Pick a subject that you believe will provide interesting results (i.e., one you legitimately don't know what the outcome will be for at least a number of the questions) because this will give you an incentive to try your best in collecting data.
For any requirements given, try to do double. This advice isn't in order (primarily) to try and get a better grade, but if you follow my first piece of advice it will allow you to have more interesting/representative findings. The bare minimum can return neat results, but if you put more time into the project you'll feel like you come away with more as well.
While procrastination may work on your 10-15 page paper you'll do overnight, I cannot fathom doing this whole project at the last minute and I wouldn't recommend anyone being brave and trying to figure out if it's possible. Hammer out your survey questions, have a methodology in order to obtain your sample (at all costs, avoid email; however, it may prove to be the only reasonable option so try your best to send out as many emails as you possibly can stand), and begin inputting data right away. Some of the process is tedious, but it's very front-loaded so that once you're about to get fully aggravated, the interesting part (interpreting data and creating hypotheses) kicks in and it becomes an overall enjoyable experience.