“Does your major really matter?”
Students in their junior or senior years have likely said to themselves “duh,” upon reading the question above. For them, the harsh realities of the job and internship market have prevailed, leaving them convinced that their choice of school and major did, for better or for worse, matter.
But for the young student just beginning college, these are legitimate questions — as they should be. Figures are often published through Forbes or CNN showing that anthropology majors have the highest unemployment or that Princeton graduates have the highest expected salaries. Yet a careful statistician will note that these figures on college majors are crucially flawed.
To be sure, it is well documented that students from more selective schools earn higher salaries later on. Why this is the case, however, is far from obvious. Some would argue that more selective schools have better professors, higher endowments and thus provide an all-around better education for their undergraduates. Or is it that more selective schools attract students that are already hard-working and intelligent, and thus likely to be successful purely by their own merit?
Empirical studies suggest the reality is almost entirely the second case. A well-known 1999 study by economists Stacy Dale and Alan Krueger found that the expected salary of students at highly selective schools could be explained almost entirely by admission into such schools. For example, students who went to Yale made on average the same amount of money as students who turned down Yale for Penn State.
A second version of this study, published in 2011, found the bolder result that salary could be explained almost entirely by the schools students applied to. In other words, students who went to Yale made on average just as much as students who applied to Yale, but did or did not get in. There is, however, this caveat — for students of lower income, certain minorities, or students whose parents did not attend college, attending a more selective school does yield an intrinsic benefit.
The reality is that highly qualified and well-connected students enter highly ranked schools and leave just as they entered — qualified and well-connected. The actual education that these schools provide adds little for the majority of students who attend. In other words, your school could very well not matter. The raw data, unfortunately, suggests the opposite at a glance.
The underlying issue here is what is called self-selection bias in statistical theory. When we want to establish a causal relationship between two variables — in this case, college selectivity and income — we try to identify groups of people that are more or less identical, and observe how they change with differing levels of some variable — education, in this example. We hope that individuals are sorted into groups randomly, as this would ensure that each of our populations are roughly the same on average.
However, if people self-select into the groups — as they do when applying for colleges — they will tend to differ, if preference for entering a group is correlated with other meaningful variables, such as income or ability. Thus it is difficult to isolate the effect of a certain college on your income, happiness, success or any variable, for that matter, as more than one trait is varying between groups. In effect, we would be comparing oranges and apples.
The same principle applies to college majors, which are — if anything — self-selected groups. For example, it has been documented that women are overrepresented in certain less lucrative majors such as education and the humanities. Furthermore, it is well documented that women sometimes face discrimination in the workplace irrespective of their qualifications, resulting in lower salaries. Thus, the average salaries of majors in the humanities and education — already relatively low — will be skewed even further downward, below their true value.
This is just a motivating example. There could be any number of selection effects at work when students choose their majors. This means that while we have a strong answer to whether your choice of college really matters, we have less of a response as to whether a major choice matters.
There is quite a bit of research in the education, economics, and sociology literature examining how the expected salary of a given major changes based on demographic data, but there is little work trying to correct for the selection bias problem. Without further efforts on the part of researchers, these figures will remain distorted.
Perhaps due to the recent economic climate, there has been pressure on universities to provide undergraduates information on earning potential by major. It seems that at this point, such information would be of little use. Perhaps colleges should start by first trying to account for the selection bias problem, and then providing their students with data that more accurately reflects the nature of the job market.
*Because of an editing error, the original excerpt for this article has been changed. WSN regrets the error.
Nickhil Sethi is a contributing columnist. Email him at [email protected].
WSN sucks • Dec 17, 2013 at 3:06 am
I hope you don’t consider yourself a careful statistician. This article is absolutely terrible. You figured out NOTHING. Yes, people self select into their majors, that’s how a major works. All the earnings statistics show is that smarter people who enter smarter majors enter smarter careers and therefore make more money. There SHOULD be a correlation between the intelligence of people who go into, say, engineering, and their salaries; claiming otherwise is just a poor attempt by a student who just took intro to psych or statistics to invalidate earnings causality purely on “self-selection bias”, which is not at all relevant to the accuracy of the statistics.