Paper Reading #20 : The Aligned Rank Transform

Author

Jacob Wobbrock, Leah Findlater, Darren Gergle, James Higgins

Occupation

Jacob O. Wobbrock is currently an Associate Professor in the Information School and an Adjunct Associate Professor in the Department of Computer Science & Engineering at the University of Washington. 

Leah Findlater is a postdoctoral researcher in The Information School, working with Dr. Jacob Wobbrock. She holds a PhD from the University of British Colombia.

Darren Gergle  is an Associate Professor at Northwestern University and has a PhD from Carnegie Mellon University.

James J. Higgins is currently a Professor at Kansas State University and holds a PhD from the University of Missouri-Columbia.

Location

Published in the CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems.  at NYC

Summary

Hypothesis

Aligned Rank Transform is a useful and easily accessible tool for pre processing nonparametric data so it can be manipulated in a level beyond current non parametric tests.

Method

five step procedure:

compute residuals: for each raw response Y computer residual = Y - cell mean

compute estimated effect for main effects: Ai is the mean response Yi for rows where factor A is at level i. AiBj represents the mean response. Yij represents rows for where factor A is at level i and where factor B is at level j, etc

compute the aligned response Y', and assign mean ranks Y" where Y' = residual + estimated effect

perform full factorial ANOVA on Y"

Results

the author look into three different ART procedures to show it applicability. One case showed how the usage of ART can show interaction effects that might not be shown with Friedman tests. One case showed how it can allow analysts from going through distributional assumptions of ANOVA. The last case showed nonparametric tests of repeatedly measured data. 

Content

The authors presented their ART tool as a applicable means of nonparametric analysis for factorial experiments. There is a indeph discussion on the process and show examples of how it can be applied effectively with real data

Discussion

Frankly this paper was very confusing and i understood exceptionally little of it due to the large amount of jargon that was passed around during its explanation. From what i can see i think the authors did a good job in presenting their cases with examples to show how their product can be applied but i was still very confused in terms of the inner working that they described in the paper.