|
Grad school is an interesting place.
The object is to become so smart,
and the way you do it is by feeling
really dumb a lot.
I've finally been feeling pretty smart though; I've become pretty good at Bayesian Analysis (that is, making a nice neat model and working through to a conjugate - or perhaps MH - posterior analysis - it's pretty fun and cool). There are a variety of philosophical (and aparently practical) arguments to prefer the Bayesian approach to others, but the world has seemingly somehow previously been able to keep on going without it. I wish there was some sort of magic bullet in Bayesian statistics that would make its advantages necesary - like, I don't know, have it somehow correct all the mistakes in the world that stastics has had a part in creating. Sadly (or not), that seems to be up to individual people. Rather, Bayesian Statistics appears to me to just be the next step in a series of improvements to "satistical methodology." I have put this in quotes because statistics is correct only in the vacuum of mathematical assumptions. We don't work under these presupposed assumptions. This flakyness is unsettling, not to mention the potential for abuse the whole field of statistics poses. I'm angry and confused that the world doesn't hold the same (aparently radical) values I do, such as wanting an equal distribution of wealth and hating the tolerance of exploitation. My hope then hangs on the thread of truth that statistics in the right hands can be useful in understanding the way our world works. But I want more than this. I want it to change the world. Lately, more and more, it seems that education is the only way we can change how the world works. I wonder if all the high powered Bayesian machinary that I've learned is really necessary for that. It seems simple frequentist analysis might be able to educate people just as well, as far as the most critical problems go. I mean, we can probably get a long way by just showing people the simplest summary numbers. Maybe it will take more than that though. Fortunately, my whole life I've been lucky. And I think I might be lucky again. It just might be that Bayesian statistics will turn out to be just the right tool to change things. From last year... I'm a First Year Graduate Student in the Statistics Department at Duke! And that is really an awesome/great/incredible thing! "But...Why?" you might ask, naively. Well: If you ARE NOT interested in statistics or the problems/areas statistics interfaces with (which is really quite doubtful, since statistics is useful, like, FOR EVERYTHING), then the answer is simple: Duke is freaking sweet, and so is North Carolina. Why? Quit asking Why so much. Look, come visit; then, you'll be able to answer that question in your own way. If you ARE interested in statistics or the problems/areas statistics interfaces with (which is probably the case, since statistics is useful, like, EVERYWHERE), then the answer is simple: Duke is THE STRONGEST Bayesian program In The World, and being Bayesian is like The Best Thing Ever!!! Bayesian methodology is clear and straight forward and always returns to the same simple/complete/unified (ascetically beautiful) principles and approach. Meaning and interpretation in a Bayesian framework is immediately apparent; understanding and intuition do not require a logical contortion. The Bayesian approach allows for considerations and flexibility in inference not even imaginable under other methodologies which are restricted to only a limited set of prepared analyses. And finally, (because of the above) the emphasis in a Bayesian framework is on correct understanding and correct modeling of a situation, NOT the moral struggle with arcane ad-hoc methods of questionable applicability and interpretation. I'm so EXCITED about Statistics! It stands on its own as a pleasing perspective and interpretation of our world - it's an amazing tool set for studying the most interesting issues imaginable. My interest in Statistics is thus, obvious. Equally obvious is that fact that pursuing statistics as a Bayesian is clearly preferable to every other option: The kind of work you do as a Bayesian and the way you do the work as a Bayesian is so exhilarating! I want the Bayesian mindset and tool-set because of the what it means my work will be like. I want the best statistical preparation I can get, and I believe this means being trained as a Bayesian statistician. In this regard, Duke is The Best. But look, I'm a first year, and so I'm just loving learning and soaking all of this stuff in - so you'll have to take my statements with a grain of salt. The point is that I'm Amazingly Excited about what I'm learning and how I'm being prepared: I'm having So Much Fun!!! |
|