Statistical Sciences 2035 Study Guide - Quiz Guide: Dependent And Independent Variables, Confounding, Analysis Of Variance
16 Oct 2018
School
Department
Professor
Document Summary
Important: always calculate with estimated marginal means and not total means from descriptives table! Estimated marginal means for balanced & unbalanced designs. Further reasons why 1-way analysis is misleading or different from 2-way analysis concerning the group effect: 1-way analysis is based on weighted means depends on the number of participants that were in a group tests whether the difference between the 3 weighted status means is significant. 2-way analysis: does not look at weighted averages, but rather at differences between the. Estimated marginal means averages that remain after filtering out the overlap ( or contamination ) with the other factor (here: gender) Once the overlap has been corrected for, the means become smaller. Balanced design: total means = estimated marginal means. But a slightly different mean can also occur with balanced design due to calculation incorrectness (cannot do anything against that) However, f-ratio is a little different, because in 2-way analysis there is less residual/error variance.
