Comparing the two versions
Same inventory, same five domains, same thirty facets. The long version asks ten questions per facet and the short one asks four, and almost every difference between them follows from that one fact.
Free, no sign-up, results right away
If you have the time, take the full version: facets are the reason to be here, and they get sharper with more questions. If you do not, the short one is a real test and not a preview.
You want the most detailed picture the public-domain inventory can give, and you can sit with it for a while.
You want a solid profile today, on a phone, in one sitting, without setting aside most of an hour.
Everything below follows from how many questions each facet gets.
| What | IPIP-NEO-300 | IPIP-NEO-120 |
|---|---|---|
| Questions | 300 statements | 120 statements |
| Time | about 40 minutes | about 15 minutes |
| Domains | all five | all five |
| Facets | all thirty | all thirty |
| Questions per facet | ten | four |
| Percentile norms | own tables, by sex and age band | own tables, by sex and age band |
| Best for | a facet profile you will actually work with | an honest overview in one sitting |
Each version is scored against its own norms. A percentile from the 300 and a percentile from the 120 are not two measurements of different things, but they are computed from different samples, so small differences between them are expected.
Reliability here is Cronbach's alpha, computed from the same open datasets our percentile tables come from. Read it as consistency: how much the questions inside one scale agree with each other. Higher means the number moves less when you answer on a different day.
At the domain level the gap is small, because a domain is still built from sixty or twenty-four answers. Both versions land where psychometrics calls a scale strong, and for the five headline numbers the short form loses very little.
Average across the thirty facets
0.84
300 items
0.76
120 items
This is where the two versions really part. A facet built from ten answers holds together noticeably better than one built from four, and that difference shows up as wider uncertainty around each individual facet percentile in the short form.
Alphas are computed from Johnson's open datasets, the same source as our norms. They describe people who took the inventory online, not the general population.
n = 135,947 / 385,801
The 120 is not a different questionnaire. Every one of its statements is also in the 300, picked by the author as the four best markers of each facet.
That is why you cannot take one and then take the other for a fresh opinion: the second reading would repeat questions you have already answered. It is also why the two profiles usually look alike, and why a large disagreement between them is worth a second look rather than a shrug.
You can, and the interesting part is the comparison rather than either result on its own. Leave time between them: a few weeks is enough for the day's mood to stop carrying over, and answers you still remember are not independent ones.
Expect small differences everywhere. A facet moving five or ten percentiles between versions is ordinary measurement noise, especially on the short form, and it does not mean one of the two is wrong.
A gap of thirty percentiles or more on the same facet is worth reading twice. Usually the explanation is human rather than statistical: a rushed session, a rough week, or a question understood differently the second time. When two readings disagree that loudly, the full version is the one to trust.
Both versions are scored the same way and both give you the full base result. Pick the one that matches the time you actually have right now.
300 questions · about 35 minutes
Take the full test120 questions · about 15 minutes
Take the short versionFree, no sign-up, results right away