0
$\begingroup$

I am going to do an experiment with English and Dutch sentences. The same participants read both the English and the Dutch sentences. There are two options participants can choose after they've read the sentence, option A and B (both options represent a certain interpretation of the sentence).

So there are four possible outcomes: EnglishA, EnglishB, DutchA, DutchB. I've entered some fictional data in spss like this: fictional data

The numbers under each option represent how many times a participant chose this option. I can analyse my data with paired t-tests to compare Dutch_A and English_A for example. But I want to do 4 comparisons and was taught that you cannot do multiple t-tests with the same data because it influences the significance value. I don't know what other test I can use, I tried several options (assuming that my data meet the assumptions for parametric tests) in SPSS but failed to find one that works.

$\endgroup$

1 Answer 1

0
$\begingroup$

You can perform multiple comparisons and use an adjustment to the significance level. The simplest one is bonferroni where you take 0.05 / number of comparisons. So if you are doing six that means you should use a level of 0.008.

$\endgroup$
2
  • $\begingroup$ While you have addressed the issue with multiple comparison. We shouldn't do a t-test here. $\endgroup$
    – SmallChess
    Mar 29, 2017 at 15:41
  • $\begingroup$ Yes it should be a test for categorical data but I wanted to clarify on multiple comparisons $\endgroup$ Mar 29, 2017 at 18:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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