Skip to content

Comparing mixed models

6 messages · Carlos Barboza, John Fox

Original
Carlos Barboza
# 
Dear all,
why Anova type III function gives different results if I change the names
of a categorical factor? I suspect that is because contrast type but it's
something strange get different results using the same data.
thank you

Em s?b., 7 de mai. de 2016 ?s 12:26, Carlos Barboza <
carlosambarboza at gmail.com> escreveu:

  
    

      
      
    

  


  


    

    

      
      

        


  

        

      Carlos Barboza
    

    

      
      #
    

  


  

      
sorry, Anova type III function form car package in R

Em ter., 18 de jun. de 2024 ?s 06:59, Carlos Barboza <
carlosambarboza at gmail.com> escreveu:

            

      
      
      
    

        

  
    

      
      
    

  


  


      

    

    
1 day later

    

      
      

        


  

        

      John Fox
    

    

      
      #
    

  


  

      
Dear Carlos,

 From ?Anova:

"Warning

Be careful of type-III tests: For a traditional multifactor ANOVA model 
with interactions, for example, these tests will normally only be 
sensible when using contrasts that, for different terms, are orthogonal 
in the row-basis of the model, such as those produced by contr.sum, 
contr.poly, or contr.helmert, but not by the default contr.treatment. In 
a model that contains factors, numeric covariates, and interactions, 
main-effect tests for factors will be for differences over the origin. 
In contrast (pun intended), type-II tests are invariant with respect to 
(full-rank) contrast coding. If you don't understand this issue, then 
you probably shouldn't use Anova for type-III tests."

I hope this helps,
  John

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Carlos Barboza
    

    

      
      #
    

  


  

      
Thank you Jonh for you answer
I indeed used contr.sum, contr.poly before to model definition
after that I get the same results using any names for the same factors
but, what do you exactly mean when testing for factors for differences over
the origin?
best regards
Carlos


Em qua., 19 de jun. de 2024 ?s 10:14, John Fox <jfox at mcmaster.ca> escreveu:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      John Fox
    

    

      
      #
    

  


  

      
Dear Carlos,

        
On 2024-06-19 11:13 a.m., Carlos Barboza wrote:

            

      
      
      
    

        
The context here is models with both numeric and factor predictors. 
Consider a model of the form y ~ x*f, where x is numeric and f is a 
factor. A type-III test for the "main effect" of f tests for differences 
among factor levels where x = 0 (the origin).

More generally, the issues concerning "types" of tests in models with 
linear predictors are sufficiently complicated that discussing them in a 
help file or by email is likely to prove unsatisfactory. See the first 
two references in ?Anova for more details.

I hope this helps,
  John

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Carlos Barboza
    

    

      
      #
    

  


  

      
Thank you,
yes I know that, it's the same interpretation when using the summary
function when we have f*x
thank you
Carlos

Em qua., 19 de jun. de 2024 ?s 12:20, John Fox <jfox at mcmaster.ca> escreveu: