0
$\begingroup$

I am currently working on research analysing yes/no responses in a recognition memory task. The false alarm rate is quite high so I have performed a d-prime test and collected d-prime values. Now that I have these d-prime values I do not know how to analyse/report these findings as I have a d-prime value for each individual participant. Some papers show a comparison using an ANOVA but I do not know how to conduct this comparison with the d-prime data or if this is what I should be doing, should I be using the mean of my d-prime values? How should I go about reporting my d-prime results?

$\endgroup$
  • 1
    $\begingroup$ If you have updated information, edit the current post. Does the below answer help you? $\endgroup$ – AliceD Mar 13 at 9:53
0
$\begingroup$

d-prime values are usually fairly well normally-distributed. So you can use any parametric test relying on the normality assumption (in facts you should always z-transform fractions before to use a parametric test on them). If you have only 2 conditions you can use a classical t-test. From your question it seems you have only 1 value per participants so that would probably be an independent t-test. If you have more than 1 comparison an ANOVA would be the typical test to use.

For reporting results it is a bit tricky. d-primes are not that intuitive for people. So you might want to report fraction correct in the text, but run your tests on d-primes. For the same reason you might want to plot your results as fractions. Below is an example of wording (but refer to a statistical textbook for conducting and reporting statistical tests). Note that it is fine to report means and standard deviations of fractions. However parametric tests should always be conducted on z-scores. I know people use fractions in tests all the time but it is wrong.

The mean recognition rate increased from NN% to NN% between the conditions ... and ..., for a mean improvement of NN ± NN% (SE). This difference was significant in an independent t-test on z-transformed recognition rates: t(N)=N, p

| improve this answer | |
$\endgroup$
  • 1
    $\begingroup$ Can you add sources to your answer? $\endgroup$ – AliceD Mar 13 at 10:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.