About
Abram Hindle's Blog, The Personal Blog Of Abram Hindle
Abram Hindle
abezblog@abez.ca
Subscribe
Subscribe to a syndicated feed of my weblog,
brought to you by the wonders of RSS.
Flavours
There's more than one way to view this weblog; try these flavours on
for size.
Links
These are a few of my favourite links.
Thu, 06 Sep 2007
Subject: Corrupt countries were more likely to support the OOXML document format? With a correlation coeffecient of -0.31?
http://www.effi.org/blog/kai-2...
For some reason they use a Wilcoxian test and fisher t test on the data. The data has 2 random variables, corruption (continuous) and vote (yes/no)
They said they should with good p-values that there correlation between voting for approval and corruption. Well using both Pearson and Point-Biserial Correlation coeffecient I found a coeffecient of -0.3129 with a p-value of 0.009373 this means there was a weak or medium stength linear correlation between corruption (more negative) and approval votes.
What's interesting is that EFFI keeps applying the wrong tests when they
want to show to correlation. I also did the rank based correlations and
they weren't as good as the linear cases -0.29 for Spearman-Rho and 0.24
for Kendall-Tau rank based correlations. So there is a correlation but
My data and a log of what I did is at: http://churchturing.org/w/ooxm...
abram