I run the service R-bloggers:

http://www.r-bloggers.com/about/

An aggregator of R related articles, from blogs.

And wanted to encourage you to join R-bloggers.com at:

http://www.r-bloggers.com/add-your-blog/

The idea behind the project is to share readers in order to gain readers: R-bloggers already has over 800 RSS subscribers (that are growing everyday).

I built it in order to find all the R bloggers out there. So far I found over 45 bloggers, which also agreed to add there feed (and some to give a link back and post about it).

And would love it if you might agree to join as well.

Feel free to erase this comment if it clutters the blog too much.

All the best,

Tal

Is Chris Dillow the fellow responsible for the last quote in my post? Then I am not placated *at all*. His mistake was magnitudes worse than yours!

I am perfectly happy with slang, but I have to point out when it’s misleading. Your beliefs in the “importance” of inequality in society is very likely to be a value judgement, not a state parameter, and so it will not be updated in the light of new data (i.e. it cannot become a “Bayesian posterior”!). Of course the boundary between the two is blurred because you might not be sure how much you care about, say, inequality in society, because you are not sure of its effects on other parameters which you care about; but this is still an uncertainty in the utility function.

Utilities fit nicely into the Bayesian paradigm, but are certainly not exclusive to it. Mixing up their subjectivity with the subjective belief of system variables (only the latter of which are governed purely by my famous posthumous Theorem) is not recommended for the faint-hearted.

]]>“Bayesian priors” has been used for some time (it originated with Chris Dillow) as a slang for “prejudices”.

Nothing to do with mathematics or statistics at all, just a little bit of blogging argot.

]]>I’ve never seen this show, “Cash Cab”, but from your descriptions I think you might have to revise your views slightly. When people think “I came with nothing and therefore I’ll just gamble what I have anyway, because I can’t do worse than getting back to where I started”, that’s (potentially) an application of Bayesian decision theory, where the decision taken is the one that maximises expected utility, a function not only of a person’s subjective beliefs concerning system parameters but also their utilities for possible outcomes.

The definition of what is “rational” behaviour, even “Bayes rational” behaviour, is harder to pin down than you think!

]]>They almost always decide based on their general philosophy about life and their risk preferences, rather than examining the relevant prior information of how well they did on the previous quiz questions (i.e. the Bayesian approach). In other words, people who choose to gamble often rationalize that decision by saying “We came in here with nothing… it’s found money… what the heck.” Those people who choose not to gamble often say something like this: “We have a sure gain of $XXX, let’s not risk losing it.”

There are also many people who get into the Cash Cab who are very familiar with the show, and therefore with the Video Bonus. I’ve never seen anyone say, “Well… most people who try the video bonus get it right, so we probably will get it right.” (Yes, more people get the video bonus right than wrong.) That would also be a relevant prior information.

I’m not suggesting that Cash Cab contestants should do any Baysian calculations to make their decision. But it’s interesting to see how they frame the decision so that most of the relevant prior information is excluded from consideration.

]]>What is disheartening is that so many people are not willing to do any additional reading or study to learn about methods outside of their previous experience or training. The resistance seems to be cultural, philosophical, and emotional.

]]>plato.stanford.edu/entries/probability-interpret/,

§3.1, classical probability depends on identifying “equally possible” cases, amongst a larger set of cases, so that the probability of the event can be identified. However, the determination of “equally possible” (which should, I think, be changed to “equally probable,” which therefore leads to circular reasoning of course) depends on subjective assessments concerning what can & cannot occur, & with what frequency or likelihood.

That is why I think that so-called “objective” probability cannot exist independently from “subjective” probability.

]]>Beware, it is written in a kind of personal style.

I hope you enjoyed as much as I enjoyed writing it.

L.

]]>