The ICA research program
The ICA's accuracy is currently being studied. This is the work: the questions we are asking, how we plan to answer them, and every claim we make, with its status.
Three questions
- Structure. Do the four dimensions and three faculties hold together the way the model says they should?
- Stability. Do people's results stay consistent when they take the ICA again over time?
- Relationships. How do ICA results relate to established measures of personality and thinking style?
The protocol sets out the planned method for each question and what result would count as support or as a problem.
Why the questions are hard
The ICA is a forced-choice instrument: you choose among four responses, so the four scores always add to 40. Scores like these, called ipsative, describe the order of a person's own preferences, and standard statistics can mislead when applied to them (Hicks, 1970). Modern methods built for forced-choice data, such as Thurstonian item response models (Brown and Maydeu-Olivares, 2011), make proper testing possible. The glossary explains the terms.
What we publish
The protocol
Questions, planned methods and the criteria for support, set before data collection.
The changelog
Every change to items, scoring or profiles, by version. Current version: ICA 1.1.
Take part
Universities and organizations can contribute consented, anonymized results to the study. See research partnerships, or read how test accuracy is judged first.
Questions
When will results be published?
When the data supports a conclusion. We will not publish interim figures before a planned analysis is complete, and we will publish results whether or not they support the model.
Can I see the scoring?
Yes. ICA scoring is public.
Sources
Keep reading
Take the next step.
The ICA research program: three questions on structure, stability and links to established measures, with the protocol, claims ledger and changelog.