PSY 370 Lecture Notes - Lecture 39: Convergent Validity, Discriminant Validity, Construct Validity
Document Summary
Need to make sure that our results are about the construct, not the method. If we"re measuring a construct, we should get (essentially) the same result no matter what method we use. Measures of the same thing using different methods should be highly correlated. Even if the techniques to get those measures are different. If we get the same results for every construct when we use a particular method, we"re probably measuring the method instead. Measures of different things using the same method should be only slightly correlated. Measure a whole bunch of traits in a whole bunch of methods. And then look for patterns in the data. To assess convergent validity, a good way to look at this is the correlation between: Method 1, trait a with method 2, trait a. *the strongest correlations should be between the same trait and different measurements (methods) *we would want this correlation to be high.