Good point! A nitpick:

Believing that Internet Explorer is safer from malicious websites than Google Chrome won't decrease your computer's infection rate.

I don't think that this is a good counterexample. If people who want to avoid malicious websites tend to believe that IE is safer from them and thus choose IE, then IE should get better results (ceteris paribus) since people who care about malicious websites are also more likely to take care to type paypal.com directly into the address bar, etc.

In other words, one's attitude to safety from malicious websites has as much effect on infection rates as one's attitude to safety from car crashes has effect on crash rates. (Or more, I would guess offhand.)

Am I missing something?