After downloading "Most Recent Data by Field of Study" data, almost all the entries are "privacySuppressed". Does this mean that there's actual meaningful data to suppress or is it a cop out to not being able to provide actual data?
1 Answer
It's standard practice -- and a good idea -- to suppress cells that have small numbers of observations. Mainly for protecting the privacy of the individuals whose data are being reported, but also because statistics from small samples are unreliable due to their large standard errors.
I've only looked at a few dozen schools' data. It appears that the College Scorecard suppresses cells with 20 observations or fewer. On the one hand, that's a larger minimum size than is usually used elsewhere (10 is a common number, but studies with small sample sizes might use 5 or even fewer). On the other hand, recent research into "differential privacy" shows that even simple descriptive statistics with fairly large sample sizes can inadvertently reveal private information to a determined investigator (or hacker).
-
I see the reasoning why although it will now be hard to effectively assume accurate data for small schools because of the 20 observation minimum.– JohnCommented Dec 5, 2019 at 6:16