From Kelly Blue Books, www.KBB.com, I would be able to research vehicles and all the statistical data about them including MSRP, Miles per gallon, horsepower, weight, etc. Using this data I could hypothesis test how much horsepower would expect to have a certain amount of miles to the gallon, using the weight and horsepower, I would be able to hypothesize how well certain cars would do on some performance tests such as 0-60, quarter mile, skid pad, slalom, etc.
If I were to conduct my own statistical research, I would randomly survey about 100-200 people asking questions about the cars they own including year, company, car, trim level, the miles per gallon they would expect their car to get, any aftermarket modifications, any tickets they had received, whether or not they use radar detectors, and obviously demographic questions. From this data, I would be able to find what types of vehicles (SUV, Sedan, Coupe, etc.) consumers in the specified survey area invest the majority of their money into, as well as how much they invest in keeping their vehicles "up to date" with modifications.