Announcement

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

  • melogit (so different from spss genlinmixed?)

    melogit health2013 ib(first).leeftijdscat_4 oplmet || nohouse_encr:, vce(robust) or

    (predicting poor health from age and education taking account of nested structure (individuals within households))

    Several things do not matter so much, robust or not, mean/variance or mode/curvature gauss-hermite, integration 7 or 50. At least, that is what I tested.

    Below you see STATA and SPSS output from this multilevel logistic regression.

    Are the differences in Odds ratios and random effect variance plausible? Particularly the difference regarding the random effect variance (0.539 in stata versus 0.113 in spss) worries me. As I am a STATA newbie, is above melogit command ok? Or did I make a mistake? Or is STATA estimation just better?

    Thanks a lot for any advice.

    Hans

    STATA output

    Mixed-effects logistic regression Number of obs = 1701
    Group variable: nohouse_encr Number of groups = 1479

    Obs per group: min = 1
    avg = 1.2
    max = 4

    Integration method: mvaghermite Integration points = 7

    Wald chi2(4) = 31.77
    Log pseudolikelihood = -780.67238 Prob > chi2 = 0.0000
    (Std. Err. adjusted for clustering on nohouse_encr)
    ----------------------------------------------------------------------------------
    | Robust
    health2013 | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
    -----------------+----------------------------------------------------------------
    |
    leeftijdscat_4 |
    2 | 1.315458 .4699117 0.77 0.443 .6531427 2.649388
    3 | 2.846147 .9623088 3.09 0.002 1.467095 5.521493
    4 | 3.065375 1.048153 3.28 0.001 1.568302 5.991528
    |
    oplmet | .8663129 .0445488 -2.79 0.005 .7832548 .9581786
    _cons | .1337556 .0501185 -5.37 0.000 .0641749 .2787781
    -----------------+----------------------------------------------------------------
    nohouse_encr |
    var(_cons)| .5396133 .5581699 .0710574 4.097849
    ----------------------------------------------------------------------------------

    SPSS output

    leeftijd2 1.30 0.67-2.53
    leeftijd3 2.68 1.47-4.91
    leeftijd4 2.87 1.56-5.29
    oplmet 0.88 0.80-0.96
    cons 0.16 0.09-0.29

    variance: 0.113 st.error: 0.201 p =0.57 95%CI 0.003-3.67














  • #2
    Your -melogit- command is a valid Stata command and it estimates a random-intercepts model. Whether it is the right one for the model you are trying to estimate, and whether it is equivalent to the SPSS model that generated the SPSS output you show--well, those are two different questions.

    To get an answer to your question, you will need to show not just the SPSS output but the SPSS commands you used. And you should really show more of the SPSS output so we can see whether, for example, the N's and numbers of groups are the same in both analyses. You should also explain just what research question you are trying to answer so that someone can help you figure out whether your Stata, or your SPSS, or neither, is the right way to go. ~~In the end, you will probably only hear from people on this list who also know mixed modeling in SPSS, a somewhat limited audience.

    Finally, to make your commands and output align in an easily readable way, please post them as code blocks. To do that, click on the underlined A button to launch the advanced editor features. Then click on the # button. Two code block delimiters will appear. Paste your commands and output between those delimiters.
    Last edited by Clyde Schechter; 21 Sep 2014, 10:38.

    Comment

    Working...
    X