Which method do you think determines someone's behavior better, watching behavior unfold in the real word or handing someone a questionnaire to fill out? In my opinion, naturalistic observation will compile the best results. These two approaches of studying human behavior have one major difference: "watching" versus "asking". The benefit of naturalistic observation is that the subject being studied doesn't know they are being observed so then the results aren't skewed. In this method of studying human behavior, the results seem to be truer as the subject is being one hundred percent themselves instead of possibly changing their behaviors or attitudes if they knew they were being studied.
Here is a short video that explains naturalistic observation really well: http://www.youtube.com/watch?v=zdXjPxOsfuo
Surveys can be an efficient approach to studying human behavior but there can be some downfalls. During surveys the participants might not put their full effort into their responses, they might not answer them honestly, and they could miss group variables such as race and gender. Overall, the method of naturalistic observation studies subjects in their elements without interference where as surveys can cause a participant to slightly alter their responses. I think these methods of observation are very important because they give scientists multiple ways of observing human behavior in a way that is best for their particular study whether that is naturalistic observation or a survey.
During high school I experienced both of these methods first hand. For one project I had to observe someone in his or her elements. While I was observing this person I had to be very careful not to be obvious that I was watching them so they wouldn't alter their behavior. For another project I had to survey a class for each subject regarding their grades and how much time they spent studying a week for that subject. Most of the results were typical straightforward answers such as a B or 3 hours a week. But then of course you had the students who didn't put answers such as "why do you need to know?" or "I don't know, you expect me to keep track". Answers like these skewed my results, as I had to take them out of my final results. These real life experiences helped me see that naturalistic behavior definitely allows the researcher into the person's behavior and life without skewing the results where as surveys can sometimes alter the outcome.
One thing that I still wonder about is, what kind of studies needs naturalistic observation and which studies would surveys work better for?