When calculating scale means, you almost always have a subset of participants who missed one or more items from the scale, but answered the majority of the questions. And of course, not everyone is missing the same items, or the same # of items. My grad student a few years ago taught me this syntax, and it makes this situation easy to handle.
Come up with a decision rule for how many missing items are acceptable. Depending on the study/scale, I’ve seen rules of allowing 20% missing or allowing <50% missing. In the example below, we used the <50% rule, so for a 10 item scale, we required data from 6 or more items. With this syntax, SPSS will calculate the means for anyone who has 6 or more of the indicated variables not-missing:
COMPUTE ESTEEM_T=mean.6(ESTEEM1, ESTEEM2, ESTEEM3_R, ESTEEM4, ESTEEM5, ESTEEM6_R, ESTEEM7, ESTEEM8_R, ESTEEM9, ESTEEM10_R).
Done. I’m too embarrassed to share how we used to calculate it before I learned this syntax.
“The post SPSS tip #1: Calculating the mean when some items are missing first appeared on Eva Lefkowitz’s blog on January 24, 2014.”