The headline percentile on your results is not the percentile of what you scored. It has been moved, on purpose, and I want to explain why before you find it and think something is broken.
The idea
Any measurement is a true value plus some error. If you score very high on a four-item facet, part of that is genuinely being high and part of it is that the four questions caught you well — you read them a certain way, you were in a certain mood, the examples that came to mind were the vivid ones.
That second part does not repeat. Take the test again and the luck resamples.
So the best estimate of your true standing is not your observed score. It is your observed score pulled back toward the average, by an amount that depends on how reliable the scale is:
true estimate = mean + reliability x (your score - mean)
A facet with reliability .90 keeps 90% of your distance from the middle. A facet with .65 keeps 65%. The less the instrument knows, the more it defers to the base rate — which is exactly right, and exactly what a person does not expect a test to do.
This is Kelley's estimate, and it is about ninety years old.
The consequence people notice
Your observed score can sit outside your own confidence interval. At the extremes it usually does.
That looks like an error and is not. The interval is around your estimated true score, and your observed score is a noisy draw that landed further out. If you scored at the 2nd percentile on something, the honest estimate might be the 7th, with a band from the 4th to the 19th — and your raw 2nd sits outside it.
I know this reads as the test arguing with itself. The alternative is worse: reporting extreme scores at face value, where they are most inflated by noise, and then watching them regress on retest while the person wonders what changed about them. Nothing changed about them. The first number was partly luck and nobody said so.
Where it bites hardest
On the shortest scales, because those are the least reliable. Which means the shrinkage is largest exactly where a reader is most likely to be excited by an extreme facet result.
It also means the headline number moves when you change comparison group, because the mean it is pulled toward is that group's mean. That is correct — "where does my true score sit in this group" is the question being asked — and it is worth knowing when you drag the age slider on your results and watch a number move for two reasons at once.