POLS 543

Advanced Political Data Analysis

 

Spring 2005                                                                              Tuesdays 3:00-5:40

DuSable 252 and SOCQRL (DU 222)

 

Professor Paul Culhane                                                           Professor Charles Cappell

828 N. Forrest Avenue                                                            Social Science Research Institute Bldg.

Arlington Heights, Illinois 60004                                           148 N. 3rd St.

(847) 392‑1233 [until 10 PM]                                                  DeKalb, Illinois

pculhane828@cs.com                                                              (815) 753-1123           

IASBO 204, 753-0311, M Tu 1:00-2:30                                 ccappell@niu.edu

 

 

The advanced quantitative analysis course in political science satisfies one course requirement for the high proficiency research tool requirement.  The course is modularized, so that students may retake POLS 543 as the module topics change.  This semester, the course consists of two modules.  First, Professor Cappell will cover maximum likelihood estimation methods for binary and other categorical dependent variables, that is, Logit, Probit, and their extensions.  In the second module, Professor will cover ARIMA methods for time-series analysis.

 

Readings

 

The required texts are:

 

J. Scott Long, Regression Models for Categorical and Limited Dependant Variables, Sage, 1997;

Tim Futing Lao, Interpreting Probability Models: Logit, Probit, and Other Generalized Linear

 Models, Sage, QASS 101, 1994; 

 

Richard McCleary & R. Hay, Applied Time Series Analysis for the Social Sciences, Sage, 1980;

David McDowell, R. McCleary, E. Meidinger & R. Hay, Interrupted Time Series Analysis,

Sage, QASS 21, 1980;

SPSS Trends manual (for your version of SPSS).

 

The Long book and the Lao and McDowell et al. QASS monographs are available at VCB and the university bookstore.  The McCleary & Hay text is out of print; it is on one-day Reserve at the university library, and copies are available on the internet.

 

It is cost-effective for doctoral students to purchase major software packages while they have graduate student status.  We will arrange for special orders of SPSS GradPack, for those who do not yet own this software and wish to obtain it this semester.  In particular, for both those who already have GradPack and those who obtain it this semester, we can special-order a licence to activate SPSS Trends, which will be used for the ARIMA course module.  In addition, we will also assist special orders for students who wish to obtain the state-of-the-art, in-print, though expensive Enders (2003) time-series book.


Course Requirements

 

Both modules will focus on practical applications of the analysis techniques, that is, construction of analysis models and interpretation of results.  Each module will include specific practice exercises, described on assignments passed out at the end of specific segments of each module, and a capstone project.  For the capstone project, students are encouraged to use problems from their own research, e.g. a preliminary (or actual) dissertation project in the case of doctoral students; if students’s current research is not applicable (i.e, no categorical dependent variables or no time series data) the professors will assist you in identifying an appropriate test data set.

 

       Material Covered                                                 Date                       Worth

Categorical: Weekly Exercises                                    Weekly                        30%

Categorical: Capstone Paper                                       March 22                     20%

Time Series: Weekly Exercises                                  Weekly                        30%

Time Series: Capstone Paper                                      May 10                        20%

 

Attendance and participation are required. The course will be fast-paced and absences from the lecture or lab sessions will break the flow of learning. Attendance and participation may count for ±5%; in particular, absences may incur up to a 1% penalty.  Auditing students may forgo one practice assignment in each module.

 

This is a SOCQRL fee-assessed course, and both modules will make extensive use of the Sociology Quantitative Research Lab (SOCQRL): software, data archives, workstations with reserved class time each Tuesday. Weekly or semi-weekly assignments will require the use of the lab in addition to that reserved during class time to complete the assignments which require lab resources.

 

Students should be familiar with the University's and Department's policies regarding withdrawal from courses and academic honesty.  While cooperation among students, especially in a lab setting, is traditional and usually useful, all work submitted for a grade (both practice assignments and module projects) must represent the individual student's work.  The Department allows "incomplete" grades only in extraordinary circumstances.

 

Course Schedule

 

January 18.      Course plan; introduction to SOCQRL; begin regression review.

 

January 23.      Review of regression foundations.

Read: Long Ch. 1-2;  review covariance and regression in your intermediate text.

 

February 1.      Nonlinear Models: Categorical Distributions and Maximum Likelihood Estimation; Introduction to Logistic & Probit Models. Notation and Functions.

Read:  Liao Ch.1-2, pp. 1-10; Long Ch. 2-3, pp. 25-54; Cappell, “Notation; Mathematical Operators; Logistic Regression” handout.

 


February 8.      Estimation and Interpretation of Logistic Models.

Read: Long Ch. 3, pp. 59-83; Liao Ch.3, pp. 10-25; Cappell, “Logistic Regression” handout.  Recommended:  Long Ch. 3, pp. 54-58.

 

February 15.    Hypothesis Testing and Test Statistics.

Read:  Long Ch. 4, pp. 85-113; Cappell, “Logistic Regression” handout;

Long Ch. 3 pp. 59-83.

 

February 22.    Ordinal Dependent Variables – The Proportional Odds Model.

Read:  Long Ch. 5, pp. 114-147; Liao Ch. 5, pp. 37-47; Cappell, “Proportional Odds Model” handout.

 

March 1.          Multinomial Logit Models.

Read:  Long Ch. 6: pp. 148-186; Liao Ch. 6 pp. 48-59; Cappell, “Multinomial Logit” handout.

 

March 8.          Recapitulation and Preview of Advanced Extensions: Loglinear Analysis, Poisson, Latent Class, Generalized Linear Models, and GLLAMM. Capstone Topics.

Recommended:  Long Ch. 7-9; Cappell, “Loglinear” handouts.

 

March 15.        Spring break.

 

March 22.        Time series introduction, the OLS model, and the limitations of D.

Reference:  Charles Ostrom, Time Series Analysis, Sage, QASS #9, 1978.

 

March 29.        ARIMA math foundations; univariate estimation.

Read:   McCleary & Hay ch. 1, 2.1-2.8; McDowell et al. §1-2.7.

 

April 5.            Debriefing of first module capstone projects; ARIMA lab.

 

April 12.          Seasonality.

Read:   McCleary & Hay ch. 2.9-2.13; McDowell et al. §2.8.

 

April 19.          Intervention estimation.

Read:   McCleary & Hay ch. 3-4; McDowell et al. §3.

Recommended: McCleary & Hay ch. 6.

 

April 26.           Multivariate ARIMA.   

Read:   McCleary & Hay ch. 5.

 

May 3.             Complete time series module.

References: Walter Enders, Applied Econometric Time Series, Wiley, 2nd., 2004.

Lois Sayrs, Pooled Time Series Analysis, Sage, QASS #70, 1989.

 


May 10.           Final exam period:  Debrief second module projects.