DRAFT: This module has unpublished changes.
User-uploaded Content
DRAFT: This module has unpublished changes.

Hello!

 

My name is Nathaniel Raley; I am a doctoral student, teacher, and researcher in Educational Psychology here at UT Austin, where I also earned my M.S. in Statistics.

 

I am passionate about learning, wherever it can be found and at every level of  analysis: in situ (the "wild", the "real world"), in vivo (the classroom, the psychology lab), in vitro (biological substrates, patterns of neural activity), and in silico (in machines). I see Bayesian statistics as learning itself formalized mathematically: in a certain sense, everything in life is data that arises from some generating process(es). We all have pet theories and working hypotheses about where this data comes from, and we implicitly update our model of the world---our beliefs about the worlds' parameters---in light of new experiences. We do this because these experiences give us evidence about their causes! But, human nature being what it is (riddled with biases, emotions, and other evolutionary baggage that makes us special) we as a species are not naturally very good at this; Bayesian inference is a tool that extends and refines our learning capacities far beyond their native levels, and is at last a fundamentally rational approach to truth-seeking.

 

My interest in this topic is long-standing, but Bayesian methods are only now becoming mainstream; as such, I am largely self-taught. However, I had the great fortune to take Dr. Stephen Walker's graduate course in Bayesian Statistics Methods (SDS 384) here at UT Austin. Though it's focus was very abstract---we never touched a computer; class-time was spent moving Greek letters around on a chalkboard---his class impressed on me a deep appreciation for the underlying theory: a mathematical elegance of the highest order. I now feel the call to go forth as an emissary and share this beauty with others (albeit in a more applied manner; I'll try to keep the Greek to a minimum!).

 

Professionally, I have used Bayesisan techniques in grant-funded educational evaluations here at UT Austin; in this Short Course, I will show you how rewarding (and easy!) it is to apply these methods to your own research. Once you have a handle on the basic concepts ("prior times likelihood equals* posterior"), you will realize that it is so general as to apply in almost every situation, and it may even completely change the way you see the world around you!

 

I am planning to teach a 4-day short course on Bayesian statistics at the Summer Statistics Institute next year. These courses are paid; they open to everyone but offered to the UT community at a discount and capped at aroud 50 participants. Having worked as an assistant at SSI for the past 2 years, I have observed what works well in these courses and I have decided to modify the traditional format by including an out-of-class online component consisting of short video lectures and exercises. For a proof-of-concept, I have created 3 ~5 minute videos with embedded exercises to be watched and completed before the first class; this is done to bring everyone up to the same level of prior knowledge (such as understanding notation and installing/using software) and to free up class-time for application activities (such as designing and performing analyses). Finally, students will make these techniques their own by applying them to their own research question(s) on the last day.

 

I have also mapped the curriculum for this course onto daily modules with in-class and out-of-class components. I have made decisions about what material to cover, how, on which day, and in what order. I have specified learning outcomes, my intended audience, course prerequisities and requirements. I use the videos and exercises to cover basics and ensure a baseline level of prior knowledge to make the best use of class time (akin to flipping the classroom).  I have also generated examples and created assignments to reinforce the material. For details about all of this, see my course page.

 

 

DRAFT: This module has unpublished changes.