Damn it - I hate these subjects. To answer them honestly brands you a troll, but to leave it alone perpetuates the “Blender is better than everything else” meme that stunts development.
I was going to avoid this (and have come back after initially typing than discarding a post); but assuming there was an honest desire to know why Maya is preferred over Blender - I’ll answer briefly and honestly.
1. Stable & Flexible User Interface
Maya has had alot more artists use their application over alot longer a time than Blender has been actively used in a professional sense (outside the original NaN context). The amount of time spent getting it right, and the profit motive to please as many users as possible with a design they could all live with, resulted in a user interface that stabilised (in terms of radical changes) some time ago. Combined with the capability of scripting/recording common actions that could be simply added as “shelf buttons” and the almost-a-pie-menu “Hotbox” - a majority of users could easily find their way through the interface a majority of the time.
Blender has not had the time to stabilise and is, relative to Maya, only recently undergone the intensive introspection in “professional use” required to hammer through the initial developer preferences toward an artist-centric tool. As the user-base grows, the common denominator interface is emerging from each new iteration of Blender, likely to be accelerated now that the interface is scripted.
2. Content Pipeline & Infrastructure
Blender does what it does quite well but, when it comes to working with other applications, it tends to fail somewhat dismally. Maya, though pure dint of being around longer and used by more people, has a large suite of import/export scripts & plugins that make it relatively simple to work in Maya on what it does well then move to other applications (Nuke for example) to do what THEY do well.
Blender is catching up in this regard, the addition of Collada is a good example, but there are big holes that make it impossible for a studio to adopt Blender and keep in budget.
This is ignoring the added cost of putting together a new pipeline, and testing it, for adding Blender to the mix. There are a number of established “tried and true” methods for mixing Maya with other software packages; something Blender has yet to have had the time & exposure to bring about.
3. Training & Existing Investment
Maya is one of the main applications taught at those “exclusive graphic colleges” (i.e. technical schools that make you pay through the nose to learn software); Blender does not have similar training exposure yet.
As such, it is easier for a studio to find another Maya-trained artist than it is to find a Blender-trained one. Finding trained staff and being able to replace them should things not work out is a LARGE consideration to studios both large & small. Never underestimate the cost in time (& hence money) it takes to train someone up, nor the risk one takes on when relying on a (relatively) obscure software package & related skill set. People with Maya skills are almost a dime a dozen (like all skill sets though, good ones are harder to find). Blender training is much rarer and therefore the likelihood of finding another Blender-trained artist with reasonable talent being slim to none.
Update: Yes, I am fully aware that my “summarise briefly” became “outline in this ten page essay”. Sad, but the quick “Blender doesn’t have the training, a stable user interface, or features assisting in a decent pipeline” would have just painted a target on my back for the flame-throwers.