Veronica Vieland Summary

Naimun Siraj
Ignorance
Response 1

Dr. Veronica Vieland: Measuring Statistical Evidence in Biological Research

This lecture really resonated with me because I have taken a yearlong course on thermodynamics and know a bit about statistical thermodynamics. A very important topic that was discussed was the idea of the p value, which is a value that is used to make sense of a statistical hypothesis. The significance of the p value on statistical evidence is something that needs to be revaluated due to its complicated and fluctuating nature. Although p values and likelihood ratios do give valid statistical solutions, they sometimes tend to sway away from the accumulated hypothesis. In regards to human genetics, Dr. Vieland believes that the statistical evidences lead to “undesirable properties” when dealing with complex disorders. This leads to further scientific ignorance and perhaps a new approach is of the essence in order to break this shell of ignorance.
The analogy of understanding temperature measurement shows the significance of statistical evidence measurement and how hard it actually is to develop sensible statistics.
Before my thermodynamics course, I was in a state of ignorance regarding how temperature is actually measured. I thought like everyone else; if the environment is hot the high temperature is associated with a higher value and if it’s cold the temperature is associated with a lower value. This understanding of temperature was primitive and required a thorough understanding of thermodynamics to truly understand what these values meant. Similarly, when p values and likelihood ratios are discussed in statistical evidence, we associate these values with observations and do not truly know what might be going on in the background. It seems that these p values and likelihood ratios are a comfortable area where statistical evidence lights up on the surface, but gets darker as you probe deeper into
it.

It is amazing how we can reference the laws of nature to gain insight into problems from various fields. In a paper by Dr. Vieland, she delves into creating an equation of state analogy from thermodynamics for statistical systems. [1] The thought of trying to mimic the thermodynamic system into a statistical system is absurd yet beautiful because you don’t know how the model will behave in the end. The topic of statistical evidence is itself dwelling on the surface of ignorance. Although we know something, we don’t really see the full picture. This reminds me of the uncertainty principle in physics. One can measure either the momentum or position of a particle, but not both. We might know how fast a particle moving, but we won’t be able to pinpoint it. In another case, we may know the position of the particle, but not how fast it’s moving. There are many systems where we only know so much about it and struggle to break the shell of ignorance.

The idea of breaking this shell of ignorance regarding statistical evidence is perhaps to go beyond the null hypothesis. Finding a theory that dominates over the null hypothesis is a difficult task, but its been done by the giants of science. It’s human nature when one thing doesn’t work; we tend to keep looking for more answers to break the ignorance barrier. In this case, Dr. Vieland was not content with the common ways of measuring statistical evidence and found deep parallels with thermodynamics. This sets the tone for any kind of research where statistical evidence may be needed. Don’t be content with the observations and analysis that lead to a simple result; rather keep probing till something doesn’t work.

References:

[1] Vieland, V. J., and S. E. Hodge. “Measurement of Statistical Evidence on an Absolute Scale following Thermodynamic Principles”. Theory Bioscience, 5 Mar. 2013. Web. 23 Mar. 2015.