'Hiding' one level of a dimension when creating tables using Collect

[Serena Luchenski]

Hello

The Data: I have several binary variables indicating presence/absence of a particular health condition and I am comparing them between homeless and housed individuals. For example, the variables will be 1/0 indicating presence/absence of a diagnosis of CVD, Cancer, etc. and I am interested in the proportion with the disease in the homeless population vs the housed population.

The Problem: When I create the table it gives me the proportion of people with the disease (variable = 1) as well as the proportion of people without the disease (variable = 0). I just want it to display the proportion of people with the disease in the table. So essentially, I will have a table with 3 columns - column1 = variable name (e.g. CVD, Cancer, etc.), column 2 = proportion with the disease in homeless population, column 3 = proportion with the disease in the housed population. It's a bit more complicated than this sounds, but if I can solve this simple problem, I can get the rest to work.

Solution Needed: How can get Stata to only display the proportion when the variable = 1 using the Collect command? I can't just use an if statement in the Table command preceding it (e.g. if cancer == 1) as then it won't give me the correct proportion of people with the condition.

I hope that makes sense and I'm grateful to anyone who can help. I have many variables and multiple years of data so I don't want to fiddle around in Excel deleting rows.

Serena

p.s. I have the Table/Collect commands working fine apart from this one issue. Unfortunately I can't copy my code as I am analysing my data in a secure data safe haven.

[Serena Luchenski]

So just to be clear, here's the problem and where I'm trying to get to to:

Current Display

	Homeless %	Housed %
Cancer = 1
Cancer = 0
CVD = 1
CVD = 0

Desired Display

	Homeless %	Housed %
Cancer
CVD

[Rich Goldstein]

your question is not very clear to me (e.g., what commands are you collecting from), but the following may help:

help collect_style_showbase

[Serena Luchenski]

just a simple table command (i.e. tab) for descriptive results. The showbase command works when I use it for regressions, but not for the descriptives/proportions.

Collect refers to variables as 'dimensions' and each dimension has different levels (in this case, level 1 and level 0). So I'm trying to essentially 'hide' level 0. Sorry, I'm finding it hard to explain!

[Maarten Buis]

. // open some example data
. sysuse nlsw88, clear
(NLSW, 1988 extract)

.
. // create a table
. table () (union), stat(fvpercent south collgrad) nototal

-----------------------------------------------------
                                  |    Union worker  
                                  |  Nonunion   Union
----------------------------------+------------------
Lives in the south=Not south      |     53.21   70.50
Lives in the south=South          |     46.79   29.50
College graduate=Not college grad |     77.70   67.90
College graduate=College grad     |     22.30   32.10
-----------------------------------------------------

.
. // what dimensions exist in the table?
. collect layout

Collection: Table
      Rows: result#var
   Columns: union
   Table 1: 4 x 2

-----------------------------------------------------
                                  |    Union worker  
                                  |  Nonunion   Union
----------------------------------+------------------
Lives in the south=Not south      |     53.21   70.50
Lives in the south=South          |     46.79   29.50
College graduate=Not college grad |     77.70   67.90
College graduate=College grad     |     22.30   32.10
-----------------------------------------------------

. // what levels are there for the two dimensions on the rows?
. collect levelsof result

Collection: Table
 Dimension: result
    Levels: fvpercent

. collect levelsof var

Collection: Table
 Dimension: var
    Levels: 0.south 1.south 0.collgrad 1.collgrad

.
. // now we can select the rows we want:
. collect layout (result#var[1.south 1.collgrad]) (union)

Collection: Table
      Rows: result#var[1.south 1.collgrad]
   Columns: union
   Table 1: 2 x 2

-------------------------------------------------
                              |    Union worker  
                              |  Nonunion   Union
------------------------------+------------------
Lives in the south=South      |     46.79   29.50
College graduate=College grad |     22.30   32.10
-------------------------------------------------

[Serena Luchenski]

Amazing! Thanks so much Maarten - this is exactly what I needed!

[lana chahine]

Hello. I have 2 variables, pdrepeat and dlbrepeat. They are coded 0 and 1.
I would like to make a table which shows only rows for value of 1.
I would like frequency to be in 1 column and percentage in the 2nd column.

table (var) (), statistic(fvfrequency pdrepeat ) statistic(fvpercent pdrepeat ) statistic(fvfrequency dlbrepeat ) statistic(fvpercent dlbrepeat )

------------------------------------------------------------------
| Factor variable frequency Factor variable percent
------------+-----------------------------------------------------
pdrepeat=0 | 808 71.50
pdrepeat=1 | 322 28.50
dlbrepeat=0 | 1,029 91.06
dlbrepeat=1 | 101 8.94
------------------------------------------------------------------

collect style row stack, nobinder spacer
collect recode result fvfrequency = column1 fvpercent = column2
(8 items recoded in collection Table)

collect layout (var) (result[column1 column2])

Collection: Table
Rows: var
Columns: result[column1 column2]
Table 1: 6 x 2

-------------------------------
| column1 column2
----------+--------------------
pdrepeat |
0 | 808 71.50442
1 | 322 28.49558
dlbrepeat |
0 | 1029 91.06195
1 | 101 8.938053
-------------------------------

This is the table I would like, except I would like to remove the pdrepeat=0 and dlbrepeat=0 rows.

collect levelsof result

Collection: Table
Dimension: result
Levels: column1 column2 fvfrequency fvpercent

collect levelsof var
Collection: Table
Dimension: var
Levels: 0.pdrepeat 1.pdrepeat 0.dlbrepeat 1.dlbrepeat

I then try the following code and it is not working, any input would be much appreciated.

collect layout (result#var[1.pdrepeat 1.dlbrepeat])

Collection: Table
Rows: result#var[1.pdrepeat 1.dlbrepeat]
Your layout specification does not identify any items.

[Maarten Buis]

. sysuse nlsw88, clear
(NLSW, 1988 extract)

.
. gen byte touse  = !missing(collgrad, south)

.
. table (var) () if touse, ///
>     statistic(fvfrequency collgrad ) statistic(fvpercent collgrad ) ///
>     statistic(fvfrequency south ) statistic(fvpercent south )

----------------------------------------------------------------------------------------
                                  |  Factor-variable frequency   Factor-variable percent
----------------------------------+-----------------------------------------------------
College graduate=Not college grad |                      1,714                     76.31
College graduate=College grad     |                        532                     23.69
Lives in the south=Not south      |                      1,304                     58.06
Lives in the south=South          |                        942                     41.94
----------------------------------------------------------------------------------------

.     
. collect style row stack, nobinder spacer

. collect recode result fvfrequency = column1 fvpercent = column2
(8 items recoded in collection Table)

. collect layout (var) (result[column1 column2])

Collection: Table
      Rows: var
   Columns: result[column1 column2]
   Table 1: 7 x 2

----------------------------------------
                   |  column1    column2
-------------------+--------------------
College graduate   |                    
  Not college grad |     1714   76.31345
  College grad     |      532   23.68655
                   |                    
Lives in the south |                    
  Not south        |     1304   58.05877
  South            |      942   41.94123
----------------------------------------

.     
. collect layout (var[1.collgrad 1.south]) (result[column1 column2])    

Collection: Table
      Rows: var[1.collgrad 1.south]
   Columns: result[column1 column2]
   Table 1: 5 x 2

----------------------------------------
                   |  column1    column2
-------------------+--------------------
College graduate   |                    
  College grad     |      532   23.68655
                   |                    
Lives in the south |                    
  South            |      942   41.94123
----------------------------------------