Exercise: Basics
- Q: What could be an unstable/time-varying confounder when investigating the effect of victimization on trust? Does it make a difference if a confounder operates for many or only a few individuals?
- Q: Thinking of the trajectory variables may have, what is a weakness of yearly panel data?
- Q: Do you think measurement of outcome Y at t0 could influence values of Y at t1? Examples?
- NOTE
- Fe or FD models (using the corresponding packages) allow us to crunch large amounts of data, i.e., many observations across multiple time points but…
- …sound causal analysis will probably lead us to manually constructing our comparison groups (treatment and control), so…
- …even if you have data over time, it’s probably a good idea to resort to diff-n-diff