- Underlying any statistical model is a joint distribution (remember!)
- Panel data: We get an additional dimension that benefits in drawing causal inferences → Time
- Figure 9.2 visualizes the joint distribution of our treatment (threat), outcome (trust) and time.
- Research question: What is the causal effect of victimization (threat) on trust?
- Q: What would be a cross-section in the data in Figure 9.2?
- Q: Why is the distinction between unit und observation essential here? Where would we find several observations of the same individual in this graph (if any)?
- Q: How can we built our treatment and control group?