PSYC 394 Lecture Notes - Lecture 4: Pooled Variance, Effect Size, Standard Deviation
Document Summary
N = the total number of unique individuals n = the total number of individuals in specified group (e. g. n1 = number of participants in group 1) When a dependent samples (matched/paired samples) t-test is conducted, the individual differ- ences that usually occur between samples are eliminated. Therefore, the power of the test is in- creased: test stat = effect / error test stat = systematic variance / unsystematic variance (lowering the error term in the denominator yields a larger overall value. ) A dependent samples test, however, cannot always be used. For example, if the study requires participants from two markedly different populations (e. g. university students and professors), the same/matched participants may not be appropriate. Similarly, there may be practical reasons such as participant fatigue or attrition that prevents a dependent samples (repeated measures) design. Assumptions of the t-test: both t-test designs (dependent and independent) are parametric tests based on a normal distribution: