Victrix Computational Chemistry and Medicinal Chemistry Consultancy
Contact us
  • Home
  • Philosophy
  • Services
  • Our Clients
  • Contact
  • About
    • Press Releases
  • Victrix Sounds Off
  • Blog

Sampling Bias and Self Help Books

5/28/2013

2 Comments

 
Picture
I have to admit I'm a bit cynical.  I'm a chemist, trained in theoretical organic chemistry, so it does come a bit with the territory.  I also have to say I hate self-help books.  Most particularly I hate those that purport to follow or survey successful, happy, or rich people and distill the secret of their success to some easily manageable number of traits you are urged to mimic.  Often you will hear such amazing insights as successful people think big. Or the unbelievably informative statistic all successful businesspeople think positively.  

I find it difficult to imagine that the unsuccessful among us somehow started a business thinking "I never want my business to ever make more than $25,000 a year and what's more, I will most likely fail anyway!"  Its nonsense, and this is where the statistics come in.  You need to take a large enough sample and rate the traits your  groups exhibits and at a minimum compute a p value for the hypothesis that trait A, let’s say positive thinking, distinguishes populations X and Y.  No one ever does anything remotely like this.  So despite the rather silly premise that you can think yourself rich, they don't even bother to try to test their hypotheses. 

The fact of the matter is that randomness plays a much higher role in people’s lives than they want to admit. Let’s take the example of new restaurants.  You often hear the statistic that 90% of new restaurants fail.  It's actually closer to 67%, but that still isn't very good odds.  So what is the differentiator? The successful new eateries are usually started as 2ed or 3rd establishments by an already successful chef.  The 90% failure rate applies to people who knew nothing about the restaurant business before venturing out into the mean streets. 

So is positive thinking what made the restaurants successful?  No it was understanding the business.  What made the 10% of newbies succeed? Chance.  

I'm pretty good at building homology models and using them for virtual screening.  I usually listen to Baroque opera while building them.  Therefore should you, gentle reader, rush out and purchase a selection of Lully, Campra and Rameau before your next homology modeling exercise?  Yes! Not because it will help you build models, but because the music is wonderful and will elevate your spirit.  To be successful at the homology modeling game.  Build a lot of homology models and learn from your mistakes.  Trial and error, not positive thinking is the secret of success.


2 Comments
Paul Hawkins link
5/29/2013 04:30:34 am

Admirable stance about more (some) statistics in directing the way we make decisions and live our lives. I would argue that the statement such as the one above about p-values

"You need to take a large enough sample and rate the traits your groups exhibits and at a minimum compute a p value for the hypothesis that trait A, let’s say positive thinking, distinguishes populations X and Y. "

rest on a very common misconception about the meaning of the p-value from a significance test.
Null hypothesis significance testing (NHST) does not generate a p-value for the likelihood of a hypothesis (Ho) given the data (D), P(H0|D). As such the p-value is not the posterior probability that we'd all like it be. Rather NHST computes the p-value for finding the data, given that the null hypothesis is true, P(D|H0). We have no way, with NHST to access the likelihood of H0, however much we would like to be able to.

Reply
Adam Kallel link
5/29/2013 04:46:16 am

Thanks for pointing that out Paul. I actually knew that... I know, I know. But your point is well taken. We need to do more to promote statistical literacy. Thanks for stopping by the Victrix CSO's blog and please let us know if we can help you in any way!

Reply



Leave a Reply.

    Adam Kallel Ph. D.

    Our CSO sounds off about drug discovery, computational chemistry and history

    Archives

    March 2014
    January 2014
    October 2013
    September 2013
    August 2013
    July 2013
    June 2013
    May 2013
    April 2013
    March 2013

    Categories

    All
    Ageing
    Amphetamine
    Anti-virals
    Biography
    Broken Society
    Caffeine
    Cancer
    Cd26
    Cluster Headache
    Colic
    Collaboration
    Corona Virus
    Danger Danger
    Depression
    Drug Discovery
    Drug Resistence
    Economy Of Scale
    Electron Microscopy
    Emergent Infectious Disease
    Encephalitis Lethargica
    Global Epidemiology
    Goal Setting
    HCoV-EMC
    Hiring
    Incorrect Pathologizing
    Incretin
    Influenza
    Jobs
    Me
    Merck
    Mergers
    Metabolic Disease
    Migraine
    Nerd
    Neurotransmiters
    Open Positions
    Over Prescription
    Overwork
    Pharma
    Pharmaceutical Industry
    Restructuring
    Stimulants
    Telomerase
    Tuberculosis
    Viral Replication
    Virus

    RSS Feed

    comments powered by Disqus
    View my profile on LinkedIn