I'm working on the research on the influence of physical/objective parameters of musical instruments on their tone quality.
Each instrument (from a group of 27) is described by a set of measured parameters-like mean value of sound pressure in dB (independant variables) and also a set of subjective parameters (like overall quality of sound) which were rated in scale of 1 to 10.
I need to find out which objective parameters have an influance on the subjective parameters. I'm considering using MANOVA for this but I'm afraid that it only gives a simple answer whether the group of independant variables have or have not an influence on the independant variables.
I'm looking for the better method of statistical analysis to determine which specific independant variables influence the specific independant variables.
I would much appreciate for any suggestions.
If you are willing to accept that the ratings are interval scale and the distribution is multivariate normal, and the covariances are homogenous (as you did when applying the MANOVA), then discriminant analysis would be the natural successor. It allows for evaluating the impact of each rating upon the group (instruments) differences.
Thank you very much for your reply. I will try to use discriminant analysis. And what's your opinion about multiple regression or canonical analysis in this case?
I agree with Alexandrowicz in the sense that both MANOVA and discriminant analysis will give you two answers. The MANOVA will give you the answer of a combination of variables that could showed if there is a difference in the mean of groups and the discriminant analysis will give you the linear combination of these variables that produce the difference and the contribution of each variable to this difference.
Since your response is ordinal you should use ordinal regression models.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis