Posted by Michaelson from em11_686.utsi.edu on July 01, 1998 at 15:48:54:

In Reply to: Re: Probability and Sadistics.....(that's what my students say) posted by Stephen on June 30, 1998 at 14:17:41:

: : : : About the thing I said below. Let me put it this way. Every big Indy fan I know says they like Raiders or Last Crusade. Every person that has at least seen the IJ trilogy says Temple. Ok?

: : : : Son, look at what you've written here, 'every person that has at least seen the IJ trilogy says Temple'. As if a "real fan" (whatever that means) couldn't possibly like ToD the best! As we started out below in the forum, you and I are going to tangle right now.

: : : In making your confident edict, you've obviously applied hypothesis testing to remove all doubt. Thus, you tested and failed to accept your alternate hypothesis (Real IJ fans could choose Temple as their favorite movie) and therefore had to accept the null hypothesis (Real Fans would only choose Raiders as their favorite). Brilliant! And as the statistical wonder that we all know you are, we know that you chose an instrument (will make it easy, say a 30-question survey using a Likert Scale) to assess 1) Which of the three movies was the individuals favorite (the dependent variable) and 2) whether the individual was a "Real Fan" or merely a pasty, wimp, Infidel living in their parent's basement (the independent variables). And we also know that your phrase, 'all the real fans I know' couldn't possible mean just the four guys you hang out with at the Circle K on a Friday night because for a sample size to be statistically significant it must be where n>30. So we know that you found more than 30 "Real Fans" (again not your four friends at the Circle K) and more than 30 infidels to make your testing statistically significant and thus acceptable to the world of science. But not satisfied with that, we suspect that you realized that the population standard deviation was unknown (because you couldn't possible have surveyed all the "Real Fans" and infidels in the world), therefore you had to perform a Two-Tailed Test. Here your null hypothesis was that the population means for the two samples were assumed to be equal (u1=u2). This meant that your original assumption, (Man, only "Real Fans" choose Raiders or Last Crusade as the best!) was valid if, after testing, the null hypothesis could not be rejected. And because you did find more than 30 "Real Fans" and infidels each, you selected a z score where alpha was 0.05. You defined the region of acceptance in you Two-Tailed Test as -1.96

: : : But that's what it takes in reality when you stand up and announced to the world, "A Real Fan would never like Temple of Doom the best!" Yep, that mess above is exactly what it takes. You don't know jack squat. Here's the deal. I'm not a real fan. I'm a violently real fan... who just happens to like Temple of Doom (and then barely) as the best of the three. I collect everything from cards to lunch boxes, books to props. And I know the dialogue almost by heart because I love the spirit of the films. And finally, while you're hanging out at the Circle K with your buds this July 4th weekend trying to figure out what to do, I'm going to be standing in Chiang Mai, in the Golden Triangle between Laos, Burma, and Thailand, getting ready to ride an elephant into the jungle on a trip to see the ruins of an ancient Thai temple. And I'll be on that elephant with my fedora... my Wested Jacket... my Mark VII satchel...and a big smile on my face.-CP

: : I tape master level classes in Statistics, Captain, and you've stated your case well. But in every class that I've ever taped, the professor ALWAYS rejects the Null Hypothesis. I have asked many times, if you're going to reject so much, why bother with it in the first place? I've never had a proper answer.... Anyway, interesting discussion. Regards. Michaeslon

: I teach statistics and just have time to point out two things: CP...

: 1) since the population mean and standard deviation are unknown,

: a z-test is inappropriate... you would use a t-test, but then only

: if the "real Indy" fans and the "infidels" could be randomly assigned

: to each group (which, of course, they cannot in this instance) and

: Michaelson...

: 2) whether you reject the null or not, you need to go through the

: process of hypothesis testing as described by CP; otherwise, you're

: just stating your opinion without any evidential foundation.

: End of stat lesson :-)

: Regards,

: Stephen

Stephen, that's exactly what our Chairman of the department said. I'll probably just reject it out of hand. I'm old, I'm allowed. Regards. Michaelson