Thursday, May 22, 2008

Bikini Statistics

(What they hide is ofter more interesting than what they expose) Growing up an trained in the hard sciences, statistics and scientific theory were drilled into my head. As a result, my head spins when I look at social science studies. I looked at a study of medcial malpractice coverage and my head almost exploded. In looking at a research study, we always first look at the abstract. There at the bottom will be a concise sentance that states their findings. I then look at the actual conclusion of the paper. You might be amazed to see how many times these two dont agree. The next step is to look at the methodology. If I want to compare two medications, I find a large cohort (group) of patients that are as close alike as possible. I then make sure that I isolate the group from other variables that might affect the results. I then find a way to quantify results that are empiric and this easily measured. When all this is done, we statistically look to see if there is a significant difference between the groups. When all that is done, all I can really say is whether of not there was a difference between the groups.

In the social papers. There is no isolation of variables, there are only trends. Conclusions are made from inferences that are not supported by the data. Here is a good example. "In a study of childrens accidents involving tricycles, the rate has decreased over the last few years. We therefore conclude that children are better at riding tricycles". What is not said is that few tricycles are made and being ridden and that they are now so small that when you fall you dont get hurt. Social science is based on Bikini statistics. Hard science tries to use "thong statistics". it tries to reveal all, and most of the time it reveals stuff that isn't necessary that good looking!


SeaSpray said...

Throckmorton, your work in research sounds extremely interesting. So you might be on the cutting edge of some discovery that significantly helps others? How exciting is THAT?!

My head would explode if I had to do statistics because math is not my strong suit. Although, older son says I am like Rainman because I remember numbers, even from years ago. I just see them. Give me words, contracts or a foreign language and I'm there.

The process for your research studies sounds intriguing. i love the idea of questions leading to questions to get answers. And working with people in studies...also interesting. The hunt I guess and perhaps the thrill when you've got concrete findings.

Regarding the social papers, I don't understand how they can honestly justify their conclusions if they know certain conditions (omitting pertinent info)are actually creating an illusion that something is a trend in a new direction.

I know this is apples and oranges but in your QNS post last fall you mentioned how results for studies can be skewed in one direction or another by how they collect their data. So who is supposed to benefit? The researchers because they get more funding or they make a name for themselves...or companies promoting a product? I really hate when studies are announced on the news, we start to abode by it then some other study contradicts it.

Your analogy of bikini(ahh bygone days:) vs thong statistics is a hoot and clever! LOL! :)

However, I don't understand your final conclusion. Are you suggesting that in the end (no pun intENDed-ok that was)the stats reveal undesired info after all the research and so it is back to the drawing board?

SeaSpray said...
This comment has been removed by the author.