1% If the outcome of interest is trending over time, before/after comparisons will provide a biased estimate of the law Question:

1992: 76.1% If the outcome of interest is trending over time, before/after comparisons will provide a biased estimate of the law Question: how much of the increase was the law and how much was it secular trends? Intervention occurs at time period t1 Ya If using time series, estimate of the effectiveness of the law is Yt1 Solution? Pool cross-sectional and time series data Use time series of "untreated" group to establish "trends" What would have occurred in the treatment states in the absence of the intervention?