Collapse across multiple variables combinations

[CydneyL Bruce]

Hello,

I have a dataset that looks something like this

Var1 Var2 Var3 Var4
. . . .
1 . . .
1 1 . .
. . . 1
1 . 1 1
1 1 . .
1 . 1 1

I would l like to be able to calculate the total across all of the combinations of the 4 variables so that i end up with:

Var1 Var2 Var3 Var4 Total
. . . . 1
1 . . . 1
1 1 . . 2
. . . 1 1
1 . 1 1 2


I have managed to find a long-winded way to do this where I generate a variable to represent each of the combinations and then collapse this variable similar to below

* Generate combinations *
gen tot1 = .
replace tot1 = 1 if Var1 == . & Var2 == . & Var3 == 1 & Var4 == .

gen tot2 = .
replace tot2 = 1 if Var1 == 1 & Var2 == . & Var3 == 1 & Var4 == .
....

gen tot5 = .
replace tot5 = 1 if Var1 == 1 & Var2 ==. & Var3 == 1 & Var4 == 1

* Collapse for totals *
collapse (sum) tot1 - tot5

gen N = 1
reshape long tot, i(N) j(Variable)
rename tot Total
drop N

label define var 1 "No Variables" 2 "Var1" 3 "Var1 & Var2" 4 "Var4" 5 "Var1 & Var3 & Var4"
label val Variable var

But, as there are many more than 4 variables and a lot more combinations this method is very long-winded and it doesn't quite give me the form I would like.

Any help is greatly appreciated!

Best wishes,
Cydney

[Nick Cox]

help contract 

[Nick Cox]

I wrote the original of contract and so can say something on why it was thought up,

In principle you don't need it as

bysort Var1 Var2 Var3 Var4 : gen freq = _N 

collapse freq, by(Var1 Var2 Var3 Var4)

gets you to the same place. But that is a trick that is easy to miss. The original used an invented name to follow StataCorp advice against using standard English words (advice I have ignored elsewhere) but once it was adopted as part of official Stata (in Stata 6) it seemed clear that as an opposite of expand it deserved the name contract.

[CydneyL Bruce]

Brilliant thank you very much for the help Nick!