The Power of Statistics



“Aaaaaah…..” “Oh, no!” (returning off the aspirator) “I’m sorry.” “What do we do?” (squeezing the aspirator tube) “Can we break the vacuum?” “Let’s pull together.” (pulling on the aspirator tube in two opposite directions unsuccessfully…)
On the third day of Mod 1, my partner and I aspirated our +IPTG induced pellet. I was so convinced that we should aspirate the left over supernatant with an aspirator because it worked well for the –IPTG control pellet. Why did an n=1 experiment convince me that aspirating would be safe? 

Consider a hypothetical situation where before any experimentation, I think there’s a 1/2 probability that the aspirator cannot aspirate a pellet, and 1/2 probability that aspirator can aspirate a pellet and does it half of the time. One successful aspiration should only increase the probability that the aspirator cannot aspirate a pellet to Pconditional = 0.5/(0.5+0.25) = 2/3. So a successful aspiration should increase my confidence in aspiration from 1/2 to 2/3. In reality, my confidence in aspiration went up from 50% to 98% after one success.
Do our brains always process statistical information logically? Can our brains directly tell whether we can make a conclusion based on difference in data? I think the answer is no. And I learned it the hard way as my partner and I lost our +IPTG pellet on the third day of class.
On the flip side, losing the pellet has increased my appreciation for statistics, that statistically analysis based on mathematically formulas are necessary to make sound conclusions. Speaking of statistical analysis, most of the data we got for Mod 1 were not statistically significant (including all the data form PPIase assay). What can we conclude from statistically insignificant data? How do we draft a coherent story in the data summary based on data that’s mostly insignificant?
Luckily, one of our ligands had significant effect in the DSF. My lab partner and I learned that statistically insignificant trends can be used to support conclusions drawn from significant data. Also, insignificance could mean that a ligand just doesn’t bind FKBP12, which is useful information to know for other researchers. Statistically insignificant trends can also motivate further experiments that can lead to a more confident conclusion.
Another difficulty my lab partner and I faced was due to our personal work preferences. While my partner is on top of things and completes assignments early, I have a strong tendency to procrastinate although I feel bad when procrastinating. This difference has led to “significant” stress for both of us. For the next 2 modules, I will try to work on assignments early and be less last minute. Overall, I look forward to future experiments, and for the power of statistics to shine through data analysis.

Comments

Post a Comment

Popular posts from this blog

Seeing the finish line!

The research proposal assignment start out as fairly intimidating.  Since 20.109 is my first introduction to biological engineering, I've appreciated being told exactly what to do.  To be tasked with crafting a research project that solves a biological problem, is original, and creative sounded like a lofty goal.  However, the more papers I read, the more I realized the real challenge was not having to come up with one idea, is was having too many ideas/options.  Even after Masrur and I had decided upon a pathway to target, there are so many diseases that the pathway is involved in that we then had to choose the most compelling disease to focus on.  Since we spent so much time deciding upon what exactly we were studying, we left little time for ourselves to define the research aim and detail the methods.  Overall, though, the project was really interesting and engaging.  I even found myself disappointed that the proposal wasn't real and that I wouldn't...
Read more

Presentation

The presentation went well. I think it was super helpful to fake present to classmates and instructors. Getting feedback from people who didn't know very much about our topic was really important for overcoming the bias of information. That’s when you know more than your audience and because of this assume a certain level of literacy with your topic that your audience doesn’t necessarily have. We practiced a lot to different people and I think that paid off and turned a presentation that was initially inscrutable and complex to something that was logical and understandable. 
Read more