Studying Statistics

Well, there's less than 4 weeks left of summer. As the day approaches when I get to attend my first live college lecture, I am taking things fairly easy. This week I started working on Udacity's Elementary Statistics course in conjunction with reading Nate Silver's The Signal and The Noise : Why So Many Predictions Fail But Some Don't. The first two chapters have conjured up memories of the Model Thinking course I took way back in November 2012 (it seems like a long time ago, anyway). Silver mentions Philip Tetlock's analogy of prediction modeling types in which those who consider a lot of information about one thing are likened to "hedgehogs", whereas models that consider a little bit of information about a lot of things are coined as "foxes". It comes from a quote by ancient Greek poet Archilochus, "the fox knows many things, but the hedgehog knows one big thing". Silver explains that when it comes to crafting prediction models, foxes often outperform hedgehogs. In fact, hedgehog models don't fare that much better than random chance. Also of note is how much better election polling models perform as election day approaches.

 

On the MOOC topic, here are the lessons I've completed in the ST095 course at Udacity:

1. Intro to statistical research methods
   This was basically just an orientation with the course goals and statistics terminology
   Defining constructs, operational definitions, double blind methods and control groups
2. Frequency Distributions & Visualizing data
   Graphs, histograms, spreadsheets
3. Central Tendency
   Mode, Mean, Median
4. Variability
   Spread, range, IQR, variance, standard deviation, outliers
5. Standardizing
   Normal distribution, z-scores
6. Normal Distribution
   Probability Density Function, converting z-scores
7. Sampling Distributions
   Central Limit Theorem
8. Estimation
   Confidence intervals, margin of error
9. Hypothesis Testing
   Null Hypothesis, alpha levels, one and two tail tests, conducting hypothesis test

I may try to read a chapter of The Signal and The Noise, do one Udacity lesson per day and briefly summarize the day's teachings to try to solidify my understanding of this stuff. Data analysis is something that I not only find fascinating, but it is likely to be a key skill to master in my pursuits. I've learned so much over the past year but I haven't had an opportunity to put many of those lessons into practice. My critical thinking abilities have certainly improved immensely by listening to podcasts like The Skeptic's Guide to the Universe and Penn's Sunday School, but mostly from taking the Mathematical Thinking, Intro to Logic, and Think Again courses on Coursera and reading a lot of science books. The need to find a discipline that I can focus on is still burning in me. Until I find that focus, I'll just have to be like the fox.