Where Do College Students Live?
Problems in Migration Data
All migration datasets have faults of one kind or another. I tend to think the American Community Survey is better than the IRS SOI data… but it’s still imperfect. One place where survey data really fails to capture the story is very high mobility individuals: seasonal workers, nomadic peoples, individuals who move multiple times in one year, and, of course, college students. Because these groups do not always have a firmly fixed “residency” and may not even know how to correctly report their own status, it is generally thought that annual surveys of them have large errors. I’ll skip the problems with all of these groups and jump straight to college students.
But first, why I am I talking about this?
Well, it’s been too long since I wrote about migration. No new data comes out until September. So I asked Twitter for suggestions, and got this nerdy request:
Sounds right up my alley.
The Problem of Group Quarters
ACS Tries Very Hard But Is Imperfect
How does the ACS track college students?
Well, that depends on their residential arrangement. College students living in dormitories are classified as living in “non-institutional group quarters.” This classification includes monasteries, rehab centers, workplace housing, and college dorms. Institutional group quarters would include prisons, medical facilities, and group homes for juveniles. These two groups are surveyed separately and, unsurprisingly, ACS has a much easier time getting data from institutional than from non-institutional groups.
But not all college students reside in group quarters! Some reside in houses or apartments near campus.
So we have two different sampling groups into which college students could fall. Some in group quarters, some in non-group-quarters.
Why does this matter? Simple: because ACS uses these groups (group vs. non-group) for the imputation of nonresponse characteristics, especially at the local level, as is most significant for university-related migration.
College students are uniquely bad respondents. Think to your college days and the number of surveys you did not respond to. If you attended during a Census year, think about those valiant but futile Census workers trying to get you to fill out your forms correctly (and how many of you mutated into whatever ethnicity you found most entertaining to claim on that day?). College students are very bad ACS/Census respondents.
The result is that the ACS has serious nonresponse problems in group quarters. Here’s a 2009 notice of low response rates, here’s a 2008 notice, here’s a 2008 statement that group quarters county figures are probably just flat-out unreliable, and here’s ACS’ current fix from 2011 where, kid you not, they don’t even sample the group quarters population of 46% of tracts with such populations, and then do a large-scale imputation.
All that to say that, ahem, the group quarters population component of college students may be very, very unreliable. Unfortunately with this type of error there’s no way we can say exactly how unreliable, because we don’t know if our imputed GQ tracts might be systematically different from our sampled ones. This is an out-of-sample question, not a standard error within the sample.
The Question of Residence
How Do You Know When You Live Where You Live?
Once upon a time, I had the honor of responding to the American Community Survey. It was one of the most exciting moments of my life when, in March of 2013, I got a letter inviting me to respond to the American Community Survey online.
But then they asked me where I lived. And I was perplexed. See, I was in college (yeah I’m a young whipper-snapper). I attended college less than 20 miles from my family home, and semi-regularly still slept at home, especially if I was hosting an event there with friends (my parents love being hospitable for large groups of their childrens’ friends). Even if I didn’t sleep there, I got most mail there.
However, the ACS instructions say that you “reside” somewhere if you usually sleep there for 2 months. I had been at college more than 2 months, so I resided at my university, in group quarters. Except that for group quarters you technically reside there if you live there at all. This is odd: it means that if I’ve lived somewhere a short amount of time, I may be a resident, or I may not, depending on if it’s a group housing situation or not.
It’s also odd because, had the ACS surveyed me in June, I would be a resident somewhere else. Yet I would not have claimed I was a migrant because, really, I was just a resident in two places at once. As of January 2015, the ACS has noticed that this kind of weird residency-change-without-migration situation is fairly common, and they’ve issued guidance to data users.
Many college students face this problem when sampled. Where do you live? Uh, well, multiple places. Nowhere. Here for now. But if you ask me this summer then somewhere else I guess.
ACS Samples All Year
What I Wish They Would Show Us
ACS samples year-round, every month. By and large this is a good practice, as it gets a good average of a place for a year. But with college students it’s a weird thing. The same person can be resident in multiple different places depending on when in the year they’re sampled, yet when asked “Did you live somewhere else a year ago?” they’ll probably answer “no.” Or maybe “yes,” if they “feel” like they migrated.
And what about college students in other housing arrangements, those in houses or apartments? GQ college residents are “resident” if they sleep even 1 night in the quarters where they are sampled. Non-GQ college residents must do 2 months. This could create inconsistent treatment within the “college student” universe.
All of this adds up to a fairly simple problem: ACS data may have very large errors if you want to estimate “college student” characteristics because the sampling frame is inconsistent, student residence can be ambiguous for respondents, and response rates are very low.
This doesn’t mean ACS is useless! It’s just that the ACS is not designed to track college students. It’s designed to track GQ- and non-GQ populations. College students straddle this line, and have unique characteristics that make them hard to study. This isn’t the fault of the ACS, it’s the fault of data users ignoring the source’s limits.
And on a final note, IRS SOI data is so hopelessly bad at tracking college students it’s barely worth noting. Many students are still dependents, and even those who are not may still file returns from their parents’ address or from an accountant’s office. Many students may also be filing for the first time. All of this adds up to IRS SOI being fundamentally useless for tracking most college students. Basically, it only correctly tracks college students who have a job while in college and complete their own taxes. How many college students you know fit that description?
Check out my Podcast about the history of American migration.
If you like this post and want to see more research like it, I’d love for you to share it on Twitter or Facebook. Or, just as valuable for me, you can click the recommend button at the bottom of the page. Thanks!
Follow me on Twitter to keep up with what I’m writing and reading. Follow my Medium Collection at In a State of Migration if you want updates when I write new posts. And if you’re writing about migration too, feel free to submit a post to the collection!
I’m a graduate of the George Washington University’s Elliott School with an MA in International Trade and Investment Policy, and an economist at USDA’s Foreign Agricultural Service. I like to learn about migration, the cotton industry, airplanes, trade policy, space, Africa, and faith. I’m married to a kickass Kentucky woman named Ruth.
My posts are not endorsed by and do not in any way represent the opinions of the United States government or any branch, department, agency, or division of it. My writing represents exclusively my own opinions. I did not receive any financial support or remuneration from any party for this research. More’s the pity.
