Consider not only on the vehicle, but the target demographic of the vehicle, as well as performance perceptions of the vehicle.
Ask yourself, is your stereotypical Mustang buyer going to share the same risk adverse behavior as a stereotypical Toyota Camry buyer?
If you asked Mustang and Camry buyers to rank the importance of performance, style, handling as opposed to say gas mileage, operating costs, reliability, would the results differ? Additionally, is it possible there could be an over expectation of the capabilities of a vehicle. A particular person's perception of the Mustang as a performance sports car, could lead to more risk taking. All just my personal, non-professional thoughts for discussion, not scientifically proven.
I agree with t32b (and his wife), there are many variables to consider, and that correlations by themselves rarely tell the full story, but do make for interesting dialogue. I am glad that Kayvan raises such questions. Or it can be constructed relatively safe, but used in an unsafe manner more frequently than the average car. There are so many different factors at play, but if the statistics are accurate, and consistent, then perhaps there is a correlation due to some determinable cause.
And collisions are often the result of multiple contributing factors, occurring close in time.
I previously worked professionally in an nationally recognized not-for-profit organization that was designed to reduce incidents and injuries, by providing the operators of a specific class of motor vehicles with mental and physical strategies to analyze, recognize and instinctively respond or avoid potentially dangerous situations. We were focused on both active and passive safety measures taken by the operator. Not all contributing factors are always truly known in an accident or incident. Do you think all drivers readily disclose "I am so sorry officer, I was stupidly looking for a CD in my glovebox and didn't see the bicycle when I swerved due to my lack of proper attention."
In response to your question about how can the car receive 5 stars for crash ratings, but still be considered unsafe, I am not intimately familiar with the current NHTSA and IIHS studies, so I may be off base here, but consider this:
A car can be rated to relatively safe in an impact (passive safety) but poor in incident avoidance (active safety). Other unknown factors may be leading to an over-representation of a particular vehicle.
Additionally, I see a 4 star, not 5 star rating for the 2006 Ford Mustang for front side impact, on the NHTSA website:
http://www.nhtsa.dot.gov/ncap/cars/3490.html