PSY 370 Lecture Notes - Lecture 34: Standard Error, Psy, Conscientiousness
Document Summary
We often care about the difference between two test scores. *but we have to remember that these tests are measured with error, and the score might have had less error at one point (worked in your favor) and more error at another point (worked against you) or vice versa. Standard error of the difference tells us how different two scores need to be to be considered truly different. Se(diff) = sd * sqrt(2 - reliability1 - reliability2) Multiply the se(diff) by the z-score for your significance level. Differences bigger than this value can be considered "real" differences. Differences smaller than this may just be due to measurement error. An outcome that should be affected by your construct. Distinct, though we expect them to be related. Examples: job performance, graduate school success, successful completion of treatment. *inference that the construct comes before the criterion. Something that exists at the same point in time as your construct.