There were a few weaknesses in my study. One was that I only received 36 surveys back out of the 63 that I sent out. This caused a rather large response bias. I also did not account for certain confounding variables, such as the amount of time each person spends driving, how much driving experience they have, if they have their own car, and also whether or not the accident was their fault. I also received some response bias because not every student that I surveyed actually had their license.
I would feel comfortable extrapolating my results to the population of any high school because our school has a large diverse population, and my sample was randomly chosen so it would be a good representation.
To further research my topic you could first conduct this study with a larger sample. Some other ideas are that you could extend this survey to adults also, and even compare genders from different high schools. Another idea would be to be more specific with the time the person has had their license to help reduce some of the confounding variables.