You scored 132. You came back the next day, scored 121, and the internet will not let it go. The honest answer is some combination of four well-understood effects, in roughly this order:
1. Regression to the mean
A score is a draw from a distribution around your true ability, plus noise. When you score noticeably above your usual level, some of that lift came from the noise — an easier-than-average random selection of items, a favourable day. The next draw on the same distribution is, on average, closer to your true level — which means lower. This isn't a psychological effect; it's an arithmetic one. Even with perfect consistency in your ability and your effort, you'd expect the second score to drift toward your long-run average.
Practical estimate: if your first attempt was 1 SD above your true ability (call it +15 IQ points), the expected second attempt is roughly halfway back — about 7 points lower — assuming nothing else changes. Standard error of measurement on a short instrument is around ±5 points either side of that.
2. Item-bank exhaustion
A short test (say 30 items) has a finite bank. On the first attempt, the easy items are easy because you've never seen them; the hard items are hard because the structure is unfamiliar. On the second attempt, the easy items are also familiar, so they don't actually screen you anymore. You essentially get points you'd have gotten anyway, but the items that previously hurt the most still hurt — so your raw score compresses toward the middle.
Mitigation: alternate forms. LabTest IQ's adaptive mode and Form B exist partly for this — if you're re-testing, ask for a different selection of items.
3. Practice effects (the asymmetric one)
Practice effects work upward, not downward — if you actually practised between attempts. But many people retake without training. They do worse on the second go simply because they spent the intervening time on a busy week, didn't sleep well, and approached the test with less adrenaline. The first attempt was a focused 30 minutes; the second was a distracted 30 minutes. The number reflects that.
4. Confidence calibration
On a first attempt, most people guess optimistically. On a second, having already “passed” the test, many people guess conservatively — they leave more items blank or pick the safest-looking option instead of the one they slightly prefer. On a multiple-choice instrument without negative marking, this hurts the score. Mathematically, your expected score is best when you commit to your best guess on every item. Confidence is part of the test.
What to do about it
- Average your attempts rather than treating any single one as canonical. The mean of your last three scores is a better estimate of your true ability than the highest or the lowest.
- Wait at least a week between attempts. Same-day retests are dominated by memory of specific items rather than ability.
- Use an alternate form when one is available. If you trained between attempts, retake on a different form so the lift isn't just familiarity with the same items.
- Don't leave items blank if there's no negative marking. Our scoring rewards correct answers and penalises nothing for wrong ones — so a guess always beats a blank.
The bigger frame
Your IQ is not a single number — it's a confidence interval. The clinical literature reports IQ with an error margin (typically ±5 IQ points on a clinical instrument; wider on an online screening). If your attempts cluster between 115 and 128, your real position on the distribution is somewhere in that band. The fact that one specific attempt came out at 121 instead of 132 is not, by itself, evidence of anything — it's the noise around a measurement that's inherently noisy.
Take it seriously enough to train the weak link. Don't take it so seriously that any single attempt becomes your identity.