POL101Y1 Lecture Notes - Confounding, Random Assignment, Internal Validity
Document Summary
Obstacles to establishing causality: reverse causality and simultaneous relationship, confounding variables spurious, misguided, and hidden relationships. Confounding variable is both related to x and y. Research designs: strategies that researchers can use to empirically test whether x cause y: experiments, observational studies. Great care should be given when you assign a particular value of x (=assign a treatment) to different groups. Otherwise, you might invite confounding factors into the process. Internal validity: the degree of confidence in the causal conclusion drawn from the analysis. External validity: the degree of confidence in generalizability of the causal conclusion drawn from the analysis. Laboratory experiments: relatively high internal validity but relatively law external validity. Using a random sample chosen from the entired population, we can external validity than laboratory or field experiments. In addition, we would expect high internal validity, because of a random assignment of treatments to sampled individuals.