I have speech recordings which have been assigned opinion scores (1-5). These recordings were performed by 10 speakers, 5 male and 5 female. I want to investigate the influence of both the specific speaker and the gender of the speaker. How should I model this in a univariate ANOVA (using SPSS)?

When I include the variables speaker (10 levels) and gender (2 levels), the gender variable appears to be ignored (df = 0, type III sum of squares = 0 etc). I assume this is because it contains a complete subset of information available in the speaker variable?

I have also tried modelling the speaker as (5x2) by including the variables speaker2 (5 levels) and gender (2 levels), with the assumption that I am most interested in gender x speaker2. I get results, but if this is correct, how should I report the results for each individual voice quality? Using the F-statistic, partial eta and p values for both speaker2 and gender x speaker2?



I dont think u should use ANOVA for this case, because (if i understand your experiment) you dont have repetitions (you want to studie the differences between the speakers). I think the best solution is the use a Chi-Squared test comparing the male vs female, and then for the speakers (separately)

| cite | improve this answer | |

Your Answer

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

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