Imagine a time series study evaluating the effects of a helmet law on head injuries among hockey players in amateur city leagues across the nation. Head injuries were significantly lower in the year immediately after the law was passed than in the preceding year. Construct four time series graphs, one for each of the following patterns of results.
1. The helmet law worked.
2. The helmet law seemed to work initially, but its effects were short‐lived.
3. The helmet law had no effect; the apparent drop was probably just the result of regression to the mean.
4. The helmet law didn’t really work; the apparent drop seemed to reflect a general trend toward reduced violence in the sport.
In the section on interrupted time series designs, we described several variations on the basic design. How might each of those be used to strengthen the hockey study?