part I (In-class open-book exam) 20% Final Exam Part II (take home) 20% Grading System The standard of School of Social Work s interpretation of grades and numerical scores will be used. H = 94-100 P = 80-93 L = 70-79 F = 69 and below Policy on Incomplete and Late Assignments Assignments are to be turned in to the professor by 5pm of the due date noted in the course outline. Extensions may be granted by the professor given advance notice of at least 24 hours. SOWO 918: Applied Regression Analysis and Generalized Linear Models Spring 2014 Page 2
Late assignments (not turned in by 5pm on the due date) will be reduced 10 percent for each day late (including weekend days). A grade of incomplete will only be given under extenuating circumstances and in accordance with University policy. Policy on Class Attendance No official record is taken to track class attendance. However, students are expected to attend all class sessions. Missing one class will result in big loss in learning, and missing two sessions will make the understanding of course materials almost impossible. Policy on Academic Dishonesty Students are expected to follow the UNC Honor Code. Please include the honor code statement along with your signature on all assignments: I have neither given nor received unauthorized aid on this assignment. Please refer to the APA Style Guide, the SSW Manual, and the SSW Writing Guide for information on attribution of quotes, plagiarism and appropriate use of assistance in preparing assignments. If reason exists to believe that academic dishonesty has occurred, a referral will be made to the Office of the Student Attorney General for investigation and further action as required. Policy on Accommodations for Students with Disabilities Students with disabilities which affect their participation in the course may notify the instructor if they wish to have special accommodations in instructional format, examination format, etc., considered. Course Outline (Topics, readings, and assignments): Jan. 8 1. Introduction and Review Course overview Review of important concepts: Association versus causation Univariate, bivariate, and multivariate analyses Statistical properties Causal inferences Summation operator Neter et al: 1.1-1.2 Guo, S. (2008). Quantitative research. An entry of the Encyclopedia of Social Work, 20 th Edition. New York, NY: The Oxford University Press. Jan. 15 No Class SSWR Conference for Social Work Students and Faculty Jan. 22 2. Simple Linear Regression Optimization: minimization versus maximization Least-square estimator SOWO 918: Applied Regression Analysis and Generalized Linear Models Spring 2014 Page 3
Interval estimation and hypothesis testing (t-test) Review of Stata basics Kutner et al: 1.3-1.7, Chapters 2 & 4 Jan. 29 3. Basic Matrix Algebra Why matrix? Matrix operations Example: solve for b=(x X) -1 (X Y) Stata Lab 1: Stata basics and running regression Kutner et al: 5.1-5.9 Assignment 1 out (Due: 2-5-14): In this Assignment, you will solve problems pertaining to simple linear regression and basic matrix algebra. You will run Stata to solve some of the problems. Feb.5 4. Multiple Linear Regression Model specifications Estimation of regression coefficients Inferences concerning Kutner et al: Chapter 6 (the first half) Assignment 2 out (Due: 2-19-14): In this Assignment, you will solve problems pertaining to multiple linear regression models. You will run Stata to solve some of the problems. Feb.12 5. The ANOVA Table and R 2 Review of variance, covariance and correlation Decomposition of total sum of squares, F test R 2 and adjusted R 2 Kutner et al: Chapter 6 (the second half) Guo, S. (1996). Determinants of fertility decline in urban China, in Alice Goldstein and Feng Wang (Eds.): China, Many Facets in Demographic Change, Westview Press. Feb.19 6. Properties of OLS and the Five Assumptions BLUE criterion and properties of OLS Maximum likelihood estimator The five OLS assumptions Stata Lab 2: Running multiple regression Kutner et al: 1.8 Kennedy, P. (2003). A Guide to Econometrics, 5 th Edition, Chapter 3, Cambridge, SOWO 918: Applied Regression Analysis and Generalized Linear Models Spring 2014 Page 4
MA: The MIT Press. Assignment 3 out (Due: 3-5-14): In this Assignment, you will conduct a Monte Carlo study that shows consequences of OLS regression under various conditions of data generation. Running Stata is required. Feb. 26 7. Stata Lab 3: Diagnostic Tests in Regression Analysis Residual analysis Looking for influential data Multicollinearity diagnostics Variance Inflation Factor A Monte Carlo study demonstrating OLS properties Kutner et al: Chapter 3 Guo, S., & Hussey, D.L. (2004). Nonprobability sampling in social work research: Dilemmas, consequences, and strategies. Journal of Social Service Research, 30(3): 1-18. Mar.5 8. Violating OLS Assumptions and Remedial Measures I Specification errors and selection of Predictors Influential data and outliers Multicollinearity Model validation Kutner et al: Chapters 9 and 10 Midterm exam out (Due: 3-26-14): Use data sets provided by the course or data set you choose to run a multiple linear regression model. Write a paper (no more than 14 pages, double spaced) to present findings. The paper should include: (1) research questions and data that are suitable to a regression analysis; (2) methods and specification of the regression model; (3) diagnostics detecting at least two problems pertaining to violation of regression assumptions and discussion of remedial measures; (4) interpretation of findings; and (5) presentation of findings that answers research questions effectively and efficiently. Mar. 12 Happy Spring Break! No Class Mar. 19 9. Violating OLS Assumptions and Remedial Measures II Heteroskedasticity Weighted Least Squares Kutner et al: Chapter 11 Gujarati, D. N. (1995). Basic Econometrics, Third Edition, McGraw-Hill, pp. 365-389. Mar.26 10. Other Topics in Regression Analysis -- I Regression through origin Partial-correlation coefficient Standardized regression coefficient SOWO 918: Applied Regression Analysis and Generalized Linear Models Spring 2014 Page 5
Apr. 2 Functional form, curvilinear relationship and polynomial regression models R 2 increment: Hierarchical regression analysis Kutner et al: Chapter 7 White, M., Biddlecom, A. and Guo, S. (1993) Immigration, naturalization and residential assimilation among Asian Americans in 1980s, Social Forces,72(1): 93-118. Assignment 4 out (Due: 4-2-14): In this Assignment, you are required to solve problems pertaining to interpretation of standardized regression coefficients, diagnostics of heteroskedasticity, performing partial F test, and testing interactions. Some of the problems require running Stata. 11. Other Topics in Regression Analysis II Dummy variables as predictors: Interpretation of regression coefficients Comparison of several regression equations Testing interactions: Mediator versus moderator Interaction, joint effect and moderator ANCOVA Experimental design Concomitant variables Stata Lab 4: Diagnostics of OLS regression Kutner et al: Chapter 8 Apr.9 12. Logistic Regression Analysis -- I Violations of OLS assumption when dependent variable is dichotomous Logistic response function Maximum likelihood estimator Kutner et al: Chapter 13 Apr.16 Assignment 5 out (Due: 4-23-14): In this Assignment, you are required: (a) to perform maximum likelihood estimation using Excel; (b) to solve problems pertaining to interpretation of odds ratio and running binary logistic regression; and (c) do a group exercise on critical review of a study using regression. 13. Logistic Regression Analysis II Odds ratio Relative risk Predicted probability Stata Lab 5: Running logistic regression models Kutner et al: Chapter 14 SOWO 918: Applied Regression Analysis and Generalized Linear Models Spring 2014 Page 6
Agresti, A. (1990). Categorical Data Analysis, pp. 13-19. Hoboken, NJ: John Wiley & Sons, Inc. Final exam part II out (Due: 5-2-14): Use data sets provided by the course or data set you choose to run a binary logistic regression model. Write a paper (no more than 14 pages, double spaced) to present findings. The paper should include: (1) research questions and data that are suitable to a logistic regression; (2) methods and specification of the logistic regression model; (3) diagnostics detecting potential problems in your data that might violate the logistic regression assumptions; (4) tests of interactions; (5) interpretation of findings; and (6) presentation of findings that answers research questions effectively and efficiently. Apr. 23 14. Group Presentation on Critical Review of Regression Application Logistic Regression Analysis III Multinomial logistic regression Ordinal logistic regression Overview of the generalized linear models Common pitfalls in conducting and reporting regression analysis Why Berk claims that the practice of regression analysis and its extension is a disaster? What to do? Where to go after this course? Agresti, A. (1990). Categorical Data Analysis, pp. 313-322. Hoboken, NJ: John Wiley & Sons, Inc. Schwartz, I., Guo, S., & Kerbs, J. (1993). "The impact of demographic variables on public opinion regarding juvenile justice: Implications for public policy". Crime & Delinquency 39(1): 5-28. Goldstein, A., Goldstein, S., & Guo, S. (1991). "Temporary migrants in Shanghai household, 1984". Demography 28(2): 275-291. Barth, R.P., Greeson, J.K, Guo, S., Green, R.L., Hurley, S., & Sisson, J. (2007). Changes in family functioning and child behavior following intensive inhome therapy, Children and Youth Services Review, 29(8): 988-1009. Berk, A.R. (2005). Regression Analysis: A Constructive Critique, Chapters 1, 5, 11, Thousand Oaks, CA: Sage Publications. Apr. 30 May 2 Final exam part I (in-class open-book exam) 9:00-11:00 (Students with English as 2nd language will have 30 minutes more) Final exam part II due SOWO 918: Applied Regression Analysis and Generalized Linear Models Spring 2014 Page 7