Hi James M. Clark Professor of Psychology 204-786-9757 204-774-4134 Fax j.cl...@uwinnipeg.ca
>>> "Mike Palij" <m...@nyu.edu> 01-Sep-09 10:44:08 AM >>> On Tue, 01 Sep 2009 05:43:36 -0700, Jim Clark wrote: >>> "Mike Palij" <m...@nyu.edu> 31-Aug-09 2:12:45 PM >>> On Sun, 30 Aug 2009 22:18:52 -0700, Jim Clark wrote: >>>These lists, especially by themselves, do NOT allow the kinds of >>>inferences Mike appears to make. >> >>I'm not sure I understand what kind of inferences you're referring to. >... >JC: >I was referring to inferences like Mike's in the next few lines. Unless you're psychic or can see the future the text you quote cannot be the inferences you were referring to in your post because these comments were made in response to your "inferences" comments which, of course, were posted AFTER your inferences comment. Either you have amazing powers to warp the space-time continuum or you either neglected to use the "inferences" I made in some prior posts or I actually hadn't made such "inferences" in earlier posts and you decided to use statements AFTER your inferences comment. JC: What I said was "inferences Mike APPEARS to make" ... for example, in drawing such conclusions as "Do not get a PhD". You then chose to make those implied inferences explicit in a later message. >JC: >My earlier posting presented evidence that in fact PhDs are over-represented >in Mike's list, being about 1% or less in the general population and 4% in the >list. The percentage I've calculated from CPS 2008 (Detailed Tables) which is available at: http://www.census.gov/population/www/socdemo/educ-attn.html is 1.25% (=2,472/196,305) and this number has sampling error associated with it. Is 1.25% significantly different from 3.96% (i.e., 4/101)? Maybe, maybe not. If the proportion of people at different levels of educational attainment is same for both the richest 101 US citizens and the rest of the population, then this suggest that educational achievement has nothing to do with becoming the richest persons in the U.S., unlike the situation with, say, Nobel Laureates. If one want to say that 3.96% of a group represents an "over-representation" relative to 1.25%, I'll grant that but remind one about the difference between statistical significance and practical significance. JC: Assuming population p = .0125 (it would be lower in years prior to 2008) and sampling 101 people, one gets the following binomial probabilities for number of Xs (people with PhDs). px = probability of that number, cpx = cumulative probability, and upx = 1 - cpx. UPX is the relevant value. The probability of 4 or more people with phd is .009. x px cpx upx 0 .2807 .2807 .7193 1 .3589 .6396 .3604 2 .2271 .8667 .1333 3 .0949 .9616 .0384 4 .0294 .9910 .0090 5 .0072 .9982 .0018 As to the difference between statistical and practical significance, that would be a trickier question. We know, for example, that effect size can be a misleading indicator of practical significance (as in the classic aspirin study). I guess we could do something like getting the probability of being a billionaire given PhD and no PhD, both of which would be extremely small probabilities, and then taking their ratio. If having a PhD makes it 3x (hypothetically) more likely to be a billionaire than not having a PhD, is that of practical significance? >For those over 65, people not completing HS were UNDER-represented >in Mike's list compared to the over 65 general population. JC is assuming that educational attainment in the group of the richest people should mirror the proportions in the general population. It is not clear to me why this would be the case given that we know that the richest people are NOT a random sample from the general population and SHOULD differ from them in systematic ways -- afterall they are the richest and should differ on a number of dimensions though educational achievement does not appear to be particularly relevant. JC: No Mike, I am TESTING the assumption that HS nongraduates are equally or over-represented among billionaires. That assumption only makes sense relative to some comparison group. Take care Jim --- To make changes to your subscription contact: Bill Southerly (bsouthe...@frostburg.edu)