PSYCH339 Lecture Notes - Lecture 5: Extraversion And Introversion, Confidence Interval, Job Performance
Document Summary
Is the light in the room affecting you: before we worry about validity, we have to make sure we"re consistent, observed score = true score + some error, example of extraversion. If r is 0. 8, that means that 80% of observed variance is attributable to true variance. If r = 0, then the variance comes completely from error. If r =1, then there"s no error we measured everything perfectly. Since we can"t compute the equation, we have to estimate reliability. You can do this by: test retest = administer the same test a few different times and see how close the scores correspond. How well different items on the test correlate with each other: ex. Split half consistency(prevents from effects from test fatigue), alpha. Correlation is the degree to which the ratings agree: often not practical. Or difficult to find people willing to evaluate others: ex. Figure skating is evaluated like this: parallel forms = different method, same construct.