sum values accross rows only if none missing

[Rodolfo Hernandez]

Hello, I am trying to generate a new variable equal to the sum for a number of observations only if none of the observations are missing (see data below) and equal to missing if any observation is missing. I cannot use a simple STATA command such as sum or total as these take missings as zero. Any help on this is very much appreciated.

I need to do this after using mi impute, so should hopefully something that can be use with mi passive.

Thank you very much in advance.
Rodolfo.

StudyNo	Var1	NewVariable
1	0	.
1	1	.
1	1	.
1	2	.
1	1	.
1	.	.
2	0	.
2	2	.
2	2	.
2	.	.
3	0	16
3	1	16
3	2	16
3	1	16
3	3	16
3	4	16
3	5	16
4	0	.
4	0.07	.
4	0.01	.
4	0.01	.
4	0.03	.
4	0.02	.
4	0.05	.
4	.	.

[Nick Cox]

.egen nmissing = total(missing(var1)), by(studyno)

.egen wanted = total(var1) if nmissing == 0, by(studeyno)

[Rodolfo Hernandez]

Thank you Nick for your answer. The code does work with the simplified data example. However, it does not seem to work with the mi dataset. Below is an extract of the dataset for the first individual.
_mi_m = 0 denotes the original data with missing values for var1 (the variable I want to sum up). _mi_m >0 denote the imputed data datasets and these have complete data. However, nmissing equals 1 in all but one observation. I was expecting a zero for all observation for mi_m >0. Am I missing something?

I used
. mi passive: egen nmissing = total(missing(var1)), by(StudyNo )

and using
. mi passive: egen nmissing = total(missing(var1)), by( _mi_m StudyNo ) // would give the same results.

Thank you very much again for your help.
Rodolfo.

StudyNo	_mi_id	_mi_miss	_mi_m	temp	var1	nmissing
1	371	0	0	1	0	1
1	755	0	0	2	0.0304656	1
1	2548	0	0	3	0.0150765	1
1	2851	0	0	4	0.0172303	1
1	3318	0	0	5	0.0161954	1
1	3279	0	0	6	0.0188946	1
1	3381	0	0	7	0.0269923	1
1	2573	0	0	8	0.0188946	1
1	3751	0	0	9	0.0620822	1
1	2377	0	0	10	0.0080977	1
1	3177	0	0	11	0.0215938	1
1	5540	0	0	12	0.0512853	1
1	6082	0	0	13	0.0377892	1
1	6824	1	0	14	.	1
1	371	.	1	1	0	1
1	755	.	1	2	0.0304656	1
1	2548	.	1	3	0.0150765	1
1	2851	.	1	4	0.0172303	1
1	3318	.	1	5	0.0161954	1
1	3279	.	1	6	0.0188946	1
1	3381	.	1	7	0.0269923	1
1	2573	.	1	8	0.0188946	1
1	3751	.	1	9	0.0620822	1
1	2377	.	1	10	0.0080977	1
1	3177	.	1	11	0.0215938	1
1	5540	.	1	12	0.0512853	1
1	6082	.	1	13	0.0377892	1
1	6824	.	1	14	0.6516068	0
1	371	.	2	1	0	1
1	755	.	2	2	0.0304656	1
1	2548	.	2	3	0.0150765	1
1	2851	.	2	4	0.0172303	1
1	3318	.	2	5	0.0161954	1
1	3279	.	2	6	0.0188946	1
1	3381	.	2	7	0.0269923	1
1	2573	.	2	8	0.0188946	1
1	3751	.	2	9	0.0620822	1
1	2377	.	2	10	0.0080977	1
1	3177	.	2	11	0.0215938	1
1	5540	.	2	12	0.0512853	1
1	6082	.	2	13	0.0377892	1
1	6824	.	2	14	0.6516068	0
1	371	.	3	1	0	1
1	755	.	3	2	0.0304656	1
1	2548	.	3	3	0.0150765	1
1	2851	.	3	4	0.0172303	1
1	3318	.	3	5	0.0161954	1
1	3279	.	3	6	0.0188946	1
1	3381	.	3	7	0.0269923	1
1	2573	.	3	8	0.0188946	1
1	3751	.	3	9	0.0620822	1
1	2377	.	3	10	0.0080977	1
1	3177	.	3	11	0.0215938	1
1	5540	.	3	12	0.0512853	1
1	6082	.	3	13	0.0377892	1
1	6824	.	3	14	0.5359771	0

[Nick Cox]

Surprisingly or not, I don't ever work with data imputation in this sense and so will leave #3 open for others. If you don't get a good answer within 24 hours, I suggest that you

1. start a new thread mentioning imputation in the title to attract attention from people who use mi

2. there cross-reference this thread

3. post a cross-reference in this thread to the new thread.

Alternatively, this question could be one for StataCorp technical services.

[Rodolfo Hernandez]

Just adding a link to the related thread mentioninikg multiple imputation. Thank you. Rodolfo.