Heterogenous relationships between working speed and ability on the Reasoning and Intelligence Online Test (RIOT)

The first peer-reviewed article using RIOT IQ data was published in a scholarly journal. In this study, Dr. Russell T. Warne analyzed how response times on IQ test items related with performance on the same questions. Some of the findings include:

:right_arrow: Each person has a default test-taking speed that is mostly unrelated to test performance. But the nature and strength of that relationship varies across tasks.
:right_arrow: On verbal subtests (e.g., vocabulary), fast responders tend to be more accurate. In other words, you either know the answer quickly or you don’t know it at all.
:right_arrow: On fluid and spatial reasoning tasks, the reverse is true. Slower responders are more accurate. This is an artifact of these items being longer and more complex, especially as they get more difficult. (Training people to take more time on these items won’t improve their score.)
:right_arrow: It is possible to identify people with unusual response patterns on the test, which can help flag cheaters or careless responders.

Link to original post: https://x.com/RiotIQ/status/2008970249664618827?s=20

Full article: https://doi.org/10.1016/j.intell.2025.101990

Oh wow, this actually explains SO much about my test-taking anxiety! I always felt guilty for taking forever on those spatial reasoning puzzles while everyone else seemed to zoom through them. Turns out slower IS better for those? That’s oddly validating lol. Though I’m still probably overthinking the vocabulary ones… classic me :sweat_smile:

This is really fascinating stuff! The part about verbal vs spatial tasks having opposite speed-accuracy relationships makes total sense when you think about it. Like, either you know what “ubiquitous” means or you don’t - staring at it for 5 minutes won’t help. But those matrix reasoning puzzles? Yeah, those need some quality pondering time. My brain hurts just thinking about them honestly.