jackman.stanford.edu/blog
• 113th U.S. Senate
• ideal point estimates pdf csv 7/31/14
• scatterplot against 2012 Obama vote share pdf
• roll call object: RData
• 113th U.S. House
• ideal point estimates svg pdf csv 8/4/14
• scatterplot vs Obama vote share svg pdf
• roll call object: RData
• Bayesian Analysis for the Social Sciences Wiley; Amazon; errata as of 5/23/13

## Wednesday October 31, 2007

Filed under: general — jackman @ 2:18 pm

About 10mins ago, a decent shake here, about 5-10 seconds in duration. Kids were in the bath, I was in the next room. Scary. USGS preliminary estimate is a 5.6 over in the East Bay, not far from here at all. Less than 20 miles.

### the ABC has a bigger budget than you

Filed under: Australian Politics — jackman @ 8:31 am

Nice report on Bennelong on last night’s 7.30 Report; following up on a good story re Eden-Monaro last night. Nothing particularly quant in either story, good color pieces. But I did ooh-ahh at this animation showing Bennelong changing over successive electoral redistributions; grabbed from about 4.35 into the almost 10 minute piece, thumbnails below.

Kudos 7.30, Ben, Clay and Kwokie (as Janet still refers to you), the graphics department, et alia.

### Scary Candidates for Halloween

Filed under: general — jackman @ 3:44 am

At least that’s what a new, Halloween-themed Associated Press/Ipsos Public Affairs poll seems to suggest. When asked to choose which ’08 hopeful would make the most frightening Halloween costume, a 37-percent plurality chose Clinton, who left second-place Rudy Giuliani (14 percent) in the dust. All the other candidates scared up only single-digit percentages, with John Edwards bringing up the rear at just 2 percent, which just goes to show there’s nothing scary about great hair.

## Tuesday October 30, 2007

Filed under: Australian Politics — jackman @ 12:41 pm

I blogged on today’s Newspoll over at the Bullring.

Market reactions have been interesting. The national headline market has seen a little money come in for the Coalition on the back on the poll, shaving the implied probability of a Labor win down to .700 from .715 (today noon, versus yesterday noon).

But the story of the day in the betting markets is some movement in Coalition seats on once-were-safe margins of 5 to 10 percent: e.g., Robertson, Sturt, Paterson, Bowman and Page all moved towards Labor today. Some summaries appear here. I’ve been wondering if and when this might happen…I’d say some of that seat-specific polling (both public and private) is starting to bite.

Roberston, with some big movement at portlandbet, and in fact there is sufficient variation in the prices there across houses so as to open up a tiny arbitrage opportunity:

+-------------+------------+------+------+---------+---------+
| agency      | date       | ALP  | Coal | Other   | ALPprob |
+-------------+------------+------+------+---------+---------+
| centrebet   | 2007-10-29 | 1.77 | 1.95 | 158.246 |  0.5211 |
| centrebet   | 2007-10-30 | 1.77 | 1.95 | 158.246 |  0.5211 |
| portlandbet | 2007-10-29 | 2.15 | 1.63 |     141 |  0.4284 |
| portlandbet | 2007-10-30 | 1.75 | 1.97 |     141 |  0.5261 |
| sportingbet | 2007-10-29 | 2.35 |  1.6 |      26 |  0.3908 |
| sportingbet | 2007-10-30 |  2.1 | 1.67 |      26 |  0.4277 |
+-------------+------------+------+------+---------+---------+


Sturt, with some hefty movement at both portlandbet and sportingbet, getting prices in line with centrebet:

+-------------+------------+------+------+---------+---------+
| agency      | date       | ALP  | Coal | Other   | ALPprob |
+-------------+------------+------+------+---------+---------+
| centrebet   | 2007-10-29 |  2.3 | 1.55 | 100.679 |  0.3989 |
| centrebet   | 2007-10-30 |  2.3 | 1.55 | 100.679 |  0.3989 |
| portlandbet | 2007-10-29 | 2.65 | 1.43 |     251 |  0.3492 |
| portlandbet | 2007-10-30 |  2.3 | 1.54 |     251 |  0.3996 |
| sportingbet | 2007-10-29 | 2.75 | 1.42 |      26 |  0.3287 |
| sportingbet | 2007-10-30 | 2.35 | 1.55 |      26 |  0.3837 |
+-------------+------------+------+------+---------+---------+


Page, with again, portlandbet the big mover:

+-------------+------------+------+------+---------+---------+
| agency      | date       | ALP  | Coal | Other   | ALPprob |
+-------------+------------+------+------+---------+---------+
| centrebet   | 2007-10-29 |    2 | 1.72 | 130.924 |  0.4591 |
| centrebet   | 2007-10-30 |    2 | 1.72 | 130.924 |  0.4591 |
| portlandbet | 2007-10-29 | 2.35 | 1.54 |     101 |  0.3923 |
| portlandbet | 2007-10-30 |    2 | 1.75 |     101 |  0.4624 |
| sportingbet | 2007-10-29 |  2.5 |  1.5 |      26 |  0.3619 |
| sportingbet | 2007-10-30 |  2.5 |  1.5 |      26 |  0.3619 |
+-------------+------------+------+------+---------+---------+


## Monday October 29, 2007

Filed under: Australian Politics — jackman @ 4:53 pm

I haven’t posted about arbitrage in the betting markets for a while. But I got an e-mail over the weekend from a reader which prompted me to start paying a little more attention.

My arbitrage tables only look for on-the-spot, cross-bookie underrounds (overrounds less than 1.0) between Coalition and Labor prices. I screen out seats where Independents are doing well (e.g., Kennedy), but haven’t screened out Forrest or Corio or Lyons, where for various reasons (!) Independents and minor party candidates are thought to have a real shot. I should fix my code to screen out these seats too.

Forrest is particularly interesting: the independent candidate (local newsreader Noel Brunning) is on 3.75 at Portlandbet and 4.00 at Centrebet, certainly still an outsider, but a race to watch on the night. In Lyons, if you ignore the prospect of the former Lib Ben Quin winning the seat, then there is an 8% profit opportunity there. But at this point you’re gambling, which is not what arbitrage profits are supposed to be about (the same is true everywhere a minor party or independent candidate runs, strictly speaking). Lyons will also be an interesting one to watch on the night, but the punters have currently got Labor at about a .75 to .79 chance of retaining the seat.

But several small but “real” arb possibilities have opened up in the last couple of days, as money has started to come in for Labor, but unevenly across bookies. For example:

1. Robertson (NSW): Centrebet has 1.95 for the Coalition, but you can get 2.35 for Labor at sportingbet, an arb profit of 6.6%
2. Makin (SA): Centrebet has the Coalition on 4.95, and sportingbet has Labor on 1.30), a 3% opportunity.
3. Bennelong (NSW): Centrebet has John Howard on 1.57 in Bennelong, but sportingbet has Labor on 2.85, a tiny 1.23% arb profit there.

### new Oz Politics sites

Filed under: Australian Politics — jackman @ 3:20 pm

At the request of colleagues and correspondents, here are two new Australian politics sites that are up and running:

1. A new site dedicated to political analysis of Tasmanian politics http://tasmanianpolitics.blogspot.com
2. An archive of Australian political advertising at the Univ of Melbourne. For a while this was amazing collection was simply a large collection of files on Sally Young’s laptop computer, but has now got a web presence at Soapbox. Congrats to Sally and her colleagues at Melbourne (Stephanie Youane and Mary Helen McIlroy) on getting this valuable resource up and running.

## Sunday October 28, 2007

Filed under: Australian Politics,statistics — jackman @ 5:51 am

On the Roy Morgan research website there are often these interesting remarks from Gary Morgan about methodology. For instance, in a report of two polls (one face-to-face, the other phone):

For the academics: Telephone polls have two inherent biases:
1. The sample (by design) only includes those who have telephones (approximately 1% bias toward L-NP).
2. The sample achieved has a much lower response rate than face-to-face interviews (approximately twice as many people refuse to answer as refuse face-to-face). The ‘bias’ caused by this rate is less ‘tangible’ and ‘predictable’. Historically, Roy Morgan Research has evidence that supporters of the party that is ‘out of favour’ tend to be over represented among those who ‘refuse’ to be interviewed. However, a compelling event or news story can generate a desire for people to ‘have their say’ — so create its own bias.

Lets ignore the “oh, if you must” sentiment underlying the phrase “For the academics”: the issues here aren’t moot, or solely of interest to the Scholastics.

Surely post-stratification weighting can take care of these likely sources of bias? That is, once I weight on, say, age/region/gender/education, then I’d tend to think differences in sample composition (phone vs face-to-face) can be wiped out (i.e., after you tell me your age, region, gender, education, landline vs mobile-only has no predictive power for your vote intention, meaning that conditional on those demographic characteristic, the phone/mobile thing is said to be ignorable). And that kind of post-stratification might also wipe out any of the non-response bias GM is alluding to in his point #2.

Design effects. Buried at the bottom of Morgan’s reports you’ll also routinely see the following:

The margin of error to be allowed for in any estimate depends mainly on the number of interviews on which it is based. The following table gives indications of the likely range within which estimates would be 95% likely to fall, expressed as the number of percentage points above or below the actual estimate. The figures are approximate and for general guidance only, and assume a simple random sample. Allowance for design effects (such as stratification and weighting) should be made as appropriate.

To Morgan’s credit, this reference to design effects is one of the only places I’ve seen a non-academic pollster refer to the fact that the standard calculations of marginals of error are based on the assumption of simple random sampling (which is almost never true for polling in the real world); the actual confidence intervals are wider, depending on how hard one is weighting the data (see my earlier crack at this here).

What would be nice is some assessment of just how big and variable the weights are, such that the knowledgable poll consumer might actually be able to compute the right confidence intervals. Right now, the reference is design effects is kind of like a “your mileage may vary” disclaimer you get from a car-maker.

## Friday October 26, 2007

Filed under: Australian Politics — jackman @ 6:35 pm

Peter (Mumble) Brent and I worked on the enrolments issue earlier in the year. We’ve been watching the enrolment story since then, over the course of the AEC’s ad campaign, and of course now into the campaign. Peter emailed me to say that the AEC released its close of rolls numbers this afternoon (AEDT; its of course past 1am here in California).

Peter was speaking to a SMH journalist who may have more on this over the weekend [update: see here]. But the bottom line is a roll of 13,645,073. See the updated graph, thumbnail below. More details over the fold.

### Possum and Me in the Courier

Filed under: Australian Politics — jackman @ 1:55 pm

## Wednesday October 24, 2007

Filed under: flight nerdery — jackman @ 4:07 am

This is back in the news with the release of a report from Indonesian officials on the crash: pilot error is the official cause.  The accident report at the link above is pretty compelling.

At 10.1 miles from runway,  altitude was 3,927 feet,  the approach calls for 2,500ft.  Airspeed was 283 knots.  I’m not a pilot, but I know enough to know that below 10,000 ft it is rare to be above 250 knots, and that if you are high and fast, you’ve got to shed that energy somehow (speed brake, throttles back, wait until the airspeed gets down to a speed where flaps can be deployed, request a 360 degree turn to rejoin the pattern at a slower/safer speed, something).  Instead, this cowboy put the nose down and acquired more airspeed, and the flaps never got extended past 5 degrees due to the speed that the aircraft was carrying.

The aircraft crossed the runway threshold at 89 feet above the runway at 232 knots.  The required landing speed was 135 knots in a flaps 40 configuration.  At touchdown the aircraft was traveling at 221 knots, which is amazingly fast.  When the aircraft overran the runway it was traveling at 110 knots, which is more than 200 km/h.  21 fatalities resulted from the crash and the resulting fire.

Next Page »