Explanatory Modeling
- theory → hypotheses → statistical test → effect size & result
- Counterfactual
Conditions
- X: independent
- Y: dependent
- Y follows X temporally
- Y changes as X changes and the relationship is significant
- No other cause should eliminate the relation between X and Y
- changes in X produce changes in Y
- ceteris paribus (holding all else constant)
Selection Bias
- Errors and Biases
- groups needs to be random and representative
- one normally cannot randomly assign traits, only memberships to groups
- e.g. one cannot assign a different nationality, but we can create randomly mixed groups
- how to circumvent:
- randomly select from population you want to study
- randomly assign to different groups