Which test should I use for a within-design with only one measuring point/time?


I will lay out the experiment I am analyzing shorty and a little bit simplified:

Participants were asked to debate a topic with an experimenter. The participant's debating strategies were observed and scored. For example, every participant obtained a score for "negative strategies" (NS) and "positive strategies" (PS). The scores are the duration of time the participant spent using that strategy, in relation to total time of the discussion (= percentage; scores could range from 0 to 1). In total, 5 different strategies were coded and all of them were mutually exclusive.

I now want to investigate if participants, who scored higher in anxiety (questionnaire), are using higher rates of NS than participants with lower scores in anxiety.
Which test could I use for this? The sample is not clinical, so there are no groups of "non-anxious" vs. "anxious". My sample size is small (N=28) - should I split the sample by anxiety median and then calculate differences between the groups? Or can I just simply use a regression?

I apologize because this is probably a very basic question but I am a little bit lost (also I am not an English native speaker, so sorry about that too). I would be very happy to get some answers or ideas! Thank you in advance.


TS Contributor
Do not split the anxiety score, that would be nonsense. You can correlate the NS-rate with the anxiety score, using Pearson's r or Spearman's rho.
In addition, make a scatteplot with NS-rate vs. anxiety score, in order to obtain a visual impression of the association.

With kind regards

Last edited: