question about comparing rates before and after an intervention


I am planning an analysis of child abuse rates in a state before and after a law was passed to increase the number of people obligated to report child abuse. My hypothesis is that the number of reports per 10,000 children will increase significantly after the policy is implemented, but the number of substantiated reports in which the suspected child abuse was determined to have actually occurred per 10,000 children will not increase significantly after the policy is implemented. I will have two numbers per year (number of reports per 10,000 kids and number of substantiated reports per 10,000 kids). Statistically, how do I get at whether or not the changes observed for each of these things differ significantly from year to year.

For example, if I find that in the year prior to the intervention, the number of reports is 2 per 10,000, and after intervention it is 3 per 10,000, how do I know if that's statistically significant and not due to chance?

Thanks in advance for your help!


Omega Contributor
The most basic thing you can do is construct a contingency table and calculate relative risks with 95% confidence intervals. Better and more complex methods like an interrupted time series with a negative control group (say rates from another state) would provide more conclusive results.