Choosing a version
IPIP-NEO-300 or IPIP-NEO-120?
Both versions measure the same five domains and thirty facets. They differ in how many questions stand behind each score.
| IPIP-NEO-300 | IPIP-NEO-120 | |
|---|---|---|
| Items in total | 300 | 120 |
| Items per facet | 10 | 4 |
| Items per domain | 60 | 24 |
| Published by | Goldberg (1999) | Johnson (2014) |
| Time to complete | About two and a half times as long | Median 8 min 10 s in a JobCannon sample of 3,079 |
| Facet scores | Ten items each | Four items each |
Length and precision
A scale built from more items is more reliable, because random noise in single answers averages out. The standard way to estimate the effect is the Spearman–Brown formula. In a JobCannon sample of 3,079 completed IPIP-NEO-120 results, the median facet alpha was .774 and the lowest was .565 (technical manual). Lengthening a four-item facet to ten items of similar quality predicts an alpha of about .90 for the median facet and .76 for the lowest. These are predictions from the formula, not measurements of the 300-item version.
Domain scores are far less affected. A domain in the 120-item version already pools 24 items; across the five domains alpha ran from .797 to .903 in the same sample.
Which one to use
- Facet-level work on individuals: the 300-item version gives each facet ten questions of evidence.
- Domain-level work, surveys, screening or any setting where time is short: the 120-item version covers all thirty facets in a sitting of about eight minutes.
- Comparing people with a reference group: use the same version as the reference data. A 120-item score and a 300-item score are on different scales.
A facet from the 120-item version is best read as a pointer to a part of a domain, and the questions behind it are listed on the facets page.