one-way repeated measures ANOVA in groups

[Urmeli Joost]

Hello everyone!

I need some assistance with one-way repeated measures ANOVA. I am using Stata 13 in Windows 10.
I want to use a one-way repeated measures ANOVA to understand whether there is a difference in body mass index (BMI) in three different genotypes of a certain transcription factor (AP2bgeno). I have measured BMI in three diferent time points (subjects 15 years, 18 years and 25 years old). How do I build up the command?

I am quite new to Stata. I tried this:

anova BMI AP2bgeno / ID|AP2bgeno time, repeated(time)

The command Works, but I am not certain if I am getting what I intended?

 anova BMI AP2bgeno / ID|AP2bgeno time, repeated(time)

                           Number of obs =    2630     R-squared     =  0.9723
                           Root MSE      = 2.53114     Adj R-squared =  0.9529

                  Source |  Partial SS    df       MS           F     Prob > F
             ------------+----------------------------------------------------
                   Model |  347928.226  1082  321.560283      50.19     0.0000
                         |
                AP2bgeno |  84.8214041     2  42.4107021       3.23     0.0400
             ID|AP2bgeno |  14161.9889  1078   13.137281   
             ------------+----------------------------------------------------
                    time |  299146.515     2  149573.258   23346.42     0.0000
                         |
                Residual |  9911.14691  1547  6.40668837   
             ------------+----------------------------------------------------
                   Total |  357839.373  2629  136.112352   


Between-subjects error term:  ID|AP2bgeno
                     Levels:  1081      (1078 df)
     Lowest b.s.e. variable:  ID
     Covariance pooled over:  AP2bgeno  (for repeated variable)

Repeated variable: time
                                          Huynh-Feldt epsilon        =  0.8147
                                          Greenhouse-Geisser epsilon =  0.8121
                                          Box's conservative epsilon =  0.5000

                                            ------------ Prob > F ------------
                  Source |     df      F    Regular    H-F      G-G      Box
             ------------+----------------------------------------------------
                    time |      2 23346.42   0.0000   0.0000   0.0000   0.0000
                Residual |   1547
             -----------------------------------------------------------------

Thank You in advance!

[Marcos Almeida]

Welcome to the Stata Forum / Statalist,

You unfortunately didn't give much information about the data display.

Therefore, I recommend you take a look at this text.

[Bruce Weaver]

Do you have data for the same subjects at 15, 18 and 25 years of age? If so, then I think your model is the same as the one shown here:

• https://www.stata.com/support/faqs/s...nova/#wsanova2

Here's the code for that example:

anova lhist group / dog|group time time#group if dog!=6, repeated(time)

Replace lhist with BMI, replace group with AP2bgeno, replace dog with ID, and remove if dog!=6.

anova BMI AP2bgeno / ID|AP2bgeno time time#AP2bgeno, repeated(time)

HTH.

[Urmeli Joost]

Thank You very much Bruce!

Yes, I have longitudinal data from a cohort study, so I have measured the same subjects.

I was actually reading the same text that You both referred to, but could not figure it out.

I am very grateful for the assistance!


Best regards,
Urmeli

[Marcos Almeida]

The link Bruce and I shared goes on the very same verge, as you remarked.

That said, I wonder whether a mixed model wouldn't "fit" better to your needs.

For this, you may type in the command window:

. help mixed

I believe a mixed model approach under - mixed - has more "plasticity", so to speak, if compared to a mixed ANOVA, let alone the issue of the more restrictive assumptions concerning ANOVA