Student's T-Test

So while wikipedia-ing (why yes, I did just do that to the English language) the t-test to check up on a couple of things I found a nice little anecdote about how the Student's t-test got its name. Of course its validity is of question because it is from wikipedia, but I really hope it is the truth. Apparently Guinness, in its quest to make the best beer, would secretly hire top biochemists and statisticians to monitor the beer and overall contribute what they could to the beer making process. Well this at the time was Guinness's big trade secret, so when one of the statistician's came up with a new test, he had to publish it under a pen name. Ergo, it is the Student T Test not the Gossett T Test. Moral of the story, all good statistics and all good beer obviously go hand in hand.

http://en.wikipedia.org/wiki/T_test#History

A very different transformation

I have been corrupted. Now when I read journal articles I find myself investigating interesting statistics I have not seen before. Thanks a lot Mike. So, while reading a journal for my trends class I came across the Principal Component Analysis (yes, I just realized it is spelled principal, not the principle I would have expected. That is actually how it is spelled). For the article they were looking at the gene transcription activation that takes place after the introduction of various hallucinogenic and non-hallucinogenic drugs. They examined 19 different transcripts and were able to take this 19-dimension data and condense it down to a two-dimensional figure. How is this possible??? They used an orthogonal-linear transformation. Huh? What is that? Oh thank you Wikipedia (http://en.wikipedia.org/wiki/Principal_components_analysis). In sum, it looks like it involves attempting to weed out those values that have the least effect on the statistics. In the end they got the this lovely graph showing that there is a correlation between the increase in the expression of c-fos and egr-2 associated with the hallucinogenics (HCs) and not with the non-hallucinogenics (NHCs). I just found this really neat.

Image from: J. Gonzalez-Maeso, N.V. Weisstaub, M. Zhou, P. Chan, L. Ivic and R. Ang et al., Hallucinogens recruit specific cortical 5-HT(2A) receptor-mediated signaling pathways to affect behavior, Neuron 53 (3) (2007), pp. 439–452.

On a completely different note...

Found this story today on my news feed and thought it was too perfect to not be shared on an academic blog.

http://www.msnbc.msn.com/id/23870761/

It is a wonderful story on the pitfalls of plagerism. Enjoy!

So while browsing some of the blogs and thinking about how to wrap my head around all these concepts with the biostat tests and wondering how I am ever going to keep them straight, I got inspired. I saw Claire's lovely flow chart and thusly did an image search for "biostatistics help" and found this, so I thought I would share. It is cited in the picture and comes from the following webpage:

http://www.parc.net.au/AERF_toolbox.htm

Mean, median, mode, again?

Is it just me or does it seem like we have been talking about mean median and mode for FOREVER in this class? I guess it could be because we started before Mardi Gras and just finished up with it (hopefully) today. I feel like ch. 4 could have been much more interesting had S&R actually given some more examples of the usefulness of the more advanced mathematical techniques. Right now it seems like there is an innundation of concepts with little practical focus. As I say this I know I am going to get to the homework and that will be a big help, but moral of the story is I am very much looking forward to moving on and getting down to more of the dirty work and less of the conceptual algebra.

Trial Post

Ah, the first entry in this brand spankin' new biostat blog. I will admit, setting this up was a nice reprieve from the reading but back to it I must go.