Author Message
gabefarkas
Joined: 31 Dec 2004
Posts: 1313
Location: Durham, NC
PostPosted: Mon May 03, 2010 12:16 pm Post subject: Reply with quote
Mike G wrote:
A book does generally fit in the bathroom better than a computer.
Depending on the type of paper used, it usually can be more useful, too.
Back to top
View user's profile Send private message Send e-mail AIM Address
schtevie
Joined: 18 Apr 2005
Posts: 408
PostPosted: Mon May 03, 2010 12:18 pm Post subject: Reply with quote
That's harsh....
Back to top
View user's profile Send private message
HoopStudies
Joined: 30 Dec 2004
Posts: 705
Location: Near Philadelphia, PA
PostPosted: Mon May 03, 2010 1:27 pm Post subject: Reply with quote
Ed Küpfer wrote:
Mike G wrote:
More to the point, who can write a book fast enough that it's not outdated when it gets to print?
I was talking to DeanO in Boston about his book. Because he focused on methods, the only way it would be outdated is if his approach was wrong or misguided, or if new data made his methods obsolete. Yet here we are in 2010, and we're still mostly using boxscore stats. His book still retains most of it's value.
There is room for another book that focuses on methods, and incorporates data sources that weren't available to Dean back in the day.
I also wrote a lot of methods to be estimates, things that can be improved upon with more stats available. So, as some have done here, you can take new stats and improve the estimates in BoP.
There are nonetheless things that sit "out there."
- Some information doesn't fit within the framework of BoP. Raw plus/minus data on its own doesn't, yet components of it do (especially defensive stuff, but some offensive stuff). Deflections aren't in the framework. Nor are contested shots explicitly.
- Methodology variations. As we've talked about, there are a lot of player valuation methods out there. What hasn't been settled, though, is two things. First, how to evaluate them against each other? It's been discussed, but not a lot and definitely not settled. Second, how much do individual valuations really turn into the team total (we all assume linearity in valuations).
Having had to deal with these issues since BoP, I have knowledge but no desire at this time to follow up with another book.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website
timmys24
Joined: 19 Feb 2009
Posts: 9
PostPosted: Mon May 03, 2010 1:56 pm Post subject: Reply with quote
basketballvalue wrote:
BobboFitos wrote:
It is a little bit annoying when random acquaintances refer me to Berri's blog, though.
Regardless of what we think of his analysis, he's clearly effective at getting attention/respect by the general public. If we want to be able to point mainstream folks to a different book, one of us (here's looking at you kp0) needs to write it. Sadly, folks are always looking for the next new book, so it can be hard to point to Basketball on Paper six years later, as good as it is.
Seriously, a Basketball Prospectus version of Baseball Between the Numbers would be useful book to point to the next time someone asks us about this one.
Thanks,
Aaron
Berri's mode of attack is disappointing.
Good info and methods is easily found and it is out there.
While some may want to find a great book to reference, those truly interested in the information and methodology will find it as long as it is available somewhere.
Though, as others have said, sitting in the bathroom with your laptop can be a little uncomfortable...
Back to top
View user's profile Send private message
Eli W
Joined: 01 Feb 2005
Posts: 402
PostPosted: Mon May 03, 2010 9:52 pm Post subject: Reply with quote
DLew wrote:
Yeah, to me it is very troubling that he has not addressed Eli's findings (replicated by Ryan Parker). That is a very well constructed, sound study, and if Berri were actually interested in the truth he should have at least addressed why he thinks it is flawed. Instead he has refused to address it because it contradicts his viewpoint. That is not good science.
Berri now has posted some criticisms in the comments to this thread:
http://dberri.wordpress.com/2010/05/01/ ... g-on-wins/
As far as I can tell, his objections are as follows:
Objection 1: A general objection to any kind of simple linear regression or correlation-based analysis
This seems bizarre to me, though maybe from an economist's perspective multiple linear regression with buckets of independent variables is the norm. But I think that single variable regressions (or basic correlation analysis) can be very useful for answering a lot of questions, or at least providing a starting point. I think it comes down to the specifics of the problem being looked at. To figure out whether there's an issue in using simple regression to study usage and efficiency in this way, one has to look at what important variables may have been omitted. Simple correlation shows low usage lineups under-perform their projected offensive efficiency, but maybe some missing variable is the real cause of this relationship.
A commenter mentions the defense of the opposing lineup as something that could be missing, which is valid (I mentioned it in the original study). Berri doesn't mention any other specific variables he thinks need to be included in the study. But if people have suggestions of other variables worthy of inclusion I think it would definitely be worth it to re-run the analysis with them included.
Objection 2: Lineup studies are flawed because particular 5-man lineups don't play that many minutes together in a season, and players aren't always trying to outscore their opponents of any given stretch of the game
These can definitely cause problems but I don't see how they plague the specific study I did. I basically was aggregating lineups together based on the players' combined usages, so I don't think there were sample-size problems. And the vital benefit to looking at the lineup rather than the season level is that it allows you to isolate situations where players were forced to increase or decrease their usage. Also, I don't see how the second issue would skew the results though one might want to see if omitting garbage time minutes changed anything.
Objection 3: A study Berri did for Stumbling on Wins controlled for many factors and found a much smaller relationship between increased usage and decreased efficiency
I think Berri is talking about the study mentioned in Endnote 32 to Chapter 8 (page 206). Unfortunately the description of the study doesn't mention any of the variables he controlled for, and the additional notes for the Chapter 8 on the book's website (
http://stumblingonwins.com/ ) don't discuss the study.
In the Endnote he says that if you "regress shooting efficiency...on shot attempts" (presumably on the player-season level) you don't see a negative relationship between usage and efficiency. He then says that there is a small negative relationship when you look at individual players' changes from season to season (i.e. players who increase their shots from one year to the next tend to have a decrease in efficiency). An increase from 16.3 shots per 48 minutes to 25.3 shots per 48 minutes would correspond to a decrease in adjusted FG% from 48.4% to 47.1%.
I'll leave it to others to comment on the merits of these kind of studies, which have been discussed a lot here in the past.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Ryan J. Parker
Joined: 23 Mar 2007
Posts: 708
Location: Raleigh, NC
PostPosted: Mon May 03, 2010 10:36 pm Post subject: Reply with quote
There is nothing wrong with fitting ZERO independent variables in a model (i.e. intercept-only).
There are both theoretical and practical issues with the model Eli and I have used to examine this, but the idea that there aren't meaningful models with a single (or zero) predictors is untrue.
There are plenty of important things you can try to control for to get a clearer picture of the exact value of usage vs efficiency, but his example isn't an apples-to-apples comparison. Anyone that's read BoP would realize that the analysis is flawed, and it's not because of a single predictor variable. Instead, it's because we know we'd instead want to regress net efficiency to win% even though that's still a single predictor!
PS: Props to JSill for attempts at interaction.
_________________
I am a basketball geek.
Back to top
View user's profile Send private message Visit poster's website
Guy
Joined: 02 May 2007
Posts: 128
PostPosted: Tue May 04, 2010 9:16 am Post subject: Reply with quote
The idea that Eli's study can be dismissed because it has "one independent variable" is just nonsense. That's merely a function of how Eli constructed the model. His DV is actually the difference between actual and projected efficiency, and his indep variable is the sum of 5 players' projected efficiencies, weighted by their expected usage. So he could have easily had 6 independent variables -- but it would be the same model!
In fact, Eli effectively controls for a lot of things by controlling for every player in the lineup. It could be that something else is going on when low-usage players are on the floor together and forced to take more shots, and Eli concedes this. But the most obvious one -- blowout games -- would suggest weaker opponents on the floor and thus makes his results even more impressive. Unless Berri can point to a plausible omitted variable that could account for such a large usage impact, the study's results should stand.
But let's not take this at face value. Berri doesn't really have a problem with a single variable model -- he uses them himself. For example, he argues that NFL teams do a lousy job of drafting QBs by looking at the correlations of draft rank and passing metrics. He is simply looking for a pretext for dismissing Eli's study (as he did with the Lew/Rosenbaum paper previously). Berri is "all in" on denying any significant diminishing returns on rebounds, and any real usage/efficiency tradeoff. If he concedes error on those, WP and 5 years of work collapses. David says this "is not good science," but it's not really science at all. Berri is not offering intellectual engagement -- it's a desperate, goal-line stand.
The NFL draft study poses a less severe test of Berri's intellectual honesty. Brian Burke has done a nice study at Advanced NFL stats, showing conclusively that draft order IS correlated with QB talent, contra Berri. (I think the Basketball Ref blog folks did a similar study showing that Berri was also wrong about the NBA draft.) Berri has promised to post a reply. This is a great chance for Berri to say, for once, "I got this one wrong. Nice work Brian. NFL decision-makers are making rational choices in this case, and I missed it." But I will go out on a limb here and predict that instead Berri will identify some minor flaw (real or imagined) in Brian's analysis, point out it isn't "peer reviewed," and then politely signal his intent to ignore it. But maybe Berri will prove me wrong on this one.....
Back to top
View user's profile Send private message
Fire
Joined: 12 Dec 2007
Posts: 5
PostPosted: Tue May 04, 2010 9:45 am Post subject: Reply with quote
Well it seems clear to me that constructive argument and criticism will not be happening from both sides. The only way to truly make your point is to go through and intermediary... and Berri is able to do this through peer reviewed articles. It is too bad nobody here can/is willing to challenge him through that process
Back to top
View user's profile Send private message
DSMok1
Joined: 05 Aug 2009
Posts: 587
Location: Where the wind comes sweeping down the plains
PostPosted: Tue May 04, 2010 10:14 am Post subject: Reply with quote
Fire wrote:
Well it seems clear to me that constructive argument and criticism will not be happening from both sides. The only way to truly make your point is to go through and intermediary... and Berri is able to do this through peer reviewed articles. It is too bad nobody here can/is willing to challenge him through that process
The peer-reviewed system is a bit overrated, in my opinion. The people here are more capable at reviewing Berri's papers than those doing the peer-review, for those here are specialized in the field with which he is dealing.
Having the "credentials" from a university is extremely overrated in our society--common sense and a solid understanding of the issues discussed is far more critical.
I say that as someone with multiple degrees, who has published in a peer-reviewed journal.
Back to top
View user's profile Send private message Visit poster's website
Ed Küpfer
Joined: 30 Dec 2004
Posts: 785
Location: Toronto
PostPosted: Tue May 04, 2010 10:19 am Post subject: Reply with quote
I have stayed out of every Berri-related exchange so far, but I have to say that a regression model with only one predictor variable is somehow not valid is the dumbest thing I've heard.
edit:
Fire wrote:
Well it seems clear to me that constructive argument and criticism will not be happening from both sides. The only way to truly make your point is to go through and intermediary... and Berri is able to do this through peer reviewed articles. It is too bad nobody here can/is willing to challenge him through that process
We've talked about this before. The thing is that posting stuff here or other places is peer review. And more to the point, the people doing the review here are more knowledgeable about the topic than at the journals Berri publishes with. So the only reason to publish in a journal is to satisfy Berri, which isn't exactly the best reason in the world.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
Guy
Joined: 02 May 2007
Posts: 128
PostPosted: Tue May 04, 2010 10:50 am Post subject: Reply with quote
Quote:
to say that a regression model with only one predictor variable is somehow not valid is the dumbest thing I've heard.
Berri isn't content to argue that a 1-variable model is a mistake in this case. He goes on to say that the failure to spot this obvious mistake shows how poor the skills of the "on-line people" are in general:
It should be troubling to people that no one noticed that this model only had one independent variable. So nothing was held constant in this analysis. Again, this really is a fundamental issue in regression analysis. And the fact everyone missed this lends credence to my hypothesis. The on-line people do not do a very good job of evaluating statistical models. Given the training many of these people have received (and their general lack of peer-reviewed research experience), this is not a surprising shortcoming.
At some point, you just have to laugh....
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 3535
Location: Hendersonville, NC
PostPosted: Tue May 04, 2010 12:15 pm Post subject: Reply with quote
Not that it necessarily means anything, but:
Has any NBA team offered DB a consulting job?
Is it even imaginable?
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
schtevie
Joined: 18 Apr 2005
Posts: 408
PostPosted: Tue May 04, 2010 12:25 pm Post subject: Reply with quote
Mike, again, the problem is the faulty specification of the regression. You need another independent variable besides the Dave Berri dummy in explaining the probability of employment as an NBA statistical consultant.
Back to top
View user's profile Send private message
Fire
Joined: 12 Dec 2007
Posts: 5
PostPosted: Tue May 04, 2010 12:29 pm Post subject: Reply with quote
Well, all I am saying as in impartial observer it is hard to say who to trust. I have a fairly limited knowledge of statistics, considering I literally just took my introduction to econometrics final a few hours ago. And after taking this class, I must say I like the idea behind Berri's metric (regressing everything on efficiency differential). Then again, I am very skeptical of the model as it just seems off, and a lot of what you guys put out there makes sense to me. But than again, for the most part Berri has refuted what you guys say, on the basis that the tests you guys run are invalid. And after all, at least as far as a I can tell, Berri is the highest academic of people contributing to basketball statistics, and he also has a laundry list of published academic articles on many different aspects of sports economics. But than again a lot of his criticisms don't seem to have very solid footing (although due to my inexperience with stats and econometrics I really don't know).
In conclusion, if the evidence disproving Berri's theory really is there, I think it should go out into the academic world. The fact is, like it or not, Berri is on his way to becoming the next Hollinger. With the release of his second book, and this past decade basically becoming the coming out party for advanced metrics, as long as there is no definitive evidence against him, Berri is only going to gain more and more clout in the public's (and especially media's) eye. It can already be seen with Henry Abbots constant linking to Berri's blog and stats. If he and his methods bother you now, it is only going to get worse.
Back to top
View user's profile Send private message
DSMok1
Joined: 05 Aug 2009
Posts: 587
Location: Where the wind comes sweeping down the plains
PostPosted: Tue May 04, 2010 12:48 pm Post subject: Reply with quote
Fire wrote:
Well, all I am saying as in impartial observer it is hard to say who to trust. I have a fairly limited knowledge of statistics, considering I literally just took my introduction to econometrics final a few hours ago. And after taking this class, I must say I like the idea behind Berri's metric (regressing everything on efficiency differential). Then again, I am very skeptical of the model as it just seems off, and a lot of what you guys put out there makes sense to me. But than again, for the most part Berri has refuted what you guys say, on the basis that the tests you guys run are invalid. And after all, at least as far as a I can tell, Berri is the highest academic of people contributing to basketball statistics, and he also has a laundry list of published academic articles on many different aspects of sports economics. But than again a lot of his criticisms don't seem to have very solid footing (although due to my inexperience with stats and econometrics I really don't know).
In conclusion, if the evidence disproving Berri's theory really is there, I think it should go out into the academic world. The fact is, like it or not, Berri is on his way to becoming the next Hollinger. With the release of his second book, and this past decade basically becoming the coming out party for advanced metrics, as long as there is no definitive evidence against him, Berri is only going to gain more and more clout in the public's (and especially media's) eye. It can already be seen with Henry Abbots constant linking to Berri's blog and stats. If he and his methods bother you now, it is only going to get worse.
I got my start in advanced basketball statistics from the Wages of Wins blog. It made a lot of sense, at first, though some of the conclusions seemed a bit outlandish.
After a while, I started to realize there were perhaps better methods out there. Statistical Plus/Minus, one of the methods I like, mirrors what DBerri did--only on an individual instead of a team basis. The issue is that the statistics that matter on a team level are not exactly what matters on an individual level.
A basic example is that on a team level, a defensive rebound = an offensive rebound. However, on an individual level, it is different. On an individual level, 2 other things could happen--one of the other 4 players on his team gets the rebound, or the other team gets the rebound. And the probability of those other things happening is different depending on the end of the court. On the offensive end, most of the time if the player fails to get the offensive rebound, it's going to the defense. On the defensive end, there are often multiple defensive rebounders in position. The MARGINAL value of the offensive rebound is higher.
This is borne out in the Statistical Plus/Minus regression, which maps the box score stats onto the (admittedly noisy) adjusted plus/minus data. This regression smooths out the warts of the APM data, while retaining a better valuation of the player's contributions than what the Wins Produced metric yields.
page 3 of 3
Author Message
Ilardi
Joined: 15 May 2008
Posts: 265
Location: Lawrence, KS
PostPosted: Tue May 04, 2010 1:20 pm Post subject: Reply with quote
Ultimately, this is a conversation about metric validity - i.e., to what extent each metric actually measures what it purports to. And for most scientific disciplines, predictive validity is by far the most salient facet thereof. Unfortunately, precious little out-of-sample testing of the predictive validity of various player evaluation metrics has occurred thus far.
Dan Rosenbaum and Dave Lewin published one such analysis a few years ago (a Nessis paper that should still be available on line somewhere), and Berri's Wins Produced yielded poor prediction: if memory serves, it could not even outperform an efficiency-adjusted minutes-played variable in predicting next-season team performance. Of course, PER and APM (noisy, single-season version) also performed poorly, so it's not just a knock on WP.
More recently, Joe Sills reported some nice (as-yet-unpublished?) out-of-sample testing with a "ridge regression" enhancement of APM, which yielded much more encouraging predictive validity - even better than that of multi-season APM (with 4 seasons' worth of data).
Joe, perhaps - in all your free time - you could see how well Berri's WP metric does with your out-of-sample test, and then make the resultant data publicly available? It would, I believe, go a long way toward settling this issue in the manner in which it deserves to be settled: with data.
Back to top
View user's profile Send private message
Joe
Joined: 27 Sep 2009
Posts: 94
Location: Long Island, NY
PostPosted: Tue May 04, 2010 1:52 pm Post subject: Reply with quote
Fire, I recommend you read these two threads before coming to a conclusion about the merits of Wins Produced and Berri's methodology:
http://sonicscentral.com/apbrmetrics/vi ... wages+wins
http://sonicscentral.com/apbrmetrics/vi ... wages+wins
There's 36 pages of posts between the two of them. I recommend only reading posts by Dan Rosenbaum and HoopStudies in order to accelerate the time it takes to read that (and that should give you about 99% of the pertinent information anyway, without many tangents). There's also one great post by mikez somewhere in there detailing the difference between a model's descriptive, ascriptive, and predictive qualities, if you can find it.
The first thread started when the original Wages of Wins was first released, so it gives a nice timeline for the discovery of issues with the model and you can see the thought process of various people from the basketball community in evaluating the model (Rosenbaum is a consultant for Cavs, HoopStudies aka Dean Oliver is an executive for the Nuggets, and Mike Zarren is the Ast. GM of the Celtics).
Really, though, anyone who hasn't read those threads should make it a priority to do so. Both are incredibly informative, not just about Wins Produced, but about evaluating basketball in general. It may take a little under an hour to get through it all, but well worth it. Some of the best threads in the board's history in my opinion.
_________________
http://www.hoopdata.com
Back to top
View user's profile Send private message Visit poster's website
Mike G
Joined: 14 Jan 2005
Posts: 3604
Location: Hendersonville, NC
PostPosted: Tue May 04, 2010 4:31 pm Post subject: Reply with quote
Actually, the introductory discussion was a few months earlier:
http://sonicscentral.com/apbrmetrics/vi ... .php?t=681
I guess we summarily dismissed it, as it was all so implausible. It immediately failed the 'laugh test', etc.
Now it's as though someone opened a retail business in a strip mall near you, and they spent 20 grand on a huge and beautiful sign: SHOES, say.
And you go in the store, and there aren't any shoes; just junk that happens to fit into shoe boxes: cracked cups, rubber chickens, vacuum cleaner attachments, etc. You realize there aren't any shoes here, and what the heck is going on?
Meanwhile, people drawn in by the gaudy sign tend to spend more than a few minutes looking and looking for shoes. The salespeople talk to you about this style and that style, as customers try to put their feet into old picture frames, wads of plastic, and other trash.
You consider telling others that these are not shoes, this is not a shoe store, etc. But it's so obvious; what is the necessity?
It's not common for anyone to spend 20 grand on a sign and have nothing to sell, and it's odd to spend the time to write a book with nothing to offer. Unless you happen to know that if you do it, they will come.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
DLew
Joined: 13 Nov 2006
Posts: 224
PostPosted: Tue May 04, 2010 6:16 pm Post subject: Reply with quote
Fire, for the record it seems pretty unlikely Berri is going to be the next Hollinger. Berri is an academic publishing papers about economics, that are reviewed for the validity and relevance of the economic concepts, not really for the validity of the basketball analysis. Hollinger on the other hand, is a journalist who writes about basketball and is judged (by the public) on the accuracy of his analysis. I don't think either wants to be in the other's shoes.
It's worth noting that Berri really struggled to get his derivation of Wins Produced published in a peer-reviewed journal (I think he did recently after 10+ years of trying, right?). Most of his publications are for things unrelated or minimally related to the Wins Produced metric (i.e. scoring explains salaries, wins explain attendance, etc). However 95% of what he writes in non-peer reviewed forums (website, book) is about Wins Produced. The reason for this discrepancy is that no academic publication cares about Wins Produced, they only care about the questions one might investigate using any reasonable measure of player performance (like Massey and Thaler did with the NFL Draft). Academic economics journals can't, and don't, evaluate player performance models.
Back to top
View user's profile Send private message
gabefarkas
Joined: 31 Dec 2004
Posts: 1313
Location: Durham, NC
PostPosted: Fri May 07, 2010 10:46 pm Post subject: Reply with quote
Guy wrote:
Berri isn't content to argue that a 1-variable model is a mistake in this case. He goes on to say that the failure to spot this obvious mistake shows how poor the skills of the "on-line people" are in general:
It should be troubling to people that no one noticed that this model only had one independent variable. So nothing was held constant in this analysis. Again, this really is a fundamental issue in regression analysis. And the fact everyone missed this lends credence to my hypothesis. The on-line people do not do a very good job of evaluating statistical models. Given the training many of these people have received (and their general lack of peer-reviewed research experience), this is not a surprising shortcoming.
The fact that he writes "on-line" with the hyphen makes him sound like he's 85 years old, has trouble using the tele-phone, isn't sure how to log-on to the inter-net, just finished reading William Howard Taft's bio-graphy in the news-paper, enjoys going to the cinema to watch the moving-pictures, will soon be in a wheel-chair, and wishes he had a ro-bot butler.
But most of all, he just doesn't consider us true gentle-men.
Back to top
View user's profile Send private message Send e-mail AIM Address
gabefarkas
Joined: 31 Dec 2004
Posts: 1313
Location: Durham, NC
PostPosted: Fri May 07, 2010 10:48 pm Post subject: Reply with quote
Fire wrote:
And after all, at least as far as a I can tell, Berri is the highest academic of people contributing to basketball statistics, and he also has a laundry list of published academic articles on many different aspects of sports economics.
I have published several papers and posters as a statistician in the field of pharmaceutical research and development. Does that make me qualified to analyze basketball statistics?
Back to top
View user's profile Send private message Send e-mail AIM Address
Joe
Joined: 27 Sep 2009
Posts: 94
Location: Long Island, NY
PostPosted: Sat May 08, 2010 10:07 am Post subject: Reply with quote
gabefarkas wrote:
Fire wrote:
And after all, at least as far as a I can tell, Berri is the highest academic of people contributing to basketball statistics, and he also has a laundry list of published academic articles on many different aspects of sports economics.
I have published several papers and posters as a statistician in the field of pharmaceutical research and development. Does that make me qualified to analyze basketball statistics?
Only if you agree with David Berri.
_________________
http://www.hoopdata.com
Back to top
View user's profile Send private message Visit poster's website
mtamada
Joined: 28 Jan 2005
Posts: 377
PostPosted: Sun May 09, 2010 1:05 am Post subject: Reply with quote
Fire wrote:
And after all, at least as far as a I can tell, Berri is the highest academic of people contributing to basketball statistics
Dan Rosenbaum was, like Berri, a tenured professor of economics, until he spent a year on the President's Council of Economic Advisors and subsequently took a job at the OMB. Rosenbaum like Berri applied econometric techniques to study basketball but IMO with much more broad-mindedness and more useful results. He was also a skeptic of Berri's models and results (and Winston's as well -- not so much a skeptic of the model, but of trying to draw premature conclusions from the model, which Winston seems to do continually).
Back to top
View user's profile Send private message
BobboFitos
Joined: 21 Feb 2009
Posts: 200
Location: Cambridge, MA
PostPosted: Sun May 09, 2010 1:50 am Post subject: Reply with quote
mtamada wrote:
(and Winston's as well -- not so much a skeptic of the model, but of trying to draw premature conclusions from the model, which Winston seems to do continually).
WW just needs someone to teach him the concept of sample size. Once he learns that, his work will be really useful.
_________________
http://pointsperpossession.com/
@PPPBasketball