The Humble Bundle's statistics [1] have offered an argument against that perception for a while. On average a Linux user pays $11.02 for a Humble Bundle, while Windows users pay $5.93 and Mac users pay $8.43. Linux and Mac users are also overrepresented among the Humble Bundle's buyers (having made 8% and 11% of the total number of purchases respectively) compared to the gaming community at large.
That said, I'm not sure how widespread this myth really is outside of select Internet communities. Enterprise Linux users are known to pay for their software (even if somewhat indirectly, e.g., through support contracts), and the biggest myth concerning desktop Linux users seems to be "huh, are there any?".
I think most Linux users see HIB as a form of activism or donation (specifically to show that games on Linux can be profitable) rather than as a purchase. Heck, I bet most people who get HIB games (on Linux or otherwise) never even play them.
I only play one or two games from most of the bundles, though every bundle I've gotten has had one or two that made it worth it. And like most other Linux users I've usually paid around $10. But I haven't bought any bundle that didn't have at least a game or two that I knew I'd actually play a bit.
That doesn't really tell us anything because Windows users have much more choice as to where their gaming money goes compared to Linux users and Mac users. Same goes with the overrepresentation.
It does contradict the claim that "Linux users will not pay for software" directly.
You also have to note that while Windows users are more likely to own the games prior to any given Humble Bundle sale (esp. given that many games were first ported to Linux specifically for a Bundle) at the time of the sale they are unlikely to find a better deal on any of the games either separately or as a package, even games unlocked for beating the average payment. That would make buying those games from the Bundle a rational choice if you are a Windows user and want even one of them.
Of course, the above argument doesn't negates the fact that Windows gamers have more choice. For them the opportunity cost of paying more for a HIB could be not buying another game on Steam, GOG or elsewhere.
I guess in the end it all hinges on how much the buyers (all of them: Windows, Linux and Mac users) are interested in playing those games specifically vs. playing any games at all. If they are interested specifically in the titles sold, then yes, the statistics would mean that Linux users are more generous. Otherwise it's probably more about opportunity cost.
I don't think that was the OP's point. The conclusion that is usually drawn from the myth that Linux users aren't willing to pay for software, is that there isn't a viable market there. There clearly is, however, as the HIB has demonstrated.
Even if you ignore the overrepresentation of Linux users, the fact stands that when you look at total revenues, Linux rivals OS X.
HIB definitely demonstrates that Linux users are willing to pay for software (which is awesome!), but I'm not sure that there's enough data to use that to conclude there's a viable market there.
We know that the HIB has drawn in significant revenue from Linux customers, but that doesn't give us the faintest clue how profitable they are (remember, that average $11 Linux sale has to be shared with multiple developers and charities). You might take the fact that they keep offering more HIBs as a sign that Humble Bundle (the company) considers its humble bundles (the promotions) to be successes, but it's not a guarantee that the metric they're using to define success is profitability. They've raised nearly $4.5 million in funding, so they're not strapped for cash; it's very possible they're more focused on using the bundles as a marketing channel than as an actual revenue stream, convincing consumers to trust the Humble Bundle brand name and developers/publishers that they're a meaningful distribution platform.
tl;dr: the fact that a successful venture-funded startup is selling software to Linux users, when we don't know if they're profitable or what their long-term strategy is, doesn't necessarily imply that there's a viable market for selling software to Linux users.
That said, I'm not sure how widespread this myth really is outside of select Internet communities. Enterprise Linux users are known to pay for their software (even if somewhat indirectly, e.g., through support contracts), and the biggest myth concerning desktop Linux users seems to be "huh, are there any?".
[1] See http://cheesetalks.twolofbees.com/humble/ for a great report.