The-costs-of-crime-victimisation-in-Aotearoa.pdf [pdf, 2 MB]
...using only never-treated units or all not-yet-treated units as control groups. We use respondents who do not report any victimisation as the control group in our main analysis, and provide an alternative not-yet-victimised control group in a robustness check. With many cohorts and time periods, this approach results in many treatment effects, so the ATT (g, t) can be aggregated in event-study parameters that provide average treatment effects at different periods after treatment, or an average t...