10
$\begingroup$

FINAL EDIT: You can find the transcript here.

It is high time to continue the CVJC series.

To remind for those who are new to the idea:

  • CVJC is a whole day meeting on chat where we discuss some paper and its theoretical/practical surroundings.
  • As mentioned above the event is whole-day (00:00-23:59UTC), but there are three meet-up sessions at:

    • 1:00UTC = 19:00 on Thursday CST (session A)
    • 9:00UTC = 3:00CST (session B)
    • 16:00UTC = 10:00CST (session C)
  • ...on which most talking take place; they are spread over day to put at least one CVJC session in reach regardless of time zone.

  • The paper must be OpenAccess or a (p)reprint suggested previously on a meta thread like this one and selected in voting.
  • I would try to invite the author (it worked last time).

So, please suggest papers (each in one answer)! The deadline for suggestions is traditionally a week ahead, so 23:59UTC 2.11.2011 23:59UTC 2.20.2011; the CVJC itself will probably take place at the 2.25.2011.

EDIT1: The CVJC3 will take place on 2011-03-11.

EDIT2: You can register (highly encouraged; this way everybody will know the approximated traffic on each session and you will get an e-mail reminder prior to the event) HERE.

$\endgroup$
  • $\begingroup$ Could you add the UTC times of each session here? When I click the 'HERE' above to go to the registration page I can see there are three sessions A,B and C but I can't see any times. $\endgroup$ – onestop Mar 9 '11 at 10:08
  • $\begingroup$ @onestop I've edited this question to make it clear; unfortunately I can't edit the registration page, there is only this "in xxx" form. $\endgroup$ – user88 Mar 9 '11 at 10:43
  • $\begingroup$ Tnanks mbq. Can you tell us which session(s) Sir David has said he'll be participating in, or is that allocation deliberately concealed?? $\endgroup$ – onestop Mar 9 '11 at 12:05
  • $\begingroup$ @onestop I don't know yet. $\endgroup$ – user88 Mar 9 '11 at 12:59
9
$\begingroup$

Let me suggest Statistical Modeling: The Two Cultures by Leo Breiman.

Abstract

There are two cultures in the use of statistical modeling to reach conclusions from data. One assumes that the data are generated by a given stochastic data model. The other uses algorithmic models and treats the data mechanism as unknown. The statistical community has been committed to the almost exclusive use of data models. This commitment has led to irrelevant theory, questionable conclusions, and has kept statisticians from working on a large range of interesting current problems. Algorithmic modeling, both in theory and practice, has developed rapidly in fields outside statistics. It can be used both on large complex data sets and as a more accurate and informative alternative to data modeling on smaller data sets. If our goal as a field is to use data to solve problems, then we need to move away from exclusive dependence on data models and adopt a more diverse set of tools.

$\endgroup$
  • $\begingroup$ Inviting author may be a problem. $\endgroup$ – user88 Feb 15 '11 at 20:46
  • $\begingroup$ @mbq: why is that? $\endgroup$ – Boris Gorelik Feb 16 '11 at 5:59
  • $\begingroup$ @bgbg , it appears he died in 2005, en.wikipedia.org/wiki/Leo_Breiman $\endgroup$ – Andy W Feb 16 '11 at 6:17
  • $\begingroup$ How sad. I sent a mail to the Berkeley Statistics dept manager to update Prof. Breiman's homepage stat.berkeley.edu/~breiman $\endgroup$ – Boris Gorelik Feb 16 '11 at 6:29
  • 8
    $\begingroup$ How about inviting one of the original discussants of the paper?? Sir David Cox, Bradley Efron, Bruce Hoadley or Emanuel Parzen? The first two might be particularly interesting as, to quote Breiman's rejoinder, "Brad Efron has serious reservations and D. R. Cox is in downright disagreement." Could ask if their views have changed at all in the intervening ten years! $\endgroup$ – onestop Feb 16 '11 at 11:55
  • $\begingroup$ @onestop Good idea. $\endgroup$ – user88 Feb 16 '11 at 18:46
  • $\begingroup$ @mbq, @onestop, any update on the participation of any of the discussants? $\endgroup$ – cardinal Mar 9 '11 at 1:13
  • $\begingroup$ @cardinal Only David Cox declared to come. $\endgroup$ – user88 Mar 9 '11 at 7:36
  • $\begingroup$ @mbq, that's great he is willing to participate. $\endgroup$ – cardinal Mar 9 '11 at 13:13
  • $\begingroup$ That's amazing! I am going to have to try to come ... en.wikipedia.org/wiki/David_Cox_%28statistician%29 $\endgroup$ – Ben Bolker Mar 9 '11 at 15:09
  • $\begingroup$ @mbq, maybe we can embed a link in the two cultures portion of the text on the banner on the main page. (?) $\endgroup$ – cardinal Mar 9 '11 at 16:06
  • $\begingroup$ @cardinal Good idea, done. $\endgroup$ – user88 Mar 9 '11 at 17:33
  • $\begingroup$ @mbq, nice. I might transpose "famous" and "Breiman's" as well. Sorry to nitpick. $\endgroup$ – cardinal Mar 9 '11 at 18:03
8
$\begingroup$

Another suggestion: Mindless Statistics by Gerd Gigerenzer

Abstract

Statistical rituals largely eliminate statistical thinking in the social sciences. Rituals are indispensable for identification with social groups, but they should be the subject rather than the procedure of science. What I call the “null ritual” consists of three steps: (1) set up a statistical null hypothesis, but do not specify your own hypothesis nor any alternative hypothesis, (2) use the 5% significance level for rejecting the null and accepting your hypothesis, and (3) always perform this procedure. I report evidence of the resulting collective confusion and fears about sanctions on the part of students and teachers, researchers and editors, as well as textbook writers.

$\endgroup$
5
$\begingroup$

In case nobody comes up with a good paper, I would suggest going through Strata 2011 conference and discussing some of the main contributions. It's all about how to deal with "big data" (social networks, and the like) which might be of interest to many of us.

$\endgroup$
  • $\begingroup$ This is not a perfect solution -- AFAICT this conference have not proceedings and is pretty wide. That's why I decided to extend the voting time by a week before, well, trying to accommodate this. $\endgroup$ – user88 Feb 13 '11 at 21:21
  • $\begingroup$ @mbq That's a fair point, indeed. I agree this is not the panacea, but still we can found some of the slides -- ok, that also means that we would have to provide a list of them and a rough TOC, not very convenient at all... I have some papers in mind that I will add within few days. $\endgroup$ – chl Feb 13 '11 at 22:14
3
$\begingroup$

I suggest "A Better Lemon Squeezer? Maximum Likelihood Regression with Beta distributed Dependent Variables" by Smithson and Verkuillen

available at http://psychology3.anu.edu.au/people/smithson/details/betareg/Smithson_Verkuilen06.pdf

Abstract

Uncorrectable skew and heteroscedasticity are among the “lemons” of psychological data, yet many important variables naturally exhibit these properties. For scales with a lower and upper bound, a suitable candidate for models is the beta distribution, which is very flexible and models skew quite well. The authors present maximum-likelihood regression models assuming that the dependent variable is conditionally beta distributed rather than Gaussian. The approach models both means (location) and variances (dispersion) with their own distinct sets of predictors (continuous and/or categorical), thereby modeling heteroscedasticity. The location submodel link function is the logit and thereby analogous to logistic regression, whereas the dispersion submodel is log linear. Real examples show that these models handle the independent observations case readily. The article discusses comparisons between beta regression and alternative techniques, model selection and interpretation, practical estimation, and software.

$\endgroup$

You must log in to answer this question.