jonathaned55
03-19-2012, 06:44 AM
When I measure a construct with just two items (so two questions), can I just combine the results (take the average across the two questions), or do I first need to check if they have a high correlation? With a multi-item scale (e.g. 5 or 10 questions to measure a construct), we would check the alpha for internal reliability and also make sure its a single dimensional construct via a factor analysis). Thanks!