[R] Resampling Stats software

[bbvaughn@bellsouth.net]

Brian,

Thanks so much for your comments.  Like I said, I am pretty much a novice at this idea.  My statistics degree emphasized a lot of theoretical knowledge, and the idea of simulation and resampling was never taught.  So, now that I am more in an applied field, I realize the need to educate myself.  I was taught SAS, and my current field stresses SPSS.  But a good friend recommended I check out R, so I'm trying to get into that now.

A few questions for you though.

> An introduction to what?  (It seems to confuse resampling and
> simulation-based inference.)
> 

Since I am new at this, could your clarify this just a bit?  Do you think this is a poor book?  Is resampling more to do with sampling from actual sample data?  (I could see how a lot of examples in this book are more based on simulation, and then called "resampling.")

It seems of interest the emphasis that statistics should be taught from this perspective (of simulating everything).  I'm not sure I agree with it totally though, because it always seems like you will get approximations all the time, where as probability formulas (if correctly specified and applied) will give you more exact measures.  It seems to possibly "dumb down" the understanding of statistics.  But then again, simulation/resampling does have the potential of answering questions for which there are no mathematical models.

What is your take on this?  Is a balance better?  I've used simulation before in teaching, but always to demonstrate theoretical knowledge (like the Central Limit Theorem).  This is the first I have seen of an emphasis on simulation as the goal of teaching.

> As a matter of terminology, this is not resampling as usually defined, so 
> I do wonder exactly what it is you are after.  For resampling in the usual 
> sense, I would echo Jason's recommendation of Davison and Hinkley's CUP 
> book.

What I'm after?  I wish I really knew!  :)

Let me see ... basically, I'm trying to understand the practical side of resampling.  I'm trying to get a basic balance of how all these differ ... such as simulation vs. resampling vs. bootstrap vs. Monte Carlo methods vs. Markov Chain Monte Carlo ... etc.  I'm trying to get a very practical knowledge before digging into the theory, but understand they probably will both come at the same time.

I want to be able to do statistical research using simulation.  There really aren't any specific goals in mind.  I'm just a learner.  I would like to do some research on robustness of certain educational measurement scales, and I'm sure this would involve simulation.

Does that make any sense?  Thanks for your help and suggestions.  I will check out that book for sure.

Brandon Vaughn
Chipola College