There are big differences in generic data collecting and personal data collection. Most companies that compile these databases of information are doing so anonymously - meaning that you visiting Tom's Hardware right now isn't tied specifically to you as the user (specifically your real name, address, phone number, email address, etc....) - but to a number (i.e. user 102068321). The information collected is to determine how these individuals are utilizing the computer - do they play games, surf the web, use microsoft office, etc.....
The end result is they optimize the operating system to the majority of the uses of the computer (i.e. if everyone spent 8-10 hours a day on Tom's Hardware, they might optimize connectivity to Tom's Hardware).