I was reading an article that reported a Cronbach's alpha of .85 for a two-item scale. The scale was measuring self-perceived competence. Each item was on a 7-point Likert-type scale.
Intuitively, that seemed like quite a high Cronbach's alpha for a scale with only two items.
- Is it plausible to get an alpha that high on a two-item psychological scale based on Likert-type items?
- How through equations or simulations can I better understand what degree of consistency would be required between the items to achieve a reliability coefficient of .85 with only two 7-point items?