Multiway Table, hide certain levels of variable

[Castor Comploj]

I am using this manual, on p10:
rtablemultiway.pdf (stata.com)

I want to create this table (see img), but only showing the percentages with "yes" (the right columns for each variable). How do I do this?

[Andrew Musau]

help collect levelsof

use https://www.stata-press.com/data/r18/nhanes2l, clear
label define yesno 0 "No" 1 "Yes"
label values highbp diabetes heartatk yesno
label variable diabetes "Diabetes"
collect clear
quietly: table (sex agegrp) (diabetes), nototals statistic(percent, across(diabetes)) name(table1)
quietly: table (sex agegrp) (highbp), nototals statistic(percent, across(highbp)) name(table1) append
quietly: table (sex agegrp) (heartatk), nototals statistic(percent, across(heartatk)) name(table1) append
collect layout (sex#agegrp) (diabetes highbp heartatk)
collect layout (sex#agegrp) (diabetes[1] highbp[1] heartatk[1])

Res.:

. collect layout (sex#agegrp) (diabetes highbp heartatk)

Collection: table1
      Rows: sex#agegrp
   Columns: diabetes highbp heartatk
   Table 1: 17 x 6

-----------------------------------------------------------------------------------
              |        Diabetes         High blood pressure     Prior heart attack 
              |        No        Yes          No         Yes          No        Yes
--------------+--------------------------------------------------------------------
Sex           |                                                                    
  Male        |                                                                    
    Age group |                                                                    
      20–29   |  99.64158   .3584229    73.92473    26.07527         100           
      30–39   |  99.61039   .3896104    62.33766    37.66234    99.74026   .2597403
      40–49   |  97.37705   2.622951    55.08197    44.91803    98.03279   1.967213
      50–59   |  94.68439   5.315615     42.3588     57.6412     92.3588   7.641196
      60–69   |  91.96494   8.035062    41.49014    58.50986    86.55953   13.44047
      70+     |  88.39286   11.60714     32.8125     67.1875    83.48214   16.51786
  Female      |                                                                    
    Age group |                                                                    
      20–29   |  99.08638   .9136213     91.6113    8.388704    99.91694   .0830565
      30–39   |  97.88484   2.115159     80.6338     19.3662    99.76498   .2350176
      40–49   |  96.06657   3.933434    65.55891    34.44109    98.78971   1.210287
      50–59   |  94.19448   5.805515    48.62119    51.37881    96.66183   3.338171
      60–69   |  91.41516   8.584842    41.91818    58.08182     94.5674   5.432596
      70+     |  89.03346   10.96654    33.45725    66.54275    92.00743   7.992565
-----------------------------------------------------------------------------------

. 
. collect layout (sex#agegrp) (diabetes[1] highbp[1] heartatk[1])

Collection: table1
      Rows: sex#agegrp
   Columns: diabetes[1] highbp[1] heartatk[1]
   Table 1: 17 x 3

--------------------------------------------------------------------
              |  Diabetes   High blood pressure   Prior heart attack
              |       Yes                   Yes                  Yes
--------------+-----------------------------------------------------
Sex           |                                                     
  Male        |                                                     
    Age group |                                                     
      20–29   |  .3584229              26.07527                     
      30–39   |  .3896104              37.66234             .2597403
      40–49   |  2.622951              44.91803             1.967213
      50–59   |  5.315615               57.6412             7.641196
      60–69   |  8.035062              58.50986             13.44047
      70+     |  11.60714               67.1875             16.51786
  Female      |                                                     
    Age group |                                                     
      20–29   |  .9136213              8.388704             .0830565
      30–39   |  2.115159               19.3662             .2350176
      40–49   |  3.933434              34.44109             1.210287
      50–59   |  5.805515              51.37881             3.338171
      60–69   |  8.584842              58.08182             5.432596
      70+     |  10.96654              66.54275             7.992565
--------------------------------------------------------------------

.

[Castor Comploj]

Thanks. The original tables basically include 2 decimals, but -collect layout- loses this format.

-collect layout (male#agegrp5) (d_diab[1] d_hibp[1]), nformat(%9.2f percent)- does not work

How do I keep only 2 decimals in this case?

[Andrew Musau]

The dimension is sex and "Male" is a level of this dimension. You can apply the formatting as below:

use https://www.stata-press.com/data/r18/nhanes2l, clear
label define yesno 0 "No" 1 "Yes"
label values highbp diabetes heartatk yesno
label variable diabetes "Diabetes"
collect clear
quietly: table (sex agegrp) (diabetes), nototals statistic(percent, across(diabetes)) name(table1)
quietly: table (sex agegrp) (highbp), nototals statistic(percent, across(highbp)) name(table1) append
quietly: table (sex agegrp) (heartatk), nototals statistic(percent, across(heartatk)) name(table1) append
collect layout (sex#agegrp) (diabetes highbp heartatk)
collect style cell diabetes highbp, nformat(%3.2f)
collect layout (sex[1]#agegrp) (diabetes[1] highbp[1])

Res.:

. collect style cell diabetes highbp, nformat(%3.2f)

. 
. collect layout (sex[1]#agegrp) (diabetes[1] highbp[1])

Collection: table1
      Rows: sex[1]#agegrp
   Columns: diabetes[1] highbp[1]
   Table 1: 9 x 2

-----------------------------------------------
              |  Diabetes   High blood pressure
              |       Yes                   Yes
--------------+--------------------------------
Sex           |                                
  Male        |                                
    Age group |                                
      20–29   |      0.36                 26.08
      30–39   |      0.39                 37.66
      40–49   |      2.62                 44.92
      50–59   |      5.32                 57.64
      60–69   |      8.04                 58.51
      70+     |     11.61                 67.19
-----------------------------------------------

.

[Castor Comploj]

Thank you so much again.
Sorry to ask one more question. The manual on -collect- is very extensive and it can be hard to find some things.

How do I add the number of nonmissing observations to this? The table I create in this code includes this:
-quietly: table (sex agegrp) (diabetes), nototals statistic(percent, across(diabetes)) statistic(count diabetes) name(table1)-

[Andrew Musau]

You should focus on looking at dimensions of the table and levels of these dimensions to understand how to modify and construct tables.

help collect dims

use https://www.stata-press.com/data/r18/nhanes2l, clear
label define yesno 0 "No" 1 "Yes"
label values highbp diabetes heartatk yesno
label variable diabetes "Diabetes"
collect clear
quietly: table (sex agegrp) (diabetes), nototals statistic(percent, across(diabetes)) statistic(count diabetes) name(table1)
quietly: table (sex agegrp) (highbp), nototals statistic(percent, across(highbp)) statistic(count highbp) name(table1) append
quietly: table (sex agegrp) (heartatk), nototals statistic(percent, across(heartatk)) statistic(count heartatk) name(table1) append

*LOOK AT DIMENSIONS OF TABLE
collect dims

*LOOK AT LEVELS OF DIMENSION RESULT
collect levelsof result

*FORMAT CELLS
collect style cell result[percent], nformat(%3.2f)
collect style cell result[count], nformat(%9.0f)

collect layout (sex#agegrp) (result#(diabetes[1] highbp[1] heartatk[1]))

You can change the column labels.

Res.:

. collect layout (sex#agegrp) (result#(diabetes[1] highbp[1] heartatk[1]))

Collection: table1
      Rows: sex#agegrp
   Columns: result#(diabetes[1] highbp[1] heartatk[1])
   Table 1: 17 x 6

--------------------------------------------------------------------------------------------------------------------------------------------------------------
              |   Percent   Number of nonmissing values               Percent   Number of nonmissing values              Percent   Number of nonmissing values
              |  Diabetes             Diabetes            High blood pressure       High blood pressure       Prior heart attack        Prior heart attack    
              |       Yes                           Yes                   Yes                           Yes                  Yes                           Yes
--------------+-----------------------------------------------------------------------------------------------------------------------------------------------
Sex           |                                                                                                                                              
  Male        |                                                                                                                                              
    Age group |                                                                                                                                              
      20–29   |      0.36                             4                 26.08                           291                                                  
      30–39   |      0.39                             3                 37.66                           290                 0.26                             2
      40–49   |      2.62                            16                 44.92                           274                 1.97                            12
      50–59   |      5.32                            32                 57.64                           347                 7.64                            46
      60–69   |      8.04                           110                 58.51                           801                13.44                           184
      70+     |     11.61                            52                 67.19                           301                16.52                            74
  Female      |                                                                                                                                              
    Age group |                                                                                                                                              
      20–29   |      0.91                            11                  8.39                           101                 0.08                             1
      30–39   |      2.12                            18                 19.37                           165                 0.24                             2
      40–49   |      3.93                            26                 34.44                           228                 1.21                             8
      50–59   |      5.81                            40                 51.38                           354                 3.34                            23
      60–69   |      8.58                           128                 58.08                           866                 5.43                            81
      70+     |     10.97                            59                 66.54                           358                 7.99                            43
--------------------------------------------------------------------------------------------------------------------------------------------------------------

.

[Castor Comploj]

How do I change the label "Number of nonmissing values" to "N"?

[Castor Comploj]

Also, in addition to #7, the solution above is not complete. The number of nonmissing observations that I need in this case is the total yes/no, not only those with diabetes==yes.
See this, the 4 is the one that came out from your answer #6:

. bys sex: tab agegrp diab

-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
-> sex = Male

           |       Diabetes
 Age group |        No        Yes |     Total
-----------+----------------------+----------
     20–29 |     1,112          4 |     1,116 
     30–39 |       767          3 |       770 
     40–49 |       594         16 |       610 
     50–59 |       570         32 |       602 
     60–69 |     1,259        110 |     1,369 
       70+ |       396         52 |       448 
-----------+----------------------+----------
     Total |     4,698        217 |     4,915