Statistical models have grown increasingly sophisticated, yet comparatively little progress has been made in understanding causal relationships, interpreting the results of research, and predicting the results of policy changes. This course takes a back-to-basics approach that emphasizes collecting the right data up front, so that the statistical model can be relatively simple and the interpretation clear. Topics include sampling, confounding, effect size, causality, and real and natural experiments. Examples focus on quantitative research, although a few qualitative examples will be included to show that the same principles apply. The prerequisite is familiarity with basic statistical methods including differences between means (e.g., t tests), differences between proportions (e.g., chi-square tests), and normal linear regression.
Students will be assessed on weekly data analysis assignments (completed mostly in class) and their choice of a final exam or final project. There will be 1-2 weekly readings which are somewhat technical.