1) Regarding the question
Are we free to choose our beliefs?
Since you posted in a philosophy forum, I take it that you are not interested (only) in the empirical question pertaining to psychology.
The technical term in philosophy is doxastic voluntarism (DV), i.e. the thesis that beliefs are subject to the will and, as such, that we are able to choose what to believe. Whether DV is a defendable claim is debated in current epistemology, as well as in philosophy of religion and, a bit more surprisingly, in philosophy of science.
The consensus seems to be that DV is untenable and thus doxastic involuntarism is the standard position. This shows in the use of the technical term, which is employed – in a similar way to "relativism" and other fighting words – as ultimate objections in arguments ("Position X implies DV and is therefore untenable").
Interestingly enough, it seems that in the history of philosophy the situation was actually reversed, as DV was maintained by many central figures in philosophy (Thomas Aquinas, Descartes, Locke, ...). Although a minority position these days, there are quite important names who defended (some forms of) DV. Among them are Roderick Chisholm, Jerry Fodor and Bas Van Fraassen. Particularly the latter draws heavily from the pragmatist's toolbox, which was arguably the only tradition where DV was more or less accepted.
2) Regarding your question:
Is it possible to choose to believe something one finds absurd?
The same question arises whether one can choose to believe a claim that she agrees rationally to be false. The problem is that a pragmatic justification of beliefs (i.e. based on utility and means-ends arguments) may clash with a (lack of) epistemic justification. Consider the following:
P is false. However, believing that P has a practical advantage. Therefore one should believe that P.
It is hard to imagine that the pragmatic justification might override the belief that P is false. The debate whether this epistemic self-deception may be possible is one of the crucial topics in current debates around DV. (Again, the consensus is that it is not and thus is an objection to DV.)
P is probably false. However, believing that P has a great practical advantage. Therefore one should believe that P.
Does this change your evaluation that overriding is possible? Probably not. However, in general it may be possible to construct a case such that the epistemic justification is weaken so much (and the pragmatic justification is strong enough) to produce a tradeoff where an override is possible.
To show another route, consider a more concrete example (presupposing that pessimistic meta-induction is a sound argument):
Our best currently available scientific theory T is probably false. However, believing that T is probably false (and thus adopting an instrumentalist attitude) diminishes the empirical testability of T, while believing that T is probably true increases the empirical testability of T. Scientist consider the empirical testability of T to be crucial. Therefore, scientists should believe that T is probably true.
What do you think now? Making the practical advantage to matter epistemically, seems to kinda blur the line here. In general, making these abstract schemata more concrete could show that there are contexts in which DV might be more plausible than in others, thus pointing to the context-sensitivity of DV.
3) Regarding Pascal's wager:
Pascal's famous argument is used as textbook example of DV, but it is unclear whether Pascal actually presupposed DV in his argument. If I recall correctly, he didn't claim that such form of practical reasoning could lead to a voluntary belief in God, but that it could lead to take part in the religious practice, which in turn would create an environment where a belief in God could grow eventually.