PSYC 394 Lecture Notes - Lecture 5: Analysis Of Variance, Observational Error, Effect Size
Document Summary
Twelve participants were exposed to a tarantula, then the same twelve participants were exposed to a toupee. Anxiety levels were measured directly after the encounter, and the mean, standard deviation and variance were calculated for each group. For the difference scores calculated by subtracting each participant"s anxiety measure for condition 1 from condition 2 see the powerpoint slides from class. These difference scores can be used because the data is related; there should be no random error between paired scores because they are generated from one participant. t. = (d" - 0) / (sd n) df. For the dependent test, a significant difference was found, whereas the independent test with the same data did not provide significant results. This is due to the lower systematic error in the de- pendent sample (lower denominator in t-test = larger overall result). Both t-tests and anova tests compare means, but t-tests have more limitations.
