Thursday, December 09, 2004

A brick wall

I have been running into a brick wall lately. First, my mood has been quite low due to severe sleep deprivation. Last week I had a 3-day period where I got a total of about 6-7 hrs sleep. My beautiful daughter has never been the best sleeper, and she is so sensitive that any disturbance upsets what little sleep rythm she has. This time, I think it may be molars, but even Oragel and Children's Tylenol gave me no rest. I've been getting a little more sleep per night the past few days, but it hasn't been enough to elevate my mood.

So, now, in my sad state, I am trying to analyze my data using a specific program to get a better idea of how to evaluate the significance of my data. Statistically I have run many tests, so I can't judge the p-values I get on face value. They need to be corrected for the fact that I have run multiple tests, because doing so increases the chance that I get a "significant" result purely from chance and not from anything special about my data. Problem is, correcting for multiple testing in my data is difficult. The classic correction for multiple testing is something called the Bonferroni, but it really only applies when each test is independent, and my tests are correlated, but it is hard to determine exactly how correlated they are. To address the problem I and many others in my field have, someone wrote a snazzy little program to help quantify the relatedness of my variables. Only problem is, I can't get it to work properly. There is a nice little web interface for most people to run, but I have too much data and it gets "stuck." So, there is a script provided that you can download, but it relies on older version of another programming language that is hard to get to work on Windows or MacOSX. You can't imagine all the stuff I've been trying. I've even had Stevie help me for hours and I still keep hitting the brick wall. I even emailed the author of the program but he has ignored my plight.

Monday, December 06, 2004

The start of a blog

I haven't started a blog before because I just don't have time. Right now, I should be working on my thesis; everything else I do is a form of procrastination. It seems a shame to have a new device to pull out when the procrastination bug strikes (which is about all the time). But, the truth is, I will undoubtedly find something to do to postpone the inevitable no matter what, so why not have a blog after all. Plus, I have terrible writer's block. Maybe the blog can be an exercise in just getting something down, even if it is not just quite right.