What I do is start out with a test exposure that gets the light tones (such as skin tones) the way I want them. Then I look at the dark tones. If they're not black enough, I use a higher-contrast filter, and if they're too black, I use a lower-contrast filter.
You almost never need to try every one of them, unless you're searching for a special effect. Depending on how you develop your film and how you like your prints to look, you'll probably find that most of your negatives print with a #1-1/2, a #2, or a #2-1/2. Then, if your print looks too flat, you go up a half-number or a number; if it looks too contrasty, you go down a half-number or a number. Lather, rinse, and repeat.
Actually, if you insist on being terribly analytical about it all, you can measure the density range of the negative: measure the density of the lightest important highlight, and of the darkest important shadow (ones that have detail in them, not absolute black or white.) I've got a "spot mask" for my Minolta light meter that lets me use it for this measurement.
Once you know the exact distance between the highlight and shadow in f/stops, you convert it to log density units (0.3 log d units per f/stop, I think...) With that information in hand, you can consult the spec sheet of your paper and see what contrast filter is recommended for that log density range.
Even after I learned all that, though, I found I got better results by just making a test strip with a #2 filter, then looking at it and deciding whether I needed to go up or down!