Many of the questions that we wish to answer in the social sciences address outcomes that are limited and fixed in their answer choices. For example, do Americans agree that Atheists share a common vision of American society? How did the Great Recession affect employment inequalities across racial groups? Who do happy people compare themselves to? Which social class does the child of a blue-collar worker end up in? How frequently do adolescents use marijuana? Questions such as these cannot be appropriately answered using linear regression models, requiring more advanced techniques which will be covered extensively in Soc8811.
This course will focus on applied statistics and primarily deal with regression models in which the dependent variable is categorical: binary, nominal, ordinal, count, etc. As a catalyst for the course, we will consider flexible methods developed for introducing nonlinearities into the linear regression framework. Specific models to be addressed include: logit, probit, generalized ordered logit, multinomial logit, Poisson, negative binomial, zero inflated, fractional response, LOWESS, kernel weighted local polynomial, and mixture models.
Throughout the course, we will address common statistical issues that require special consideration when applied to nonlinear regression models, including: the calculation of predictions, interpretation of coefficients, interaction, and mediation. We will also become familiarized with techniques developed for applied research: model fit, selection, and robustness, joint hypothesis testing, weighting, clustering, and poststratification for complex survey design, and missing data.
Soc8811 covers statistical methods for analyzing social data and is designed for graduate students in the social sciences. Students are assumed to have a background equivalent to Soc5811 and thus have familiarity with linear regression models. The course will be taught in Stata, but students will have the opportunity to instead use R if they prefer.
1. Produce, interpret, and report results from complex statistical models
2. Understand how to apply data analysis to substantive research questions, and effectively present results to a general interest academic audience
3. Develop strategies and competency to conduct future studies of advanced techniques in quantitative methods
4. Build a robust, reproducible workflow to move from raw data to numerical and visual information placed in a final paper.