At 03:46 PM 10/28/2004, Jason McNellis wrote:

>I have a dataset where each case is a family. Children ages are >divided into five variables (0-2 years, 3-5 years, etc) where each >variable shows the number of children in that age group (from 0-2 >children). I would like to set up a variable that indicates the >oldest child's age group for each family. For example if a family >(case) has three children, one each in the age ranges 0-2, 5-10, and >16-17 I would like the variable oldest child to say 16-17.

So you have a list of five variables, which have an order meaningful to you (i.e., age group), and you want the highest variable in the order that has a non-zero value. It feels like there should be a one-statement solution, but I can't think of one. Here's one using a loop. (Code not tested. This is the most direct code. A loop running 'backwards', from high to low, and using BREAK, would be faster, but I can't imagine that the difference would be detectable.)

Assuming your variables are KDS00_02, KDS03_05, KDS06_10, KDS11_15, and KDS16_17, and that they are contiguous in your file,

NUMERIC TOP_KID (F2). VAR LABELS TOP_KID 'Age group of oldest child in family'. VAL LABELS TOP_KID 0 'None ' 1 ' 0- 2' 2 ' 3- 5' 3 ' 6-10' 4 '11-15' 5 '16-17'.

VECTOR GRP_KDS=KDS00_02 TO KDS16_17. COMPUTE TOP_KID = 0. LOOP #AGE_GRP = 1 TO 5. . IF (GRP_KIDS(#AGE_GRP) > 0) TOP_KID = #AGE_GRP. END LOOP.

If the variables aren't contiguous, you can use similar code with DO REPEAT. This code, by the way, treats missing counts as 0.