Like if a person likes blue, trees, and turtles. But doesn't like orange, snow/cold, or fish. They might still become a fisherman in the arctic areas.
To somewhat echo what others have said, if the above person happens to be an Inuit, and doesn't currently have an opportunity for travel to (say) the Caribbean, then, yes, they'd probably plump for the fishing job. At least for the time being.
You
could boil it down to the old criminality adage "means, motive, opportunity".
For any given job, and any given person, can that person:
a) Be able to get employed (or self-employ themself) in that job,
b) Desire to do that work (or at least desire it more than anything else available)
c) Luck out against all the others who want (or will put up with) that job, and who would beat them to it.
It needn't be boolean. Grade each test, and see what jobs each person (or, vice-versa for a given job, which people) come up with the best combined score[1]. It may be
really difficult for said Inuit to get a job building palm-wood huts for the tourists at a turtle sanctuary in the Bahamas, but with their heart set out to get the job and exceptional qualifications in carpentry and zoology gained at nightclasses[2] it might be an absolute shoe-in position for them when they actually apply!
If you don't like the MMO triplet, I'm sure you could work with any other pressures you want. (Parental influences... i.e. tendency to do what the house breadwinner does; Caste[3] expectations... 'your' kind tends to work with horses a lot; Circumstances... A minor war starts, and you go and do your duty, leading onto professional soldiering afterwards, or perhaps becoming mercenary once Your Country No Longer Needs You!, quite so much.)
[1] Straight sum/average, product of all terms, geometric mean, log sum..? .You may get different results for different combinations. To which you might pre-apply weightings to the terms[1a]. A "zero" score for one item could totally scupper some combinatorial methods from winning, and a maximum-value score for one item could allow another method's result to dominate, so tune to your own satisfaction.
[1a] Availability should be quite important, but not create an automatic up-take for the most available jobs nor necessarily be a deal-breaker for the least available. Except, of course when there's absolutely
no call for nuclear scientists, in amongst your typical meso-american indian cliff-dwelling pre-Columbian settlement.
[2] During one winter, when there's an
lot of night...
[3] Consider also gender, physique/build...