Hello!

I think you can try use linear mixed model:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4302710/

For example: (lme4 package)

lmer(reaction_time ~ personality_test_score + (1|id))

Dear Yi Li,

first you have to decide what statistical test would be good to answer your question. Linear mixed models may be a good solution, if you're expecting to see a linear relation between the RTs and and the personality test score (I'll call it PT), as I understand this what you're after. You need to have some hypothesis you are trying to test, so you probably have an idea of how conditions (triangle_absent_4 etc) would influence this relation. Depending on a software, then you have to prepare the data and transform it to an appropriate form. For example, in R for linear regression you will need a long format, so your data will look something like that:

cond RT PT

subject1 c1 123 9

subject1 c2 158 9

subject2 c1 445 11

subject2 c1 456 11

subject3 c1 ... ...

etc.

then you build a model GIVEN your hypothesis and predictions. For example, if you're hypothesising that higher PT scores will improve RTs and this is modulated by a condition in a such a way that you expect an interaction between PT and condition, you would like to build something like:

lme4::lmer(RT ~PT * cond + (1|subject) + (1|cond), data = mydata, REML = F)

Of course, remember to check the assumptions for this test, make sure the link is indeed linear, and decide on the random structure based on your design.

There are plenty of tutorials and sources online to do that.

Good luck!

Hello. It all depends on how you think in the statistical program or by the formula yourself. In any case, the data should be located: the parameters in the columns, the subjects in the rows. As I understand it, you think according to the r-Pearson criterion, the results should be measured in parametric scales and checked for normal distribution.

And I would also like to have more data on what you are doing. If you want send tabular data and I will count.