Stats Lecture 10 Factorial Design PDF

Preview

Summary

Discovering Stats Lecture...

Description

Lecture 10: Factorial Designs Designs are often just named, but the name of the tests used can be broken down. Two-Way designs mean that there are 2 IVs (three-way meaning 3 IVs etc), Independent means that there were different participants in all conditions and several independent variables is known as a factorial design. The benefit of factorial designs is that we can look at how variables interact with one another. Interactions show how the effects of one predictor might be moderated by another. Essentially we can see how the various levels of one predictor interact with the levels of another predictor.

Factorial ANOVAs look at the effect of more than one IV, with each IV having two or more conditions/levels. There are main effects for each IV but also interactions which show the moderation effects of other experimentally manipulated factors. ANOVAs answer the question of whether using the model is significantly better than just using the grand mean. F-ratio: Variance explained by the model / unexplained variance. Significant: Model is better than just using the mean on its own. Follow up tests answer the question of how much particular groups or chunks differ from one another. T-test: How far apart the means are given the SD. Significant: Two particular groups/means are likely to be from different populations.

These tests do not answer the same question, so the answers you get can be contradictory. Real data can be messy so it is possible to have a significant ANOVA but not a significant follow-up....