The Fundamentals of Political Science Research

Citation

Kellstedt, P. M., & Whitten, G. D. (2018). The Fundamentals of Political Science Research (3rd ed.). Cambridge University Press.

Chapter Summary

The third edition has twelve chapters and four short appendices (critical values of chi-squared and t, and the link functions for binomial logit and probit models), followed by a bibliography and an index, in a book that Cambridge lists at 340 pages. Every chapter opens with an overview and closes with a list of concepts introduced and exercises. The book moves in a deliberate order: causal theories first (Chapters 1-4), then measurement and description (Chapters 5-6), then probability and inference (Chapters 7-8), and finally regression (Chapters 9-12). The publisher says it takes an integrated approach to research design and empirical analysis, so that students can develop and test causal theories. The preface to the third edition adds that the new "Your Turn" boxes ask students to apply each chapter's lessons to a new situation.

Chapter 1: The Scientific Study of Politics

The opening chapter asks what political science is and presents the scientific approach as a search for causal explanations. It introduces variables and causal explanations, models of politics, and a set of "rules of the road": focus on causality, do not let data alone drive theories, consider only empirical evidence, leave ideology and normative statements at the door, and pursue both generality and parsimony.

Chapter 2: The Art of Theory Building

How to build a theory. The sections cover good theory-building strategies, interesting research questions and variation, using the literature without getting buried in it, thinking formally about causes (utility, expected utility and the puzzle of turnout), and attention to institutions. The third edition expanded this chapter with a section on three strategies for developing an original theory, including a Z variable that modifies an established X-Y relationship, and a section on building on previous work.

Chapter 3: Evaluating Causal Relationships

The chapter on causality. It starts from everyday causal language, then sets out four hurdles to establishing a causal relationship and what happens when other possible causes are not controlled for. Three examples from political science, on life satisfaction and democratic stability, race and political participation, and the evaluation of Head Start, show why the issue matters.

Chapter 4: Research Design

Comparison as the key to establishing causal relationships. The chapter covers experimental designs and their relation to the four hurdles, the difference between random assignment and random sampling, near-experiments and the drawbacks of experiments, and observational studies in cross-sectional and time-series form. The third edition added a section on dissecting research by other scholars.

Chapters 5-6: Measuring Concepts of Interest; Getting to Know Your Data

In earlier editions these were a single chapter; the third edition splits them. Chapter 5 deals with social science measurement: conceptual clarity, reliability, measurement bias, validity and their relationship, with two measurement controversies, measuring democracy and measuring political tolerance. Chapter 6 introduces descriptive statistics: the measurement metric of a variable (categorical, ordinal, continuous), ways of describing categorical and continuous variables, and the limits of descriptive statistics and graphs.

Chapter 7: Probability and Statistical Inference

Populations and samples, basic probability, the central limit theorem and the normal distribution. The presidential approval example brings in sampling issues, including random samples in the cellphone era and the effect of sample size; the third edition added a section on the challenges of random sampling.

Chapter 8: Bivariate Hypothesis Testing

Bivariate hypothesis tests and causal claims, how to choose among tests, and the logic and limitations of p-values and statistical significance. Three tests are worked through: tabular analysis, difference of means and the correlation coefficient. The preface says the logic of chi-squared tests and tabular analysis was expanded in this edition.

Chapters 9-10: Two-Variable Regression Models; Multiple Regression: the Basics

Chapter 9 introduces OLS regression with two variables: fitting a line, estimating it, measuring uncertainty (root mean-squared error, R-squared, confidence intervals, one- and two-tailed tests) and the assumptions behind OLS. Chapter 10 moves to multiple regression: interpretation, comparing effect sizes, statistical versus substantive significance, the consequences of failing to control for a variable, an example from the trade literature, and a new section on constructing and writing about regression tables.

Chapters 11-12: Multiple Regression Model Specification; Limited Dependent Variables and Time-Series Data

Chapter 11 covers dummy independent variables, interactive hypotheses, outliers and influential cases and multicollinearity. Chapter 12 treats dummy dependent variables, including the linear probability model, binomial logit and probit, and time-series issues such as lags, trends and spurious regression, the differenced and lagged dependent variable, and an example on the economy and presidential popularity.

Key Concepts

Chapters 1-4: Causal theory and design

• Variables and causal explanation: the vocabulary of independent and dependent variables.
• Four causal hurdles: the checklist in Chapter 3 for judging a causal claim.
• Experimental versus observational designs: the main division in Chapter 4.
• Random assignment versus random sampling: a distinction the Chapter 4 headings single out.

Chapters 5-8: Measurement, description and inference

• Validity and reliability: the two main problems in measuring a concept.
• Measurement metric: whether a variable is categorical, ordinal or continuous.
• Central limit theorem: the basis for learning about a population from a sample.
• p-values and statistical significance: the logic and limits of bivariate testing.

Chapters 9-12: Regression

• OLS regression and goodness-of-fit: fitting a line and judging it with R-squared and the root mean-squared error.
• Multiple regression: controlling for other variables to isolate an effect.
• Dummy variables and interactions: ways to include categorical variables and conditional hypotheses.
• Logit, probit and time-series issues: the topics of Chapter 12.

Critical Analysis

Strengths

• One thread from theory to regression. The chapter order follows a single idea: build a causal theory, establish what a good test needs, then learn the statistics that perform the test. Students see why regression is used before they meet it.
• Technical material for non-mathematical readers. The publisher says the book makes technical material accessible to students who might be intimidated by mathematics, and each chapter ends with a list of concepts and exercises.
• Substantial treatment of causality and design. Chapters 3 and 4 spend nearly 50 pages on causal reasoning and research design before any statistics are introduced.
• Practical guidance. The book includes advice on using the literature, on presenting and writing about regression tables, and on dissecting other scholars' research.
• Companion materials. Cambridge lists free and instructor-only resources and separate SPSS, Stata and R companions for the third edition, so the book can be paired with the software a course uses.

Limitations

• Quantitative focus. After the first four chapters the book is about statistics and regression. Qualitative methods and case studies are not given chapters.
• The later chapters are the most technical. Chapters 11 and 12 (interaction terms, influential cases, multicollinearity, logit and probit, time series) assume that the regression logic of Chapters 9-10 has been absorbed, so a short course may reasonably stop at Chapter 10.
• Examples tied to American politics in places. Several worked examples visible in the contents (presidential approval, turnout, presidential popularity, Head Start, race and political participation in the United States) are American, although others such as democracy and trade are comparative.
• No independent review located. The assessment rests on the contents, the preface and the publisher's description.

Where It Fits in the Course

This is the main statistics-using textbook among the research methods books. Shively and Endersby's The Craft of Political Research is the shorter, nontechnical alternative that covers design and measurement and gives only three chapters to statistics; it is a gentler first read, with Kellstedt and Whitten the better choice for students who will run regressions themselves. Halperin and Heath's Political Research: Methods and Practical Skills is a broader methods text that also covers qualitative methods and the philosophy of social science, and it suits a student planning an independent project.

King, Keohane and Verba's Designing Social Inquiry and Gerring's Social Science Methodology argue about how qualitative and quantitative work can share standards of inference; the first is best read after the causal chapters of Kellstedt and Whitten, and the second is a graduate-level account. Shepsle's Analyzing Politics introduces rational choice and formal theory, which links to the "think formally" section of Chapter 2.

How to Use This Book

The book is for undergraduates in a first course on political science research methods, and for students who need to read quantitative political science. The publisher says the book makes technical material accessible to students who might be intimidated by mathematics, and Chapter 9 includes a section on the minimal mathematical requirements of regression.

A practical plan for a semester:

• Weeks 1-4: Chapters 1-4. Read these slowly and do the exercises, especially the four causal hurdles in Chapter 3. They are useful even if you never run a regression.
• Weeks 5-8: Chapters 5-8. Measurement, description and inference. Use the "Your Turn" boxes and the end-of-chapter exercises.
• Weeks 9-12: Chapters 9-12. Regression. Chapters 9-10 are the core. Chapters 11-12 can be assigned selectively, depending on what the course data require.
• Alongside the book. Use the data sets and the SPSS, Stata or R companion that matches your course, and replicate one example from Chapters 8-10. Re-read Chapter 3 after Chapter 10 to check whether your regression results support a causal claim.