Contrast method

[fauman]

Hi - I've been using the T Test and ANOVA methods in statsample quite a bit.
I wanted to try the "contrast" method today, but first I discovered that my copy doesn't have this method - Is it not included in the 1.0.1 distribution?

By including the contrast.rb code directly into my Ruby script, I got it to work, after I added alias :msw :ms_den to Statsample::Anova::OneWayWithVectors.

This works to calculate the psi and the se. But when I go to calculate a the T statistic (c.t) I get this error:

in t_object': undefined methodnew' for Statsample::Test::T:Module (NoMethodError)

Is there a version past 1.0.1 that has contrast implemented?
Thank you

[clbustos]

I didn't release a new release. Let me work on it and I try to fix your
problem.
Can you download the last git version of statsample and test it? Will be
very helpful

Greetings

On Thu, May 19, 2011 at 3:35 PM, fauman <
[email protected]>wrote:

Hi - I've been using the T Test and ANOVA methods in statsample quite a
bit.
I wanted to try the "contrast" method today, but first I discovered that my
copy doesn't have this method - Is it not included in the 1.0.1
distribution?

By including the contrast.rb code directly into my Ruby script, I got it to
work, after I added alias :msw :ms_den to
Statsample::Anova::OneWayWithVectors.

This works to calculate the psi and the se. But when I go to calculate a
the T statistic (c.t) I get this error:

in t_object': undefined methodnew' for Statsample::Test::T:Module
(NoMethodError)

Is there a version past 1.0.1 that has contrast implemented?
Thank you

Reply to this email directly or view it on GitHub:
#4

Claudio Bustos
Psiclogo
[email protected]

[fauman]

Be glad to, but I can't find anything past the January 28th 1.0.1 release. Where do I find the latest release?

[clbustos]

Install git (sudo apt-get install git on debian/ubuntu) and pull the latest git version using
$git clone git://github.com/clbustos/statsample.git

You should point to correct directory using $LOAD_PATH with the directory of latest statsample version.

[fauman]

I'm happy to report that I've successfully installed the latest git version and the contrast method is working perfectly. I've compared the p-values for analyses of sets of 4 vectors of size 4 against values computed in R and the numbers I've checked so far have all agreed.

[clbustos]

Good! I tested the contrast routines on the suite, but is better to have multiple users approving the methods.