Hi - I love this website: http://bicyclesafe.com/ , which quotes lots of evidence, particularly under its "Articles" link. Interestingly, the author states "I would like to subtract out fatalities where the cyclist was riding at night without lights, or riding on the sidewalk, or ran a traffic signal, but I can't find the data.", so even someone dedicated hasn't managed to dig this up!
At one point it does state that "Bruce Mackey (formerly of Florida, now head of bike safety in Nevada) says that 60% of bike collisions in Florida are caused by cyclists riding at night without lights." That seems to me very unreliable, particularly given that there's no source quoted, and it heavily contradicts studies of police reports assigning reponsibility for accidents, such as this coverage of a DfT report, which states that dark clothing at night was mentioned in 2.5% of police reports, and a lack of lights mentioned in 2%. http://www.guardian.co.uk/lifeandstyle/2009/dec/15/cycling-bike-accidents-study
This university study reports a 77% reduction in the likelihood of a crash for cyclists wearing fluorescent clothes. http://www.everybody.co.nz/page-2fb0bd78-d3cd-415d-ad8e-acfc71684746.aspx
While we're on evidence, I'd just like to point out these sites which, whilst not about lights or bright clothes at night, are very good at discussing evidence on safety topics which come up again and again:
Review of research on cycle paths:http://www.cyclecraft.co.uk/digest/research.html
Against mandatory cycle helmet laws (but doesn't address voluntary usage so well):http://www.cycle-helmets.com/
For cycle helmets (but doesn't address the problem about them discouraging cyclists well):http://www.bhsi.org/stats.htm