st: nested random effects anova

[Inna Khousnoullina <inna.khousnoullina@uni-konstanz.de>]

Dear statalist-participants,


I'm  looking for an answer on the following question and hope to find it
here.
I run a two-way nested anova. My dependent variable is f1 and I'm
searching for an effect of the interviewer gender and respondent gender
on the respondent behavior (f1). so respondents are nested within
interviewers. 
I get different results depending on the residual error I choose. 
Here are my outputs:
. anova f1 sex_int / sex|sex_int /

   Number of obs =    3528     R-squared     =  0.0138
   Root MSE      = 2.34259     Adj R-squared =  0.0129

Source |  Partial SS    df       MS           F     Prob > F
            
------------+----------------------------------------------------
 Model |    269.922069     3  89.9740231      16.40     0.0000
                      
sex_int |    218.147357     1  218.147357      12.56     0.0712
sex|sex_int |  34.7409223   2  17.3704612   
            
------------+----------------------------------------------------
sex|sex_int |  34.7409223     2  17.3704612       3.17     0.0423
                         
   Residual |  19338.7718  3524   5.4877332   
            
------------+----------------------------------------------------
      Total |  19608.6939  3527  5.55959566   

. anova f1 sex_int  sex|sex_int 

  Number of obs =    3528     R-squared     =  0.0138
  Root MSE      = 2.34259     Adj R-squared =  0.0129

 Source |  Partial SS    df       MS           F     Prob > F
            
------------+----------------------------------------------------
  Model |    269.922069     3  89.9740231      16.40     0.0000
                         |
sex_int |    218.147357     1  218.147357      39.75     0.0000
sex|sex_int |  34.7409223   2  17.3704612       3.17     0.0423
                         |
   Residual |  19338.7718  3524   5.4877332   
            
------------+----------------------------------------------------
      Total |  19608.6939  3527  5.55959566   

My question is: which output should I take for calculating intraclass
correlation coefficient? In the first case my major effect (interviewer
gender') is not significant. I would prefer the second one. By are there
some rules in this case?


Thank you.

Inna Khousnoullina


*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/