Announcement

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

  • Galbraith_plot STATA 17

    Hi to every body,
    I'm trying to use the Galbraith meta command to detect potential outliers.

    Below the dataset:
    Code:
    * Example generated by -dataex-. For more info, type help dataex
    clear
    input byte Study_num int(Diabetics Smokingdiabetics) double(prev se_prev)
     1  201   60  .29850746268656714 .032276846391339466
     2   71    7  .09859154929577464  .03537948313367658
     3 3319  532   .1602892437481169 .006368152014073769
     4  107   31   .2897196261682243  .04385427912934891
     5   15    2  .13333333333333333  .08777074514725111
     6 2568  121 .047118380062305294 .004181355960040659
     7   50    3                 .06  .03358571124749333
     8  160   42               .2625  .03478449277336095
     9   84    1 .011904761904761904 .011833688064142915
    10   89   10  .11235955056179775 .033475612783467244
    11  118    4  .03389830508474576  .01665940208836309
    12  108   47   .4351851851851852 .047706572241199924
    13   26    3  .11538461538461539  .06265626800161944
    14  474   95  .20042194092827004 .018387127819684504
    15  101   81    .801980198019802  .03965294490504815
    16 1874 1254   .6691568836712913 .010869016127646519
    17  165    5 .030303030303030304 .013345015334864978
    18   40    5                .125  .05229125165837972
    19   94   11  .11702127659574468  .03315458593157387
    20   74    8  .10810810810810811  .03609685639329682
    21 1772  256  .14446952595936793 .008351683101359416
    end

    I use the code below (my effect size is the variable prev):

    meta set prev se_prev

    meta galbraithplot, random

    Click image for larger version

Name:	Galbraith plot.png
Views:	1
Size:	74.4 KB
ID:	1681282



    The potential outliers are the 2 study that lie farther away from the shaded regions.

    I use the code below:

    Code:
    quietly generate lbl = string(trial) if _meta_es/_meta_se > 2 & 1/_meta_se > 4
    meta galbraithplot, mlabel(lbl) mlabpos(12)


    Anyway, when I ran the command it pointed me to different studios. Where is the mistake? Many thanks in advance
Working...
X