Dummy Variable with Mutliple conditions

[Ayorinde Ogunyiola]

I am trying to generate a new variable to take on the values of 1 and 0. Where 1 represents alcohol abuse and 0 no alcohol abuse.

My variables are:

Alc is a variable which has information for alcohol consumption

gender takes the value of 0 and 1 for male and female respectively



for female, alcohol abuse the condition is equal or greater than 2.

for male, alcohol abuse the condition is equal or greater than 3.

Will I be right to specific the command as below:

gen ABI = 0

replace ABI =1 if alc >=3 & gender ==1

replace ABI =1 if alc >=4 & gender ==0

Clyde Schechter

[David Benson]

Ayorinde,

It looks to me like your code will get you what you want if (a) female is gender==1 and (b) you don't have any missing values for alcohol consumption. Also, I always recommend creating / renaming any gender variable to something like male or female (whichever is coded as a 1).

I made some toy data to test things:

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte(id alcohol_consump female)
 1 3 1
 2 6 1
 3 0 1
 4 4 1
 5 5 1
 6 0 1
 7 1 1
 8 6 1
 9 5 1
10 3 1
11 0 0
12 1 0
13 6 0
14 1 0
15 1 0
16 6 0
17 0 0
18 6 0
19 3 0
20 1 0
end

------------------ copy up to and including the previous line ------------------

tabstat alcohol_consump, by(female) stats(n mean median min max)

  female |         N      mean       p50       min       max
---------+--------------------------------------------------
       0 |        10       2.5         1         0         6
       1 |        10       3.3       3.5         0         6
---------+--------------------------------------------------
   Total |        20       2.9         3         0         6
------------------------------------------------------------

gen ABI = 0   // creating alcohol abuse indicator
replace ABI = 1 if alcohol_consump >=3 & female==1  // this works if people can't list fractions of drinks (like 2.5)
replace ABI = 1 if alcohol_consump >=4 & female==0
sort female alc
list, sepby(female) noobs

  +------------------------------+
  | id   alcoho~p   female   ABI |
  |------------------------------|
  | 17          0        0     0 |
  | 11          0        0     0 |
  | 20          1        0     0 |
  | 14          1        0     0 |
  | 15          1        0     0 |
  | 12          1        0     0 |
  | 19          3        0     0 |
  | 16          6        0     1 |
  | 13          6        0     1 |
  | 18          6        0     1 |
  |------------------------------|
  |  6          0        1     0 |
  |  3          0        1     0 |
  |  7          1        1     0 |
  | 10          3        1     1 |
  |  1          3        1     1 |
  |  4          4        1     1 |
  |  5          5        1     1 |
  |  9          5        1     1 |
  |  2          6        1     1 |
  |  8          6        1     1 |
  +------------------------------+

[Nick Cox]

I always recommend creating / renaming any gender variable to something like male or female (whichever is coded as a 1).

Bang on. The auto data does this, naturally: foreign indicates which cars are foreign (and which not) and that allows readable code such as if foreign and if !foreign.

I have heard various tales of this form.

Person listening to presentation: Which way is your gender dummy defined?

Researcher: I am not sure. I will check it out.

But go all the way: call your (0, 1) variables indicators, not dummies. More at https://journals.sagepub.com/doi/abs...36867X19830921