3

I'm designing a pretty simple experiment that goes like this. Participants will be shown a series of stimuli and after viewing each one they will answer a few questions where they will make judgments about the stimulus - all Likert items. There are two kinds of stimuli. Probably obvious, but the hypothesis is that there will be a difference between answers for A vs B stimuli. There will be 30 or so stimuli, with an equal number of A and B stimuli. All participants will see all the stimuli (within-subjects).

I'm wondering if there would be a benefit to counterbalancing the order in which they receive the items, vs just showing everyone the same randomized sequence of stimuli (which is easier to setup).

If there's a better method I need to consider, I'd be interested in hearing about it. I also looked into blocking designs, but this is so simple that I don't think those apply here. I'm planning to analyze with t-tests or Mann-Whitney-Wilcoxon.

paul
  • 1,342
  • 3
  • 11
  • 16
  • 30 tests will need a huge sample size to keep the experiment wise error rate low. – Michael M Oct 11 '13 at 08:09
  • Assuming you mean family-wise or experiment-wise error rate, I don't see how this is 30 tests. So I don't think that applies here, but I would be interested in hearing why you think it is. It's one test/experiment, with a single hypothesis and a single variable with two levels. One of the levels could be considered the control. – paul Oct 11 '13 at 15:44
  • You are mentioning multiple Likert-type items. I thought you would want to compare them between groups. – Michael M Oct 11 '13 at 21:08

1 Answers1

1

Randomization of stimuli order should be enough since you don´t have any other grouping variables. http://www.unc.edu/courses/2008spring/psyc/270/001/counterbalancing.html

Milica Popović Stijačić
  • 11
  • 4