PSYCO212 Lecture Notes - Lecture 22: Statistical Conclusion Validity, Nuisance Variable, Regression Testing
Document Summary
Mediating variable: a variable that helps explain the relationship between two other variables. Testing for a mediating variable: run a regression test. Reverse-mediation model: use special statistics, researchers can compare the two models to see which model fits the data the best. The third variable can be detected using multivariate designs. The third variable is external to the two variables in the original bivariate relationship, and the mediator is internal. The third variable is a nuisance variable (not of central interest). The mediator is of direct interest to the researcher. The mediator is internal to the causal variable. Longitudinal studies help establish temporal precedence, and multivariate provide evidence for internal validity. We should interrogate the construct validity, external validity, and the statistical conclusion validity. Construct validity: how well each variable is measured. External validity: how well the results are generalized. Statistical conclusion validity: the effect size and statistical significance. Simple bivariate correlations are not sufficient for establishing causation.