Tuesday, May 27, 2008

I can't sleep right now because earlier today I made a kinda-important scientific discovery and now the race is on to get it published. Does that make me a geek?

Labels: personal, science

link : 1:48 AM :  1 Comments : Links to this post

Wednesday, February 21, 2007

The High Price of Number Crunching

I do a lot of work in statistics. In fact, almost all of my work involves statistics. I don't collect data in the field. I don't do things in a lab. I sit in front of my computer working with data sets and spreadsheets. Lately I've been doing a lot of online research for a good Mac statistical package. Right now I use a combination of SPSS and Excel.

SPSS seems like the perfect thing for me (it stands for Statistical Package for the Social Sciences). But it's, in a word, bad. It's just bad. I have version 11, which ironically works on Intel while version 13 does not. Still, it follows archaic command line structure under its GUI. Variables can have a name eight letters long or less. No numbers or symbols. What? You can assign labels to variables but that's killing one bird with two stones. And that's gross.

Excel... well Excel is easier to use (which says more about SPSS than Excel) but there are several published articles on how you shouldn't use Excel due to its inaccuracy.

In the statistical package market there are a few alternatives. Aabel and Stata are both up-to-date to work with Intel Macs. The problem, they're around $500 each, and that's the student price (it's almost as expensive as.. a PS3!). If you're just some guy who wants to do statistics, it's $1500. Additional feature packages are $300 or so each. Why're they so expensive? I guess they're trying to sell to people with grants, but you'd think they would want to aim low and get students to like it first, then sell it to them again when they're professors with labs.

Open source to the rescue! Kind of. R is free, but it's strictly command line. And it's hard. The manuals and tutorials aren't very good. Lesson 2: Add and subtract. Lesson 3: Multiple regression! I got lost somewhere in there. Also, trying to find outside advice for using R is hard in itself because it's hard to Google. The internet has a lot of things, and the letter R is one of them.

Labels: anthropology, science, tech:computers

link : 11:14 PM :  0 Comments : Links to this post

Friday, February 2, 2007

How My Dissertation is Going: Part One in a Long Series

The past week has been full of numbers. I've been examining the relationship between femur length and stature, hoping to find a sound way to estimate the latter with the former. Regression is the way to go, but what kind? Most studies have settled on linear regression but there is an inherent bias in its calculation. I'm trying to find alternatives. With the help of my advisor I think I found something that eliminates the bias completely. That gave me some sense of satisfaction.

Next on my list of things to do is to finish my dissertation proposal. It sounds backwards, and it is. I had a really hard time trying to write a proposal when I finally decided on a research topic. The scientific method starts with coming up with a hypothesis, but in reality it often starts with a sudden thought in the shower, or during a sleepless night. Doing statistical work is also not something that has a definite ending where my experiment is done and my question is answered. One result leads to more thinking (and another revelation in the shower) which leads to more testing and so on. I got so caught up in the actual work that I left the proposal behind.

The graduate school requires that I do have a proposal however, as they should. It's a good organizational tool for myself to get my thoughts straight and for my committee whom would like to know what I'm up to. So ironically I'm taking a short break from my work to write a good proposal... for my work. I already have a few drafts done and comments from my advisor. I just have to find them again and read them.

Labels: personal, school, science

link : 5:30 PM :  0 Comments : Links to this post

Saturday, December 23, 2006

Go #54!

My advisor Robert Benfer made Discover Magazine's top 100 science discoveries of 2006! He is number 54 on the list, not a bad showing at all. Bob made the list for finding the oldest known site in the New World that was used for astronomical observations. These observations were important for Andean groups for planning their agricultural seasons. I unfortunately haven't seen all of the cool alignments and stuff in person since I wasn't there the field season when the discoveries were made.

Only two archaeological discoveries beat Bob's Temple of the Fox on the list: a newly-discovered tomb in Egypt and the earliest evidence of writing in the New World. Omega-3 fatty acid-producing pigs was another MU development that made the list, at a surprising (?) #38. Alternative energy in general was the #1 story.

Bob's Article from Discover Magazine
Article from the Columbia Missourian
My glorious advisor's webpage with more information on the site.

Labels: anthropology:archaeology, MU, peru, science

link : 2:25 PM :  0 Comments : Links to this post