I am trying to analyse discourse and will code my data. There will be two iterations of this in the form of a pre-task, intervention and post-task involving naturally occurring conversational data. I want to measure the effect of the intervention and so want to use a parametric t-test measure in order to investigate any statistically significant difference.
Hello,
Of course you can use T-test, ANOVA Test, logistic regression( if for example, you want to estimate the effect of certain characteristics on change of behaviour or attitude pro-intervention...)
But, I suggest to use Text mining or qualitattive analysis post and pre intervention.
A lot of R packages are a best tool for such task.
best luck
Hi Chellai, yes, this is my intention, I will use QUAL CA and thematic coding in order to generate the tags for the paired t-test.
Of course Tyrone Bishop , but try not to be limited to the simple T-test ; there are a lot of statistical methods in this field.
Hi Chellai, thanks again, I only have one group of 12 participants with two instances of measurement so the paired t-test suits my needs, for the ANOVA I would need to have repeated measures > 2.
My only concern would be the validity of the coding legend in the actual transcribed data, I suppose as long as I am consistent in counting occurrences of phenomena then I should be OK.
Thanks again for your input.
I'm sorry, I can't help with your question but I was wondering: what is "thematically coded conversation analysis data"? Do you speak of CA in the Sacks/Schegloff/Jefferson way?
Hi Max, yes that's correct. I'm looking for themes which I will tag and count in a pre-task followed by an intervention followed by a post-task. The pre and post tasks will be tagged in a consistent way so the measure is as internally valid as possible. These are the measures to which I intend to apply the paired samples t-test.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics