Hi, I'm working on a problem trying to reduce factors into a single scale. I think generally to do this, you would see if cronbach's alpha was above .6 or .7 for acceptability, but I seem to remember something about 2 items having a different test that is more appropriate, perhaps just regressing one onto the other. If I'm looking at 2 factors trying to reduce them to one representative "index", is there an alternative to cronbach alpha, and if so are there guidelines for what score is "acceptable" to keep as a single item instead of two?

Thanks,
D