POLS 208 Lecture Notes - Lecture 11: Causal Inference, Random Assignment, Control Variable
5 Dec 2018
School
Department
Course
Professor
Document Summary
They had a large sample and a strategy: excluded people who never vote and people who always vote. Substantive effect is how large the effect is which is different from external validity. An observational study is one in which th research does not attempt to manipulate (randomly or otherwise) Non-random assignment to treatment (or level of causal variable) Omitted variables bias - when you omit a variable that affects both the indendepent variable the dependent and the bias is the effect of the independent variable on the dependent variable. Other names: common cause, confounder, lurking variable. Control variables as a solution to non-random assignment. Control variable - used to hold constant potential sources of bias in observational studies (a confounder) It adjusts your sample to allow for apples to apples comparisons. Makes it more random and less biased. However, you are not randomly assigned to be here. People who are in college are different from those who are not.
