Geekfest: Bootstrapped Credible Intervals of the Mean in SAS

Its not called 'Uncommon Reaction(s)' for nothing. This will be weird to many visitors, but that's ok. If you're not into this kind of thing just move along to the next post. If you don't care for the talky parts, I've pasted the full code without interruption at the bottom.

I wrote a little program to calculate bootstrapped nonparametric confidence intervals of the mean in SAS. Yes, I know STATA users, STATA apparently has a command that just does this, but you know what? I don't use STATA and I don't feel like learning. Besides, when I write it in SAS it reinforces my understanding of what I'm doing.

I've been using this with simulation data, but you could use the same program with observational data. Its not different. There is nothing particularly special about this, I just combined a bunch of different pieces of code from different SUGI doc's. Its just when I looked there was nothing

Here goes (and please feel free to post questions in the comments).

Situational Statistics or Should You Go For Two?

Our bias towards naive empiricism leads many to falsely believe that consistently reliable predictions can generally be drawn provided data has been gathered from a controlled setting. However, as 'Coaches Still Vexed by Going For 2' illustrates, strategic decision making is inherently uncertain and context specific even in the highly controlled world of NFL Football. Social scientists have no choice but to incorporate both uncertainty and context into their conclusions. The 'takeaway' message from all of this? Beware of 'takeaway' messages.

Here's a quiz from the article. I think even nonsports fans would enjoy the full piece.

Here are 4 situations. Should you go for two?

1. Ahead by 5 at the end of the first half.

2. Behind by 2 at the end of the first quarter.

3. Behind by 10, 12:30 left in the game.

4. Ahead by 8, beginning of the fourth quarter.