IEG, publishers of www.sponsorship.com, are webcasting a seminar on measuring sponsorships on Thursday, April 19. I've seen bits and pieces of this material before. Some interesting stuff, if memory serves.
If you should partake, and if it's included, keep a keen eye out for 'Signal Detection Analysis' borrowed from the field of psychology. Essentially the technique asks the audience (or attendees) to explicitly, and often in an unaided manner, identify both sponsors and non-sponsors in cases (think logo soup situations like NASCAR) where there are many. Can they correctly identify sponsors and non-sponsors at least 50% of the time? Who's a clear winner? Who are the top performers in a cluttered field? Who's lost in the clutter? Who are the poor performers? All this is plotted on a matrix.
Be somewhat wary, however, of methodologies such as automated logo (on screen for TV) image identification technology as a sole means of sponsorship measurement as, in my view, it's the sponsorship equivalent of advertising equivalency.
The Institute for PR also provides some thinking on measuring sponsorships.
In all cases, not rocket science and no single, catch-all (nor should there be) silver bullet solution. Rather, it involves applying (and triangulating, really) multiple long pre-existing research methods; some qualitative, some quantitative.