PS296 Chapter Notes - Chapter 16: F-Distribution
Document Summary
One of the simplest measures of the magnitude of effect is (eta squared) However eta squared tends to overestime the population value, though is useful as a simple, first approximation. The differences we care about are the ones among scores that can be attributed to our treatment/group effects and are measured by ss group. Affords a less biased estimate where k stands for the number of groups. Usually produces a lower value than eta squared d-family measures: Best to restrict ourselves to two-group comparisons and to speak about the difference between specific groups or sets of groups, rather than to make a global statement about differences among all groups simultaneously. Same calculation as before:: dhat = xbar1 xbar2 / s s: Sd of any group deemed appropriate by researcher. The f distribution depends on: number of groups. Df for groups: number of observations within groups. Df for the error term: the magnitude of f.
