group command in stata

[Muhammad Ramzan]

Can anyone explain what this command is doing.

egen momid = group(v001 v002 v003)

where

v001 cluster number
v002 household number
v003 respondent's line number


Thanks

[Hemanshu Kumar]

This will give a unique number to all observations that have the same combination of cluster, household, and respondent line numbers. It can be used as an identifier for a group defined by common values of those three underlying variables.

The numbers are positive integers starting with 1.

[Nick Cox]

group() is here a function of the egen command, and not itself a command.

You don't give a data example, but here is a worked example, showing results with the groups command from the Stata Journal. That is quite separate and just a convenience to show what is going on, namely sorting by the variable(s) mentioned, assigning integers 1 up to the distinct groups of observations found, and optionally supplying value labels. See the help for other options. such as for including missing values.

. sysuse auto, clear
(1978 automobile data)

. egen group1 = group(foreign rep78)
(5 missing values generated)

. egen group2 = group(foreign rep78), label
(5 missing values generated)

. groups foreign rep78 group1 group2

  +----------------------------------------------------------+
  |  foreign   rep78   group1       group2   Freq.   Percent |
  |----------------------------------------------------------|
  | Domestic       1        1   Domestic 1       2      2.90 |
  | Domestic       2        2   Domestic 2       8     11.59 |
  | Domestic       3        3   Domestic 3      27     39.13 |
  | Domestic       4        4   Domestic 4       9     13.04 |
  | Domestic       5        5   Domestic 5       2      2.90 |
  |----------------------------------------------------------|
  |  Foreign       3        6    Foreign 3       3      4.35 |
  |  Foreign       4        7    Foreign 4       9     13.04 |
  |  Foreign       5        8    Foreign 5       9     13.04 |
  +----------------------------------------------------------+

.