One can make as reasonable a guess as is possible, but not predictable. Most college grads do not get a first job that requires a degree. Only a distinct minority get a job within the field of their degree. It can’t be shown that prediction of what few grads base choice of major on, or stick with for that matter, prove out after that 6-year period of time. Most students simply do not have a qualified clue as to what the future will be and thereby what choice they should make. You can’t even say that the great majority of grads are happy they chose the major they did.
You are probably unaware of this, but I don't need to go as far as what you have in mind to make forecasts. What I am saying is that there exists
some relationship between where you study, what type of degree you get, and the field which dominates your curriculum and your income out of college. I don't really care about why it exists, or its exact form. What I care about is approximating it well enough to have a good prediction. To be specific, I have this equation in mind:
income(i) = f(College Reputation(i), Type of Degree (i), Field of Study(i), Other personal characteristics (i)) + noise(i)
where i is an individual and we'Re looking at income over some period of time after college. My job is to approximate f(.,.,.) well enough to forecast income. I can do it with neural networks, random forests, kernel ridge regression, support vector regression or plain linear regression and I can use my dataset in a smart way to compare how well each method generalizes to observations I do not have used to perform the estimation. By reliable, I mean that I want to beat the "I have no clue model." If you have no clue, your best predictor in this context is the average income -- i.e., you do as if everyone earns the same out of college.*And by "beating it" I specifically mean that, using observations that I did not use either for model comparison or model estimation, my chosen model must beat the other in a mean square sense, i.e.:
average( prediction error(i)^2 | chosen model) < average( prediction error(i)^2 | I have no clue model).
My contention is that it should be easy to beat the "I have no clue" model.
Who cares why people choose what degree they wish? Liks I said, they so often get a job, not in their field of study. Many people of a particular political leaning believe scientists are radical.
Your choices are indicative of your personal characteristics. They can be used as information to make guesses more accurate. Look at how protestors behaved in response to conservatives speaking on college campuses. Look at the number of times they harassed, threatened and sometimes even successfully got people to be fired or ot resign because of a simple disagreement.
They are a noisy minority and it's fairly possible many students pick a humanities major because they don't know what else to do and feel like they have to get a degree. However, it doesn't take a dozen of those lunatics to cause you severe problems. It's not too bad if you can easily fire people because you can always take a chance to pick a rotten apple if you are allowed to throw it back immediately. On the other hand, if you can be seriously stuck with employees who can cause entire hosts of troubles and absolutely can also cause legal trouble if you fire them, that's a risky bet.
Well, my post in this area had to do with real scams, where a lie or some other deception was involved and where the Trump admin is making it more difficult for students to go against the perps or write-off scam debt.
Student debt has been a problem in the United States since a few decades, long before Donald Trump. Laws prohibiting writing-off student debt have also been around before Donald Trump.
Moreover, when you write off a loan, you have to understand that this is a lost to a bank and by extension to many investors. If they didn't scam you, they have no business suffering those losses. And it's not better to ask the government to step in and pay off those debt since it amounts to asking other Americans, who didn't scam you, to pay for your tuition. I agree, however, that if there
REALLY is evidence of a scam on some students, those students should be given a chance to sue those responsible for the scam. But it has to be a real scam, not an imagined scam where people try to foist the blame of their own irresponsible behavior on other people.