I have sworn by not using filters on my lenses and have made the effort to ensure that I use a decent hood.
Recently I've been using some longer, older, non-ED lenses and was curious about what was happening with Purple Fringe.
Since I use these lenses with my telescope setup, I was annoyed by PF as it makes the star images bloat and become less distinct. When I started using a light pollution filter, these star images became much sharper with much reduced PF. This got me to thinking and looking at the frequency curves on the Light Pollution filters vs the UV filters that I had been using in the past.
I did some controlled testing with my non-multicoated Hoya UV filter on a 200mm F2.5 lens. Shooting wide open at branches silohuetted against a hazy sky as well as shooting the sun reflection on chrome, I found that the PF was significantly reduced, even at wide open. There was less difference noticed at lower f numbers, starting at around F5.6 and F8.
I've since picked up a couple of MC Hoya UV filters and have been enjoying the new-found capabilities of my older lenses.
Given that these filters were purchased just for PF control, I don't forsee using them on my shorter primes.
The reason why I selected the Hoya filters is because of the specific curves shown in this test published on the web:
UV filters test - Introduction - Lenstip.com
I hope that these responses help.