ANOVAs Excel 2000

[John T]

installed the data pack, and it says that I can do ANOVAs. Each time I
setup the data, I can not get any data out of the analysisI such as Sum of
Squares, df, F ratio etc. I get error messages. Any help for me?

 

[Mike Middleton]

John T -

There may be some help, especially if you tell us which of the three ANOVA
tools you want to use, how you have the data organized, and the specific
error message. The setup is different for each of the three ANOVA tests.
And, the Microsoft web site has some Knowledge Base articles about Excel
2000 ANOVA, but maybe you've already looked at those.

- Mike
www.mikemiddleton.com

 

[John T]

Specifically, I want to do repeated ANOVAS and two-way ANOVAS. The data pack
says I can do both.

On the repeated ANOVA, I have 10 subjects, with three tiers (K): base score,
immediate score after experience, and six months later. I know that the
results will be faulty, if I do a series of repeated t tests, and compare,
because I need the "error" interaction.

The other is a simple 2X3 two-way ANOVA.

In either case, my calculations seem to be off, and I need to verify them.
(twoo hours worth, ugh!)

 

[Michael R Middleton]

John T -

For the repeated measures ANOVA, you could use "ANOVA: Two-Factor Without
Replication" of Excel's Analysis ToolPak.

For example, you could arrange your data with subject names in A2:A11;
labels "base," "after," and "later" in B11; and scores in B211. Choose
Tools | Data Analysis | Anova: Two-Factor Without Replication, enter A111
for Input Range, check the Labels box, specify Alpha, choose a location for
the output, and click OK. On the output, ignore the Rows variation, and
focus on the Columns variation.

You didn't describe your data for the "simple 2X3 two-way ANOVA." To use
Excel's "ANOVA: Two-Factor With Replication," you must have the same number
of observations for each of the six combinations of factors. Excel's tool
cannot handle unequal sample sizes.

- Mike
www.mikemiddleton.com

 

[John T]

Thanks Mike:

I will print it and try it tomorrow. I am too tired from doing manual
calculations to fool around further.

 

[John T]

Mike:

Your explanation hepled greatly, but I did run into a major problem.
I explained it in another post lost in cyberspace. Here it is in a nutshell:

The SS of each, between, subject, error and total were consiferably
different from the ones I calculated 3X for verification. (The column sums of
mean, square and total were correct, but that is where it ended.) Naturally,
the F ratio was off, showing a significant difference when there was none.

I need to use that function because my version of SPSS does not have it. Am
I asking too much of Excel?

 

[Mike Middleton]

John T -

I do not have your data, and I do not know what calculations you did for
your verification.

An example of repeated measures ANOVA is available at

http://psynts.dur.ac.uk/notes/Year2/redesdat/RepeatedmeasuresANOVA.htm

When I use Excel's "ANOVA: Two-Factor Without Replication" with the data of
that example, the Excel result for Columns variation is F = 10, identical to
SPSS's General Linear Model (the table of "Tests of Within-Subjects Effects"
of the web site example).

- Mike