[Question] How can we measure creativity?

Sta­tus: have spent about two hours on this.

As part of mea­sur­ing how marginal pro­duc­tivity changes over time, I need to know how to as­sess cre­ativity. One promis­ing test for that is the Tor­rance Tests of Creative Think­ing, in which sub­jects are given an open ended prompt, and graded on their an­swers. An­swers are eval­u­ated by a hu­man be­ing for fluency, origi­nal­ity, ab­stract­ness of ti­tles, elab­o­ra­tion and re­sis­tance to pre­ma­ture clo­sure (sam­ple ques­tions for the cu­ri­ous). But does the TTCT pre­dict any­thing we ac­tu­ally care about?

The cre­ator of the tests stud­ied their pre­dic­tive abil­ity in a lon­gi­tu­di­nal study that lasted 50 years so far. The 40 year fol­low up showed good-for-so­cial-sci­ences cor­re­la­tion be­tween child­hood TTCT scores and adult cre­ative achieve­ment, al­though IQ had a stronger cor­re­la­tion. The 50 year fol­low up (con­ducted by differ­ent ex­per­i­menters) found no cor­re­la­tion be­tween score and “pub­lic achieve­ment” un­less com­bined with IQ. Given that these stud­ies were sub­ject to the usual so­cial sci­ence weak­nesses, mul­ti­plied by 50 years and sub­jec­tive grad­ing, I do not count this as strong ev­i­dence.

A smaller study in Brazil found adult­hood scores of rec­og­nized cre­ative achiev­ers and non-achiev­ers to vary.

Ba­sic googling found more aca­demics say­ing both dis­parag­ing (be­gin­ning of page 310) and en­courag­ing things.

My de­fault as­sump­tion is that psy­cho­me­t­ric tests are in­valid, and this ev­i­dence isn’t enough to make me change my mind. But I don’t have any­thing bet­ter to use for my ac­tual goal, which is a mea­surable task that taxes cre­ativity and *noth­ing else*, and this has a cer­tain face val­idity to it. Does any­one have in­for­ma­tion to sway on the val­idity of the test, or an al­ter­na­tive test to use?