Day #80: Experimental Design
I have funding for an experiment. Well let me rephrase, I’ve had funding for an experiment. It’s new, it’s exciting, it’s everything I wanted it to be… but there is a catch. My PI and I don’t see eye to eye regarding the experimental protocol. It’s not a matter of a fledgling PhD student thinking he knows better, he is well know for losing sight of the big picture in favor of collecting as much data as he possibly can.
In my research there are a few lines of thought regarding experimental setup. In recent years the effort has been to create a simple, well thought out experiment that answers a specific question. Doing a very “clean” simple experiment eliminates confounding factors, things that we don’t want to have to control for, so we can remove the effects of them (or minimize them if we cannot eliminate).
To that end my colleagues and I designed a very simple experiment to test two very distinct conditions. It was well thought out, easy to set up, easy on the subjects, and most importantly would make the data analysis simpler. We quite literally spent months designing this experiment and when we got funding two months ago we thought we were ready.
Our PI had a different idea. Now don’t get me wrong, I love our PI, he is brilliant, kind, understanding, and when he explains something his depth of knowledge on the topic is just mind boggling. However, we are all human and sometimes we fall into certain traps, one of them being that more data is better. However, most of us in the lab feel that this isn’t always the case and I like to put it succinctly by saying, Mo’ data, Mo’ problems.
So after the first meeting with him where he completely restructured our experiment, we came up with reasons why our first experiment was better. Yesterday we had another meeting and he agreed (in fact he didn’t recall some of the changes he had proposed, which was funny).
Basically this isn’t a post to bash my PI, because he is a great guy (if not super busy). Instead it is to say this, sometimes we can lose sight of the main question we are trying to answer in an experiment. Now in a perfect world we could answer many questions with a single experiment and in some cases, you can! Unfortunately for us, this wasn’t the case and after explaining our reasoning he came to agree with us. Now we have a slightly different experiment from our first proposal, but it is still a much better experiment than what our PI first came up with.
Until next time, don’t stop learning!