There is a psychological phenomenon known as risk compensation which other posters have mentioned. The idea is people will attempt to maintain a constant level of risk in their lives. It may be one reason people in SUVs pay no attention to driving because they have been sold the idea that the SUV makes them safer. I have read that people in SUVs end up no safer than in regular sedans because of this (and the fact that people can more easily lose control and run off the road and roll them over). It is also the reason that ABS in cars did not seem to improve safety significantly, people with ABS apprently took more risks driving that people without ABS. The whole issue is whether the PERCEIVED improvement in safety is greater or less than the ACTUAL improvement in safety. If the perceived improvement is greater than the actual improvement, then the device can increase the risk to the user. I have never read any studies indicating whether ATGATT is actually increases or decreases risk. I have read that helmets do decrease risk, but I have not seen anything about armored jackets or other gear.
My personal practice is ATGATT and hope that it doesn't give me a false sense of security.
It isn't the conditions its the decisions
Don't bring a motorcycle to a car fight