If I have data from an experiment concerning reading comprehension that has a (2x2) mixed design where both variables are the same. What would be the proper way to analyze? That is, there are two groups that can be compared on one variable, and the same variable can be compared within both groups.

Can I "pool" all the subjects, or does that violate some assumptions since the two groups did not have same exact text material? Or should I do a mixed ANOVA although there only is one dependent variable.


I'll try. :) I have let people read 6 short texts and then let them write down a summary of them to try to measure reading comprehension (my dependen variable). Three of those text had inconsistent terminology and three had consistent terminology (my independen variable).

I used two groups. Group A read text #1,#3 and #5 which all had consistent terminology and text #2, #4 and #6 with inconsistent terminology. Group B read text with consistent terminology in #2, #4 and #6. And inconsistent texts in #1,#3 and #5. (phew!)

All texts (#1-6) exists in a consistent and an incónsistent version. Depending on which group the participant were in they read one of them (as described above). I used this design before realising the possible problems of analyzing the results... The reason was to try to control possible inequalities between groups and texts.

So the independent variable is consistent or inconsistent terminology and I measue reading comprehesion as the dependent variable. I have two groups that have read the same texts with some small variations of the independent variable. I also let both groups read (different) texts of both conditions. In short - it sounds like a mixed design, but I only use one independent variable... My question is how to best analyze this. Compare all inconsistent with consistent means i pool all participant into one big group althou they havent read the same texts, or use mixed ANOVA even though I only have one independent variable...

Not sure that clear things up... But I can hope. :)

Example (not the actual words used, just for giving the idea of what words were in the texts):

                       Group A              Group B
Text 1 (healthcare):   ache-ache            ache-pain   
Text 2 (healthcare):   medicine-drug        medicine-medicine  
Text 3(economy):       charge-charge        charge-bill
Text 4 (economy):      salary-salary        salary-pay
Text 5 (computer):     memory-memory        memory-storage  
Text 6 (computer):     internet-cynerspace  internet-internet
  • 1
    $\begingroup$ Can you say more about your situation & your data here? What is the design of your experiment? What do you mean by "both variables are the same"? $\endgroup$ – gung - Reinstate Monica Apr 9 '15 at 18:28
  • 1
    $\begingroup$ I edited my post. Hopefully it makes my issues somewhat more clear. $\endgroup$ – Jimmy Apr 9 '15 at 20:28
  • $\begingroup$ @gung Do you understand his question? Can we model it with six independent variables, each independent variable with two levels (consistent and inconsistent)? The independent variables will be #1, #2, #3 ... $\endgroup$ – SmallChess Apr 10 '15 at 0:09

Your Answer

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

Browse other questions tagged or ask your own question.