Here are the responses I received to my request for examples/datasets for teaching about the two-way ANOVA. Thanks for everyone's contributions! --Lise
~~~
Some data sets are available in my lecture notes, which you can find at http://www.uvm.edu/~dhowell/gradstat/psych341/schedule.html
The data have generally been created to exactly mimic data in published studies. (Well, "exactly" is a bit overstated--they have the same means and standard deviations.) you are welcome to anything there. Click on any of the links in weeks 3 and 4.
If you have trouble downloading the data, I can send it to you. IE6 will download the files, but Netscape 7 thinks it should open them, and doesn't know how. I forget what Opera does.
Good luck,
Dave Howell
~~~
For data, try DASL
http://lib.stat.cmu.edu/DASL/allmethods.html
Click ANOVA.
Jill Binker
~~~
Further to your EdStat post of October 31, some technical material
about computing sums of squares in two-way unbalanced ANOVA is at
http://www.matstat.com/ss/
Don Macnaughton
~~~
It occurs to me that a data set I used to use for examinations might be
a useful tool for introducing students to 2-way ANOVA. It's a 3x5
subset of a more elaborate design, and a graph of the 15 means shows
very strongly the idea of interaction between the two factors. If the
5-level factor (which are 5 different tasks, hence repeated measures)
is divided into (2 + 3) levels, and the two pieces analyzed separately,
the interaction virtually disappears, which again provides a stimulus
for some useful pedagogy. Originally part of a Ph.D. thesis in the
former Dept. of Applied Psychology, OISE. Haven't time to go dig it up
now; if I forget and don't get back to you in the next week, please
send me a reminder. ...
Ciao!
Donald F. Burrill
~~~
this is a case where i think it is very helpful to have students actually do some hand calculations ... and, if you make say a simple 2 by 3 design ... with only maybe 3 Ss in each cell AND, make sure the cell means and row and column means (and grand mean) work out nice and whole number neat ... it is easy (sans interaction term raw calculation) to have them find the SS components and do the actual 2 way anova ...
then, you could have them use some software ... enter the data table the right way (depending on how your software lays data out) and see the output and compare to their hand calculations
what i usually do is to give them this as an inclass hands on exercise handout ... we have the table and data in the table ... find the cell, row, column and grand means ... make a graph ... speculate on what is happening ... THEN do the SS calculations and the summary table ... THEN see how software would handle it
another thing that is useful is to use a cell mean table that ... if you plot with one factor on the baseline and the other in the graph ... and REVERSE the orderings .... that you will see a crossover pattern one way and ... NOT a crossover pattern the other way ... which shows that (as i learned the hard way) the pattern we see with a crossover where we "usually" say there is an interaction ... may not be evident when independent variables are reversed for plotting ...
terms like ordinal INTERACTION or DISORDINAL interaction are just descriptions of the graphs ... and NOT the interactions themselves .... long ago, don burrill "straightened" me out on this score
Dennis Roberts
Lise DeShea, Ph.D.
Assistant Professor
Educational and Counseling Psychology Department
University of Kentucky
237 Dickey Hall
Lexington KY 40506
Email: [EMAIL PROTECTED]
Phone: (859) 257-9884
"... for only by varied iteration can alien conceptions be forced on reluctant minds."
