Two very different things look the same
Every dataset has two ways of saying "nothing here". One is a genuine zero: the value really is nought. The other is missing: the value exists in the world, but nobody measured it, or the measurement is not reliable enough to publish. Written down as numbers, both can look like 0 — and that is how false claims get made.
Why the government suppresses figures
- Small cohorts. If only 15 students were in the measured group, an average would be both unreliable and potentially identifying. The Department withholds it.
- Privacy. Earnings come from tax records. Cells that are too small are suppressed so individuals cannot be inferred.
- Programme scope. Some programme types fall outside the data collection entirely, so no figure exists for them at any size.
None of those reasons is about quality. A college with a suppressed graduation rate may be excellent; it may simply be small. The same is true of a large college where a particular programme has few students.
The sorting trap
Missing values create a specific hazard when you sort a list. If blanks are sorted as zero, a college with no data appears at the bottom of a "worst graduation rate" list — which is a claim the data does not support. If blanks are sorted as the largest value, the same college appears at the top of a "best" list, which is equally unsupported.
On this site, missing values sort to the end regardless of direction, and the list shows how many institutions actually had a figure. That way the ranking reflects only the institutions that reported.
How to read a page full of blanks
- A blank is a question, not an answer. If a figure matters to you, ask the college for it directly.
- More blanks is not automatically worse. It usually tracks size and programme mix, not quality.
- Watch the denominators. If a page says a figure is based on 40% of institutions, the other 60% are not part of that number.
- Beware any comparison that silently treats blanks as zeros. Ours does not, and you should check whether others do.