PSYC 200W Chapter Notes - Chapter 12: Type I And Type Ii Errors, Multivariate Analysis Of Variance, Statistical Inference
16 Jan 2019
School
Department
Course
Professor
Document Summary
Analyzing complex experimental designs: the problem: conducting multiple tests inflates type 1 error, overall chance of making a type 1 error across all of our tests is higher, type 1 error = 1 - (1- alpha)^c. C = # tests/comparisons performed: bonferroni adjustment: divide desired alpha level by # tests they plan to conduct. Sum of squares: total amt of variability in a set of data. Sum of sum of squares for each of the experimental groups. Mean square between groups (estimate of between groups variance) If f = 1, iv has no effect. Test whether the b/w group variance is larger than we would expect based on the amount of w/i group (error) variance in the data. Specify 3 things: set alpha level, calculate df for the effect we are testing, calculate df for the w/i group variance. Conclude: at least one of the condition means differs from the others, thus the iv had an effect.
