I suggest you replace missing values with a number like -5, and then use

income = max (sd13a, sd13b) through transformation option in toolbar.

Dear Carina Hartz ,

Alternatively you could create a variable indicating whether sd13a has been answered with a zero for no and a 1 for yes. You can then easily compute the required value with a formula.

// Stata code

replace sd13a = sd13b if sd13a == . & sd13b != .

Note that you could simplify that as follows:

replace sd13a = sd13b if sd13a == .

Or you could use the missing() function, like this:

replace sd13a = sd13b if missing(sd13a)

There is no need to check for sd13b being non-missing before assigning 13a the value of 13b. If 13a is missing and 13b is also missing, 13a will end up being missing, which it was to start with. So no harm done.

* Equivalent SPSS code, both long & short versions.

IF MISSING(sd13a) AND NOT MISSING(sd13b) sd13a = sd13b.

IF MISSING(sd13a) sd13a = sd13b.

DESCRIPTIVES sd13a.

The DESCRIPTIVES command is not strictly necessary, but it will cause the data file to be read so the the transformations happen. Without it (or some other command that causes the data to be read), you'll see no change in your data file, and there will be a message in the bottom right corner saying "Transformations Pending".

One final piece of advice is to consider storing the income in a new variable so as to not overwrite sd13a. E.g.,

// Stata code

generate sd13inc = sd13a

replace sd13inc = sd13b if missing(sd13a)

summarize sd13a sd13b sd13inc

* SPSS code.

COMPUTE sd13inc = sd13a.

IF MISSING(sd13a) sd13inc = sd13b.

DESCRIPTIVES sd13a sd13b sd13inc.

HTH.