Political Science 205: Measurement and Data Analysis in Political Research

 

Prof. Eric Smith Office: 3711 Ellison

E-mail: smith@sscf.ucsb.edu Hours: MWF 11:00 - 12:00

Phone: 893-4328

TA: Bob Gno email: bobngo@umail.ucsb.edu

This is an introductory course in quantitative political research. In this course, we will examine basic quantitative methods and the principles underlying them. This course has three purposes: (1) to give you experience analyzing social science data and writing research papers about testable substantive theories; (2) to teach you the statistical theories and assumptions that underlie the methods, and how those theories and assumptions relate to substantive problems; (3) to teach you how to read the statistical literature so that you can go beyond the methods presented in this class.

No mathematical or statistical background beyond high school algebra is required, nor is experience with statistical applications on computers. Both algebra and use of computers will be covered in this course. We will be using SPSS, a statistical package available in the Humanities and Social Science Computing Facility. You may use another statistical program such as SAS if you wish.

The requirements are three short papers, several short homework assignments (to be announced) and a final exam. Each of these requirements will count for 20 percent of the grade (the homework collectively will account for 20 percent). The papers must be typed and double spaced. Each paper has a limit of five pages of text plus tables. The papers will be due one week after the material is covered in class. Because it is far easier to learn statistics by working through problems than by merely reading or listening, ALL ASSIGNMENTS MUST BE DONE ON TIME.

The books are available at the UCSB bookstore. I have ordered the 4th edition (© 1999) of the Keller et al. text because it is $50-60 cheaper than the current 6th edition, and there have been no developments in basic statistics in the last five years. Feel free to purchase a newer edition if you like (the authors and publisher will certainly appreciate your kind donations). I will be happy to review the book and let you know if there are any changes in the reading assignments.

Data sets for assignments include:

  • Attitudes toward nuclear power plants in California

    • Gerald Keller, et al., Statistics for Management and Economics, abbrieviated fourth edition.

    Earl Babbie, et al., Adventures in Social Research

    Data sets for assignments include:

     

    Course Outline and Assignments

    I. Univariate Analysis (2 weeks)

    Review of theory construction. Statistics, mean, proportion, variance, standard deviation, probability, sampling and sampling distributions, the central limit theorem, standard error of an estimate, hypothesis testing, Type I and Type II errors.

    Readings: Keller et al., chapters 1-2, ch. 4 (pp. 104-140 only), ch. 5, ch. 6 (pp. 205-214 only), chap. 7 (pp. 257-276 only), ch. 8-11

    Babbie and Halley, chapters 1-10

    Note: This seems like a lot of reading, but the Keller et al. book is padded with large figures, exercises, and Excel and Minitab instructions (which you should ignore). You will be able to skim through most of the material quickly. Focus on chapters 8-11.
    1. Bivariate Analysis and Causal Theories (2 weeks)

    Estimating two variable systems and theories; contingency tables, differences in means, hypothesis testing, measures of association.

    Readings: Keller et al., chapter 4 (pp. 140-147) and chapters 12 and 13

    Herman Loether and Donald McTavish, Descriptive and Inferential Statistics, 4/e, ch. 7

    Babbie and Halley, chapters 11-16Assignment 1: Write up a bivariate causal analysis. For each variable, report the frequency distribution and the appropritate univariate statistics ("appropriate" means for the given level of analysis). Report appropriate measures of association and other relevant statistics to test your theory. Additional details will be discussed in class.

    III. Multivariate Models and Tabular Analysis (2 weeks)

    Estimating multi-variable models; cross-tabulation. Interaction, intervening variables, spurious relationships, suppressed relationships and the limits of statistical explanation. Measures of association and partial correlation.

    Readings: Keller et al., chapter 15

    Babbie and Halley, chapters 17-21

    Assignment 2: Examine and write up a three-variable causal system. Comment on direct and indirect effects and any statistical interactions. Discuss all appropriate statistics.IV. Ordinary Least Squares (OLS) Regression (4 weeks)

    The ordinary least squares model (OLS), the Gauss-Markov assumptions and theorem. Unstandardized vs. standardized regression coefficients; measures of goodness of fit; dummy variables, functional form, interaction terms, analysis of residuals, autocorrelation, heteroskedasticity.

    Readings: Keller et al., chapters 16, 17, and 19

    Handouts

    Assignent 3: Using multiple regression, examine and write up a three (or more) variable causal model. Discuss the coefficients, standard errors and important summary statistics. Examine the residuals for violations of assumptions and discuss what you find.