This is a great question. There will be a lot of inference made on my part in answering this, but I will provide my view and add references to the views of others where applicable.
1. Does age, gender, or profession affect a user’s choice of browser?
In short, yes. Let us first tackle age. In my opinion, the majority of users who are 50+ are more likely to use a Windows based machine, and with that, use Internet Explorer (the default browser). Younger users, who have grown up with computers, are more likely to experiment with software and find a browser that meets their personal needs and values.
I am going to tackle profession next. If you work in technology or media, you are much more likely to be aware that their is a choice of browsers, and as you are familiar with the industry and comfortable with technology, you will choose a browser which best meets your needs and values.
A large percent of browsers users do not choose their browser. One play Microsoft made in IE4/IE5 was to invest heavily in IT deployment of IE (See IEAK). They made it very easy for large companies to rollout IE across 1,000 of desktops, including libraries, university’s and other places that control many desktops. This is likely an anchor of Microsoft’s market share as I don’t think anyone else does as much for them. This user group is notoriously slow to upgrade or change. It is hardest for Chrome, Firefox or Opera to penetrate this usage base. Big shops bet more than just the browser on IE: they bet lots of web apps and internal tools. To switch requires rebuilding that infrastructure, making the choice much larger than just browser installs. The browser is free but there’s much resting on the decision that isn’t.
Gender. I think it is safe to infer that their are more men in technology jobs than there are women. If you are in a technology role, as mentioned in profession, you are more likely to make a choice about which browser you wish to use. Therefore, I believe more men than women will make a browser choice.
Over half of the professionals in the United States are women. They account for nearly 50% more college degrees than men. With this kind of knowledge, how is it right that less than 20% of the working computer hardware engineers are men? Answer: it isn’t.
2. Are users of a certain browser more tech saavy and more easily muddle through
usability problems than users of the remaining browsers?
We covered “Are users of a certain browser more tech savvy” in profession during the first question. If you are more tech savvy you will be aware of browsers non-tech heads aren’t. Any browser that is not a default operating system install is more likely to be used by someone who is tech-savvy (unless it has been installed for them by someone who is technologically more advanced — but it could be inferred they will return to their ‘known’ browser choice).
The second part of your question “… more easily muddle through usability problems than users of the remaining browsers?”, it could be argued that tech savvy folk could be more forgiving of interface issues, and will be more likely to experiment and correct issues, gradually aligning their mental model of how a system works with that of the designers. However, I feel it is important to note that this is largely browser independent.
3. When conducting usability tests, will forcing the testee to use a browser
that is not their preferred browser corrupt the findings of the test?
The chrome of the browser shouldn’t corrupt the test. There are a number of commonalities between browsers and standards which they should comply to. A test environment should use a browser ‘out-of-the-box’, that is, no extra toolbars, add-ins and panels. You could always put the browser in to full-screen mode to entirely remove the chrome if you have any concerns.
What you should consider is how the interface within the browser has been designed. For instance, if you have a progressively enhanced design which has very different views in, let’s say IE6 from the latest stable version of Chrome, if you have the time and budget, you may wish to consider testing across variants.