My problem is basic, but it's also a bit of a time-crunch. I’m having to re-familiarize with SPSS quickly as I’ve been asked to revisit the dataset from a study I did 2 years ago in grad school. I haven’t used the program since then and have forgotten much.

I wrote a survey that included a 13 item scale (Likert-type) that I designed for the study. Another researcher wants me to break that scale up into two subscales of 3 and 10 items, and look for differences between male and female respondents (gender is coded in a separate variable), as he suspects that a subtle difference in these 3 questions may reflect a gender difference.

Should this be two independent t-tests with each batch of items in the “testing variable” window and “gender” in the "grouping variable" window? Or should this be a 1-way ANOVA? Also, should I be putting these items in individually, or should I compute 2 new variables that take an average of each batch of items?

Any help is greatly appreciated.