I need help to define which test to use. I want to know which biochemical parameters are influencing the reproductive gonad stage in a scallop.

Gonads have a great variability therefore each gonad stage was determined by taking 50 random points in each gonad and scoring the percent in three categories: "spawning", "mature" and "atresia".

My explanatory variables are a set of environmental variables: temperature and chlorophyll and biochemical composition inside the gonad and the muscle of the scallop: glycogen, protein, lipids.

Can anyone please tell me how can I analyze these data?

  • 2
    $\begingroup$ If you have multiple dependent variables, you could do multivariate multiple regression or MANOVA. This post might be helpful too. Or this one. $\endgroup$ May 31, 2013 at 6:10
  • $\begingroup$ Hi, please correct me if I am wrong but MANOVA would not be good because I have several independent variables that are continuos and MANOVA is used usually to compare groups, right? $\endgroup$ Jun 3, 2013 at 3:28
  • $\begingroup$ Also the dependent variables are percentage of gonad reproductive stage, so they are inter-related: a percentage of gonad in atresia will affect the percentage of gonad in mature stage. How to account for that? Thanks in advance, Tania $\endgroup$ Jun 3, 2013 at 3:46
  • 1
    $\begingroup$ The dependent variables should be moderately correlated with each other because if they wouldn't, you may as well do separate ANOVAs. If the DVs are strongly correlated, you may as well just use one DV for all of them: see this post. Because you have percentages, you may have to perform a multivariate GLM. Maybe this post helps. $\endgroup$ Jun 3, 2013 at 6:37

1 Answer 1


I would suggest you to clearly mention you the number and measurement scale (metric or non-metric) response and predictor variables in order to make others understand the problem correctly.

If I understood correctly, you have multivariate multiple regression problem. One possible solution is to use MANCOVA (multivariate analysis of co-variance).

Edit: R package jmv (The 'jamovi' Analyses) may be useful.

Refer: This and this


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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