Dummy creation

[Jon Grob]

Hi there,

Using the following code, i was unable to create dummy variables for this generated interval. I result does not exactly provide dummy variables.

use poverty, clear
keep xhpsu xhnum hhsize
gen hhsize_grp=""

replace hhsize_grp = "1-2 persons" if hhsize <= 2
replace hhsize_grp = "3-4 persons" if hhsize >= 3 & hhsize <= 4
replace hhsize_grp = "5-6 persons" if hhsize >= 5 & hhsize <= 6
replace hhsize_grp = "7-8 persons" if hhsize >= 7 & hhsize <= 8
replace hhsize_grp = "9-10 persons" if hhsize >= 9 & hhsize <= 10
replace hhsize_grp = "11 or more" if hhsize >= 11

Is there any I could get dummy variables for this? Thanks

[Nick Cox]

You're creating a string variable with 6 categories. That may have uses for tables or graphs, but it's not a dummy or indicator variable, at least as I understand the term.

If you use i.hhsize as a factor variable in model fitting you can think about further changes, if (for example) the pattern of coefficients suggests a simpler parameterization or you decide that some sizes have too few observations to work well as separate levels.

[Dirk Enzmann]

Because you didn't show us your data (see FAQ #12) I made up some data to demonstrate one way to do (there are other ways without using recode). As Nick Cox already said: In most situations numeric variables (properly labeled) are better to use than string variables; additionally, your code does not create indicator variables (also called "dummy variables") -- but again, most often that is not necessary because you could use factor variable notation (see help fvvarlist).

Note that I am using frequency weights only because I want to create an example data set. Have a look at the help for all commands you don't fully understand using help command :

. clear
. input hhsize freq

        hhsize       freq
  1.    1  50
  2.    2  70
  3.    3  80
  4.    4  85
  5.    5  82
  6.    6  61
  7.    7  40
  8.    8  18
  9.    9   5
 10.   10   3
 11.   11   2
 12.   12   2
 13.   13   1
 14.   .   12
 15. end

. recode hhsize (11/max = 6 "[11+")   ///
>               ( 9/10  = 5 "[9-10]") ///
>               ( 7/ 8  = 4 "[7-8]")  ///
>               ( 5/ 6  = 3 "[5-6]")  ///
>               ( 3/ 4  = 2 "[3-4]")  ///
>               ( 1/ 2  = 1 "[1-2]"), gen(hhsize_grp)
(12 differences between hhsize and hhsize_grp)

. // check the recoding of hhsize:
. tab2 hhsize hhsize_grp [fw=freq], mi

-> tabulation of hhsize by hhsize_grp  

           |                               RECODE of hhsize
    hhsize |     [1-2]      [3-4]      [5-6]      [7-8]     [9-10]       [11+          . |     Total
-----------+-----------------------------------------------------------------------------+----------
         1 |        50          0          0          0          0          0          0 |        50
         2 |        70          0          0          0          0          0          0 |        70
         3 |         0         80          0          0          0          0          0 |        80
         4 |         0         85          0          0          0          0          0 |        85
         5 |         0          0         82          0          0          0          0 |        82
         6 |         0          0         61          0          0          0          0 |        61
         7 |         0          0          0         40          0          0          0 |        40
         8 |         0          0          0         18          0          0          0 |        18
         9 |         0          0          0          0          5          0          0 |         5
        10 |         0          0          0          0          3          0          0 |         3
        11 |         0          0          0          0          0          2          0 |         2
        12 |         0          0          0          0          0          2          0 |         2
        13 |         0          0          0          0          0          1          0 |         1
         . |         0          0          0          0          0          0         12 |        12
-----------+-----------------------------------------------------------------------------+----------
     Total |       120        165        143         58          8          5         12 |       511

. // show percentages of subgroups excluding missings:
. tab1 hhsize_grp [fw=freq]

-> tabulation of hhsize_grp  

  RECODE of |
     hhsize |      Freq.     Percent        Cum.
------------+-----------------------------------
      [1-2] |        120       24.05       24.05
      [3-4] |        165       33.07       57.11
      [5-6] |        143       28.66       85.77
      [7-8] |         58       11.62       97.39
     [9-10] |          8        1.60       99.00
       [11+ |          5        1.00      100.00
------------+-----------------------------------
      Total |        499      100.00

. // generate indicator variables from hhsize_grp:
. qui tab1 hhsize_grp, gen(hhsize_grp)

. // display variable labels of the indicator variables:
. d hhsize_grp?

Variable      Storage   Display    Value
    name         type    format    label      Variable label
--------------------------------------------------------------------
hhsize_grp1     byte    %8.0g                 hhsize_grp==[1-2]
hhsize_grp2     byte    %8.0g                 hhsize_grp==[3-4]
hhsize_grp3     byte    %8.0g                 hhsize_grp==[5-6]
hhsize_grp4     byte    %8.0g                 hhsize_grp==[7-8]
hhsize_grp5     byte    %8.0g                 hhsize_grp==[9-10]
hhsize_grp6     byte    %8.0g                 hhsize_grp==[11+

. // show proportions of hhsize_grp1 to hhsize_grp6:
. sum hhsize_grp? [fw=freq]

    Variable |        Obs        Mean    Std. dev.       Min        Max
-------------+---------------------------------------------------------
 hhsize_grp1 |        499     .240481    .4278044          0          1
 hhsize_grp2 |        499    .3306613    .4709234          0          1
 hhsize_grp3 |        499    .2865731    .4526141          0          1
 hhsize_grp4 |        499    .1162325    .3208251          0          1
 hhsize_grp5 |        499    .0160321    .1257248          0          1
-------------+---------------------------------------------------------
 hhsize_grp6 |        499      .01002    .0996973          0          1