Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Mixed model postestimation

    I have a longitudinal data set of children from birth to 5 years. During the first year, %body fat is recorded every 3 months and then recorded every year until age 5.
    We are interested in testing the association of the velocity of %body fat gained during the first year and the %body fat at year 5.
    Originally I tested this by generating a variable for the change in % body fat during the first year and regressing it on %body fat at year 5.
    I suspect that I can also do this using mixed models. Although I can fit the LMM. I am having difficulty constructing the needed hypothesis test. Is there a way that I can do this using marginal, contrast or some other way?

    Thank you in advance,
    Mario

    Last edited by Mario Cleves; 09 Feb 2019, 07:28.

  • #2
    Welcome to the Stata Forum / Statalist.

    Please read the FAQ. There you'll find advice about sharing data/command/output. Thanks.

    That being said, and assuming I understood your query correctly, your are asking about postestimation concerning mixed models, yet your performed (OLS?) regression analysis having the 5th year %bodyfat as DV, then "ajusted" for the first years.

    In short, the answer is yes for both questions: yes, this seems to be a nice situation for a mixed model; yes, you can use - margins - for postestimations under a - mixed - design.

    Hopefully that helped.
    Best regards,

    Marcos

    Comment


    • #3
      Thank you Marcus. I don't think that you understood the question. I know that I can use margins after mixed model, that is not the issue. The question is how do you use marginal after mixed models to test the association between fat-accumulation velocity in the first year of life and the resulting fat at 5 years.

      Comment


      • #4
        Let me try this again. I fit the following mixed model:
        Code:
        mixed fat_mass month i.gender gestation  birthlength birthweight || id: month, covar(unstructure)
        Now I want to test the assocation between dy/dx=d( fat_mass)/d(month) at month=12 and fat_mass at month=60

        Thank you,
        Mario

        Comment

        Working...
        X