Patrick,

IDK if this is still an issue, but it's exactly what I'm trying to do, too. My setup seems like yours, but I teach stats. I have a question bank

*category* with eight questions, but those are grouped by twos. All questions are calculated.

Word Problem 1:

1a. Calculate slope of word problem 1

1b. Calculate Y-intercept of word problem 1

Word Problem 2:

2a. Calculate slope

2b. Calculate intercept

Word Problem 3:

3a. Calculate slope

3b. Calculate intercept

Word Problem 4:

4a. Calculate slope

4b. Calculate intercept

So there are eight (8) questions in the question bank category. I want to randomly select a *pair* of questions from within one word problem. It doesn't work to select, say, question 1a and 2b; those refer to completely different word problems. 1a and 2b share a edataset, so the calculated answers are consistent; 2a and 2b share another dataset, etc. I want to randomly select, a pair, but from within a single "word problem." So if I select 1a, 1b should come with it. Or if I select 3a, then 3b should come with it. There are only 4 choices.

Anyone know how to do this?

**WAITAMINNIT. **I've been thinking about the suggestion:

If I make Word Problem 1 a sub-category of its current category, and only give it 2 questions (1a and 1b), and make Word Problem 2 a sub-category at the same hierarchical level as Word Problem 1, and give it two questions (2a and 2b), and so forth--four sub-categories, each with two questions--then... can I randomly select the sub-category from those four, and then include in the selection all the questions in whichever sub-category gets randomly selected? Is that the suggestion?

**Edit:** Nevermind; I can't figure out how to do that, so if that's the answer, I hope someone can give me some guidance. Al? Anyone?