Recode if date is not missing

[Andreas Rydell]

I have a problem that help recode doesnt solve.

I have a IHD variable that is coded 0 or 1. In another column I have date of myocardial infarction. If a date is there, an event happened. If it's just a dot, no event happened. I want to recode from 0=1 if there is a date/or not a dot. Is there a simple solution?

[Rich Goldstein]

your question is not completely clear - do you want to change just to "1" (if there is a date) or do you also want to change to 0 if date is missing; here is the first:

replace IHD=1 if missing(date)

change "date" to whatever the actual name of your date variable is

in the future, please show example data using -dataex- and showing the result in CODE blocks (see the FAQ)

[Andreas Rydell]

Thanks for swift reply. It would have been easier with a picture. I try to make it easier:


MIDATE IHD
. 0
. 0
06feb2000 1
. 0
. 0
. 0
04jun1999 0

I want to record every row that has a date =1. Information about previous MI has several origin and not all make the current coding IHD=1, so I have to manually change those.

replace IHD=1 if nonmissing (MIDATE) is in theory what I need.

I recoded everyone to 1 if 0 now. Then used your code mentioned above. Seems to work.

The problem is that I have 3 columns. MI-date, STEMI-date, NSTEMI-date. The two latter doesnt have a 1 if MI-date is 0 fortunately so maybe this solved my problem. I need to double check in the larger dataset (using a smaller training set on 9K individuals now).

[Rich Goldstein]

sorry, but this increases my level of confusion; maybe someone else can help but my suggestion about following the advice in the FAQ remains

[Andreas Rydell]

sorry, but this increases my level of confusion; maybe someone else can help but my suggestion about following the advice in the FAQ remains

Ah i see now why. It didnt look like that when I wrote...

MI-Date	IHD
.
.
04jun1999	1
.
.
.
.
06feb2000	0
.

I want IHD==1 if there is a date and not a "." in column "MI-Date".