This is a link to an interesting study finding no support for the use of curfews to deter or prevent crime.
However, this study is not "solid" as some of it's statistics involve correlations and not regression analyses. As those familiar with statistics should know, correlation does not prove causation.
But the information and findings are still relevent if consumed with some caveats.
More rigorous abstract, although I have not read the entire study:
Another decent looking study..