<$BlogRSDUrl$>

Thursday, June 23, 2005

They say that virtue is its own reward 

But when that surf comes up
I'm gonna get my board.
Got my own ideas about the righteous kick;
You can keep the reward,
I'd just as soon stay sick.

Apologies to the Cramps.

-----------------------

Monday, June 20, 2005

Straight guys always tell me that they have a hard time understanding women 



I dunno, I guess it always seemed pretty simple to me.

The siren of statistical significance 

In epidemiology, and essentially any other statistically-operated decision-making science, one deals primarily with converting a hypothesis into a discrete, testable mathematical statement and subsequently judging any analytical results against a chosen window of significance (the mythic p-value we all love and hate). This p-value is usueful in theory and in practice because it essentially allows one to make a decision by quantifying the amount of uncertainty that one is willing to accept when asserting that the hypothesis under consideration is false (as, unfortunately, there is no such thing as a true hypothesis in this sort of operation; one can only fail to reject the hypothesis).

Say, for instance, that I'm willing to reject the validity of my hypothesis while allowing a 5% chance (the scientific standard de rigueur) of making this kind of error. Now, students always seem to grapple with this sort of thing, myself included, because, when it comes down to it, this completely arbitrary value has no meaning whatsoever. No, seriously, you try to put this into comprehensible English: I am 95% confident that the result I observed or a result more extreme did not occur due to chance given that the null hypothesis is true. Jesus, did you get that? What does that really mean, 95% confident? It seems to be diametrically opposed to my favorite aspect of cognitive psychology, the Belief in the Law of Small Numbers (which, in brief, asserts that an assumption that patterns observed in large samples are represented in small samples is inherent in human mentation--explained at length in the original paper); all things being equal, 95% confidence is no less probable by random chance than, say, 2% confidence.

It twists the brain, which is why it is interesting.

Today I read an amazing section of "Statistical Inference" by Michael Oakes that validated my suspicions about statistical confidence. Yes, I am a total nerd, because this small phrase made me get goosebumps. I read it again and again. It's thrilling and scary all at once, as statistical significance, especially in the human observational sciences, is the fucking cornerstone of decision-making.

"[S]ignificance tests convey almost no information that is of use to researchers... and they lend themselves to abuse and faulty inference."

Nothing too startling, but so eloquently and simply put that it makes one step back from all the theory and calculations to think, "my god, we are still living in the dark ages."

-----------------------

Sunday, June 19, 2005

Natural 

People, including some of my friends from home, tend to display a general mixture of confusion, horror and anger when I tell them that I miss my home city. It's true, and it's a detail that even caught me off guard; in fact, I think that it's a relatively nascent attraction that didn't fully emerge until around the time of my second trip to Europe and the subsequent camping binge that my travelling companions and I began in its wake.

Now that it's summer the urge to return is strong. This is the first year of my life that I haven't been walking through a backyard filled with shoulder-high tomatoes. I'm all about the urban lifestyle, but it certainly has its limitations. I don't even have a houseplant now.


This time last year, first crop

I'm not the only one, though. It seems that we all have wildly varying preferences of place; where is your secret domestic love, hometown or otherwise?


Twenty-seven views of Little Rock, aggregated from a news channel weather camera over the past year

-----------------------

This page is powered by Blogger. Isn't yours?