Announcement

Collapse
No announcement yet.
X
Collapse
  • Page of 1
  • Filter
  • Time
  • Show
Clear All
new posts
  • #1

    Combining unweighted frequencies and weighted percentages in dtable under svy

    I have a question regarding apply the dtabe command to data with complex weight.

    Here is some sample codes:
    Code:
    webuse nhanes2, clear

svyset [pweight=finalwgt], psu(psu) strata(strata)

* Without survey weight:
dtable bmi i.race, by(sex)

* With survey weight:
dtable bmi i.race, by(sex) svy
    And the results are:

    Code:
    . * Without survey weight:
. dtable bmi i.race, by(sex)

-------------------------------------------------------------------
                                           Sex                    
                           Male          Female          Total    
-------------------------------------------------------------------
N                      4,915 (47.5%)  5,436 (52.5%) 10,351 (100.0%)
Body mass index (BMI) 25.510 (4.024) 25.563 (5.600)  25.538 (4.915)
Race                                                              
  White                4,312 (87.7%)  4,753 (87.4%)   9,065 (87.6%)
  Black                  500 (10.2%)    586 (10.8%)   1,086 (10.5%)
  Other                   103 (2.1%)      97 (1.8%)      200 (1.9%)
-------------------------------------------------------------------

.
. * With survey weight:
. dtable bmi i.race, by(sex) svy

--------------------------------------------------------------------------------
                                                  Sex                          
                             Male              Female               Total      
--------------------------------------------------------------------------------
N                     56,159,480 (47.9%) 60,998,033 (52.1%) 117,157,513 (100.0%)
Body mass index (BMI)     25.480 (3.956)     25.087 (5.462)       25.276 (4.803)
Race                                                                            
  White               49,504,800 (88.2%) 53,494,749 (87.7%)  102,999,549 (87.9%)
  Black                 5,096,044 (9.1%)  6,093,192 (10.0%)    11,189,236 (9.6%)
  Other                 1,558,636 (2.8%)   1,410,092 (2.3%)     2,968,728 (2.5%)
--------------------------------------------------------------------------------
    My question: In public health literature, the convention is to report every weighted item, expect the frequencies which are often either omitted or reported unweighted. Is there a way I can achieve that with code instead of manual replacement?

    This is my "wanted" table:

    Code:
    -------------------------------------------------------------------
                                           Sex                    
                           Male          Female          Total    
-------------------------------------------------------------------
N                      4,915 (47.9%)  5,436 (52.1%) 10,351 (100.0%)
Body mass index (BMI) 25.480 (3.956) 25.087 (5.462) 25.276 (4.803)
Race                                                              
  White                4,312 (88.2%)  4,753 (87.7%)   9,065 (87.9%)
  Black                  500  (9.1%)    586 (10.0%)   1,086  (9.6%)
  Other                   103 (2.8%)      97 (2.3%)      200 (2.5%)
-------------------------------------------------------------------
    Many thanks!

    EDIT: After some digging, I was able to create something with Race:

    Code:
    dtable bmi, factor(i.race, stat(fvrawfreq fvpercent)) by(sex) svy
    which gives:

    Code:
    --------------------------------------------------------------------------------
                                                  Sex                           
                             Male              Female               Total       
--------------------------------------------------------------------------------
N                     56,159,480 (47.9%) 60,998,033 (52.1%) 117,157,513 (100.0%)
Body mass index (BMI)     25.480 (3.956)     25.087 (5.462)       25.276 (4.803)
Race                                                                            
  White                    4,312 (88.2%)      4,753 (87.7%)        9,065 (87.9%)
  Black                       500 (9.1%)        586 (10.0%)         1,086 (9.6%)
  Other                       103 (2.8%)          97 (2.3%)           200 (2.5%)
--------------------------------------------------------------------------------
    However, the numbers in the sample size (N) row remains elusive. Would appreciate any tips.

    EDIT: Woah!! My 1000th post!!
    Last edited by Ken Chui; 27 Nov 2023, 10:06.
    Tags: None
  • #2
    Alright... I ended up solving my own problem. Here is the code:

    Code:
    dtable bmi, factor(i.race, stat(fvrawfreq fvpercent)) sample(, statistics(freq percent)) by(sex) svy
    And we'll get:

    Code:
    -------------------------------------------------------------------
                                           Sex                     
                           Male          Female          Total     
-------------------------------------------------------------------
N                      4,915 (47.9%)  5,436 (52.1%) 10,351 (100.0%)
Body mass index (BMI) 25.480 (3.956) 25.087 (5.462)  25.276 (4.803)
Race                                                               
  White                4,312 (88.2%)  4,753 (87.7%)   9,065 (87.9%)
  Black                   500 (9.1%)    586 (10.0%)    1,086 (9.6%)
  Other                   103 (2.8%)      97 (2.3%)      200 (2.5%)
-------------------------------------------------------------------
    Last edited by Ken Chui; 27 Nov 2023, 10:14.

    Comment

    • #3
      This is the way.
      • 1 like

      Comment

      Previous Next
      Working...
      X