jackman.stanford.edu/blog
bannerImage

Bondi to Bronte

Tuesday March 31, 2009

Filed under: general — jackman @ 11:32 pm

Sunday afternoon. Bus to Bronte. Walk to Bondi via Tama, Mackenzies and the cliff walk around to Bondi Icebergs. With Tom, aged 4 almost 5: not a bad effort by the little bloke. Lovely.

Lots of surf photos here.

Comments (1)

congratulations Achim

Wednesday March 25, 2009

Filed under: statistics — jackman @ 5:41 pm

Use R!  Indeed. Amazon lists the publication date as August 28, 2008, but I just got mine (whoops).

img-3251731-0001

Comments Off

Barracker Bama

Filed under: Australian Politics,politics — jackman @ 12:25 pm

It is odd hearing Australians pronounce the American president’s name as “Barracker Bama”.

Perhaps this locution arises here because Australians want to be Barak barrackers (though not Burke barrackers)…?

Comments (1)

Qld election polling fallout

Monday March 23, 2009

Filed under: Australian Politics — jackman @ 10:14 am

Were the polls that wrong? I think we would find that (a) they were reasonable on primary vote share, a slight over-estimate of LNP support; (b) maybe worse on 2PP. With OPV, there is a reasonable amount of uncertainty over the 17% or so of votes that have a 1st preference for someone other than a major party candidate.

Galaxy (March 17-18) had Labor 40, LNP 43; the vote count as of Monday morning is something like 42.6 to 41.4 Neither the report of the poll in the Courier Mail nor on the Galaxy site mentions the sample size. Mumble appears to have got hold of the internal report from Galaxy, which says n=800.

I suspect we would find that any individual poll was inside the nominal “margin of error” on primary shares. The interesting thing is whether a pool of the polls would show the Newspoll/Galaxy average to be off. Mumble had a weighted average of ALP 41.1 and LNP 42.4, with the combined sample size about 2,100 (Newspoll = 1,328; Galaxy = 800). Of course, the nominal sample size isn’t the effective sample size after weighting (i.e., the so-called “design effect” or DEFF); let’s say the DEFF is 1.4, and so the effective sample size is 2100/sqrt(1.4) = 1800 or so. Then the (95% confidence interval) “margin of error” the pooled estimate of the ALP on 41.1 is +/- 2.3 and easily envelopes the true result; put differently, if the world really is 42.6 to 41.4, and you use a sampling technology that delivers the equivalent of n=1800, you will get errors at least as large as the ones obtained here about 20% of the time. I’d call this far from a train wreck for the polls, at least on primary vote share.

As for biases in the polls: again, perhaps… Qld has been growing fast, a younger, slightly more mobile population, and has seen its unemployment rate take a bump in recent months (perhaps all of this adds to mobility, or making it harder to get ALP people to take phone interviews). But thats just wild speculation. It would be a good check to see if the polls underestimated Labor support in QLD in the 2007 Federal election, too?

Finally, the bookies always had Labor in front, but so too did the poll question that asked “who do you think will win?”. That number seems to track the implied probability in the betting markets pretty well, falling as low as about low to mid 50s by Election Eve. It would seem a lot of money came into the market for the LNP late in the game, helping the bookies balance out the money they’d been taking on the Labor side. Nicely played… With the benefit of hindsight, the 1.67 Election Eve price for Labor was the value bet of the campaign.

Comments (7)

Qld state election betting markets

Tuesday March 17, 2009

Filed under: Australian Politics — jackman @ 10:18 pm

This will be interesting. I haven’t been following this very closely, but my sense is that (a) the polls have a close result looming in Qld; (b) the betting markets have been consistently pointing to a Labor win.

I just looked at Centrebet. Labor at 1.42 to the LNP’s 2.80, and thus the implied probability of a Labor win is (1/1.42)/(1/1.42 + 1/2.80) = .66. It seems that even if recent poll results (51% LNP 2PP) are right, that won’t be enough to get the LNP a majority of seats in the Legislative Assembly.

To put the betting markets in some perspective, Centrebet currently has Federal Labor at 1.37 to win the next election, to the Coalition’s 2.90, with the implied probability of a Labor win at .68; i.e., virtually identical with what the market is saying in Qld.

Interestingly, the last Galaxy poll finds that 33% of respondents think that the LNP will win, which is equivalent to the implied probability of a LNP win in the Centrebet prices.

Comments (7)

W visits Harold Stanley

Sunday March 15, 2009

Filed under: politics — jackman @ 10:42 am

One for the Rochester alums: George W. Bush dropped by Harold Stanley’s Intro to American Gov class at SMU (report via Politico; a shot from inside the classroom is on the SMU site).

 Newsinfo Stories Images Bush-Smu-Visit-24Feb2009B

Comments (2)

fuel dumping by Philippines airlines 747

Monday March 9, 2009

Filed under: flight nerdery — jackman @ 5:45 pm

As all flight nerds know, United airlines lets you listen to air traffic control on their in-plane audio. On the trans-Pacific flights this is silent for all but the first and last 30 minutes of the flight (once we’re beyond the reach of terrestrial radar cover ATC is conducted via a satellite data link with the old-fashioned HF radio serving as a backup).

Anyway, I was listening in outbound from SFO on UAL863 and caught a recently departed Philippines 747 which couldn’t get its landing gear up after takeoff and was dumping fuel over the Pacific for at least 20 mins before it was light enough to attempt a landing back at SFO.

This kind of thing happens all the time, I suppose, but I’ve never heard it happening live on ATC.

I don’t know if this triggers an emergency landing at SFO, or even it would make the news.

My other thoughts were that (a) environmental risks, they are dropping a lot of unburned Jet A 10-30 miles off the coast; (b) cost — that plane would have been carrying a lot of fuel for the flight to Manilla, and the airline is on the hook for that, plus hotels for passengers back at SFO, I suppose…

Speaking of cost and airline industry stuff: tonight’s UAL flight to Sydney has a very light load. I got a nice upstairs seat with no one next to me, no problem (that has never happened before) and the load is “very very light” according to the United people. When I was flying SYD-SFO about 10 or 11 days ago it was V-Australia’s debut flight. It is an interesting time for a bit of extra competition on these routes (heading out of summer in Australia, recession biting both countries…).

Comments (2)

Exit the King

Sunday March 8, 2009

Filed under: Australian Politics,general — jackman @ 3:35 pm

Lots of Australian accents in the audience last night at Exit the King, which had its first “preview” Saturday night at the Barrymore Theater.

I was able to snag a seat on the inner aisle, 3rd row, orchestra seats, about 2hrs before curtain; it was tough to walk past this opportunity on a Saturday night in NYC. A big tall guy next to me on the way out turned out to be Mr Sarandon aka Tim Robbins.

Early March, 16 degrees Celsius, daylight savings starts overnight, and they’re still ice-skating at Rockefeller Plaza…

Img 1594

Img 1595

Comments Off

Econometricians and the GFC

Saturday March 7, 2009

Filed under: Australian Politics,politics,statistics — jackman @ 3:58 pm

Not bad, from the ABC’s Clarke and Dawe. Includes a nice definition of econometrics.

Comments (1)

Melbourne fire photo

Wednesday March 4, 2009

Filed under: Australian Politics,general — jackman @ 5:59 pm

My brother found this, he lives in Melbourne. He’s not sure of the provenance of this, but it is quite striking (thumbnail below).

Photo Taken From The#479168

Comments (3)
Next Page »

Powered by WordPress

Bad Behavior has blocked 5645 access attempts in the last 7 days.