PSY 250 Lecture Notes - Lecture 4: Litter Box, Naturalistic Observation, Internal Validity
Document Summary
You can"t co-vary if you don"t vary, so you need a diverse group of people not a very narrow one. Using correlation implies that you actually calculated the statistic. If you didn"t, use related to or associated with. Looks for correlations among variables, can also involve more complicated statistical techniques (like regression). Does not manipulate/ control for / intervene in any of those variables. Naturalistic observation (watching what happens), surveys (depending on how they"re written, can go either way), archival data. Other variables might affect the correlation, no absolute control in correlation, correlation is a probability, correlations aren"t always perfect (no absolute). Correlation is necessary (but not sufficient) to causality. Called the third variable problem, in a controlled lab study, you can make sure nothing varies except a and b. In a correlational study, you"re measuring variables as they occur naturally you can"t control these extraneous factors. The main criticism of correlation designs is that they lack internal validity.
