Wednesday, May 31, 2006

The World Revolves Around p Value

You understand the intricacies of the subject...become overjoyed with new ideas and questions budding into your head...if you're lucky enough, your boss allows you to go ahead with them...you plan every single detail about how you would proceed with the experiment...you keep your enthusiasm intact through repeated seesions of resurrecting the protocol...overcome every single difficulty at the experimental stage...and then that stage comes when finally you sit to analyze the data you'd been gathering so far...But, you realize that all this while, gods were planning against you..the variables under scrutiny are NOT SIGNIFICANTLY DIFFERENT from each other!!!

It's very, very easy to ridicule it by saying, "but that means you were already expecting some kind of result without even analyzing the data, that's not done in research!!!"

Oh yeah! Easy to say, dude...Show me a single scientist who goes into an experiment without some expectation. Even if the experiment is totally new, howmuchever we like to avoid it, we've some pet idea about the kind of data or results we would like to obtain.

And then there's dispair and frustration. Don't get me wrong. Scientists aren't weak. But, each experiment takes a long time to get over with. And it's like your own baby, man! Everytime you come up with a good batch of results (read positive result), you feel proud...am sure like a proud dad/mom feels. And everytimg things turn sour, is the feeling parents face over the miscarriage.

One thing is beyond my understanding though. Ok well, there was some loophole or some problem in the protocol, you take it in your stride because after all, it was your fault. But what when the result is in the form of "a non-signficant difference" or "a uniform distribution" or a "no cluster" situation. Of course, that's a result, too. Just presence of something is not important. Absence of something, or results contrary to expectations are important, too.

In short, and in very simplistic tones, just the presence of some biological phenomenon is not important. It's equally important to understand that the absence of it is an important thing to study, too. Who knows that might be a possibility to a ground-breaking discovery! Fleming found out Penicillin in the same way, didn't he?

Obviously, I'm living in an utopian world. p value > 0.05 and you're doomed!

No comments: