PDC21: What Would it Take to Change Your Inference? Quantifying the Discourse about Causal Inferences in the Social Sciences

Instructor:
Kenneth Frank, Michigan State University

Fee: $105 

Participants will learn the fundamentals of inferences from quantitative analysis in the social sciences, including Rubin's causal model (the counterfactual) and statistical control through regression. Using these fundamentals, they will learn to characterize the robustness of statistical inferences and violations of assumptions from quantitative analyses. They will learn to characterize the robustness of inferences from a regression in terms of correlations associated with an omitted variable (assumed to be zero in making inferences from a regression). Participants will learn not only how to conceptualize the robustness of an inference but also how to calculate the sensitivity of inferences from general linear models using spreadsheets (Excel) or macros in SPSS, SAS, or STATA. Participants will learn how to apply the techniques to concerns about internal and external validity, as well extensions of the techniques to logistic regression and multilevel models. Participants will learn to employ a language for articulating the robustness of inferences that can be applied to their own analyses or to inferences in the literature.