jackman.stanford.edu/blog
• Bayesian Analysis for the Social Sciences Wiley; Amazon; errata as of 4/16/13
• 113th U.S. Senate
• ideal point estimates pdf csv 5/15/13
• scatterplot against 2012 Obama vote share pdf
• roll call object: RData
• 113th U.S. House
• ideal point estimates pdf csv 5/17/13
• scatterplot vs Obama vote share pdf svg
• roll call object: RData
• 2013 Australian Federal Election, betting market summary
AgencyALPCoalitionProb ALP Win (%)Last 7 days
Centrebet8.501.0611.1
sportsbet8.001.0511.6
Tom Waterhouse8.001.0511.6

## Thursday May 16, 2013

Filed under: general — jackman @ 5:55 am

…from TEDx Sydney, got picked up by the TEDx global site.

Thanks to everyone at TEDx Sydney for letting me be part of an amazing event. The organization of the event was terrific, the production value and support absolutely sumptuous (the Sydney Opera House!).

Edwina Throsby (my producer) and I had been bouncing ideas and drafts around for a couple of months; Gretel Killeen joined Edwina in knocking the talk into shape during rehearsals.

My thanks to David Broockman and Chris Skovron for letting me use their same-sex marriage example.

I also need to thank Margaret Stewart (herself a “big” TED veteran) (and check this out) and family, from whom I appropriated the “It’s Your Duty To Hack” sign appearing at the end of the talk.

And this is too beautiful not to link to (bump it up to HD, full-screen etc):

## Friday May 10, 2013

Filed under: statistics — jackman @ 4:25 pm

I just spent a couple of hours at the meeting of the West Coast Experiments group, a set of political scientists interested in using experiments.

One the speakers was talking about the need for “credible” or “honest” p-values. “This will be good”, I thought to myself…

What the speaker was alluding to are the current moves afoot in political science to help stamp out “fishing” for statistically significant results, including pre-registering research plans. The problem is that after you’ve looked at a data set multiple times, the p-values aren’t telling what you think they are. The problem – as some Bayesians would point out – is that a p-value isn’t ever what you’d like it to be, even when you’re looking at the data the 1st time…

From the Bayesian perspective, all this stuff is kind of ridiculously overblown, a consequence of an unthinking acceptance of $$p < .05$$ as a model for scientific decision-making, point null hypothesis testing, the whole box and dice. That is worth a separate post one day.

For now, I'll remark that pre-registration of research plans is a bit like eliciting very crude priors: i.e., enumerating things to be looked at in the analysis, because the effects aren't thought to be zero; enumerating things that won't be looked at, because prior beliefs over the effects are concentrated close to zero.

The best moment of Bayesian irony was when the speaker emphasized that the need for honest p-values is especially pressing in situations where the experiment is expensive or intrusive and therefore unlikely to be run very often. This was just awesome, when you think what about a p-value is supposed to measure.

More generally, its been very interesting to bring a Bayesian perspective to my teaching about experimental design and analysis, or to a meeting like the one I was at today.

To begin with, try this on: the role of randomization in Bayesian inference. At least as a formal matter, randomization plays no formal role in the Bayesian analysis of a data from an experiment, or any other data for that matter. This sounds so odd to non-Bayesians at first, particularly people who are doing a lot of experiments. But recall that repeated sampling properties like unbiasedness just aren’t the 1st or 2nd or even 3rd thing you consider in the Bayesian approach.

So just is the value of randomization to a Bayesian? Surely not zero, right? Don Rubin has written a little on this; Rubin’s point – that randomization limits the sensitivity of a Bayesian analysis to modeling assumptions – is a stronger conclusion than it first sounds, and one more of the more helpful things I’ve come across on the topic. I also found this note by J. Ghosh, a very concise and accessible summary of the issues too, summarizing some of the Bayesian thinking on the matter (Savage, Kadane, Berry etc). But my sense is that there’s not a lot out there on this. There is actually more writing on this in the literature putting model-based inference up against design-based inference in the sampling literature, which is essentially a parallel debate.

So, vast chunks of the (overwhelmingly classical/frequentist) literature on the analysis of experiments can seem very odd to a Bayesian. Randomization inference, or permutation tests. Re-randomization of assignment status if one detects imbalance. Virtually all Bayesians take the Likelihood Principle seriously, but so much of the work on experiments seems to violate it. It is also pretty obvious that experimenters are also carrying around prior information and using it: balance checks would seem to be guided by prior expectations as to likely confounders, no? Just in the same way that post-stratification weighting for non-response in a survey setting seems to be guided by an (implicit, and rather simplistic) model of response/non-response.

There is a lot to work through. Above all, it is important to keep in mind what is relevant for the applied scientist, what is more esoteric, and where Bayesian ideas can be of real practical use (e.g., Andy Gelman et al on hierarchical models for multiple comparison problems, or in the analysis of blocked or clustered designs, etc).

For now, I’m blessed to have colleagues like Persi Diaconis, Guido Imbens and Doug Rivers, who indulge (or encourage) my thinking out loud on these matters.

## Wednesday May 1, 2013

Filed under: Australian Politics — jackman @ 6:19 pm

I’m in Sydney this week prepping for TEDxSydney this weekend, which should be a blast. I’m talking about Politics and the Data Revolution, which will take in a review of some of the ways that the Data Revolution is reshaping political science research — and in particular — making that research incredibly relevant to real-world politics and policy making.

Its a very different kind of talk to lecturing, workshops, or even general audience talks. I’m learning a lot about those other, more conventional modes of giving talks from the prep I’m having to put into being comfortable with the TED format. We’ll see how it goes…

The Guardian Australia announced that I’ll be helping out with their election coverage ahead of the election here in September. It looks like a great group of people they’ve assembled (only one or two I knew about until today’s announcement).

And it seems only a month ago that I blogged about Labor’s price breaking new records for long-odds in the Australian political betting markets, at 7.30 to 1.10 on Centrebet on March 29. Think again.

Its May 2 and Labor’s out to 8.80 at Centrebet, the Coalition in to 1.05. The implied probability of a Labor win is 10.7%. Tom Waterhouse has the Coalition at 1.10.
Do ya best.

And, one of the best parts of a week in Sydney is that I usually treat myself to a workspace with a view like this:

Not too shabby.

## Thursday April 25, 2013

Filed under: general — jackman @ 10:18 am

At the risk of being a little self-serving…

7 political scientists were elected to the American Academy of Arts and Sciences this year. 4 of them (er, us) got their degrees from Rochester in the early to mid 1990s: Daniel, John, me and Alastair.

And 3 of the 4 Rochester people were born/raised outside of the US.

Oh, Meliora… And congratulations.

## Monday April 8, 2013

Filed under: politics — jackman @ 5:58 pm

After a bit of a hiatus, I’ve got the 113th U.S. Senate ideal points up and running. Links to deliverables appear above, in the blog header. It is interesting to ask where the new faces line up.

Quelle surprise, there is no partisan overlap in the estimated ideal points. Manchin is the most conservative Democrat, followed by McCaskill; Collins the most liberal Republican, followed by Murkowski.

I was a little surprised to see Flake (R-AZ) and Coburn (R-OK) not out in the extreme of the Republican ideal points, but maybe thats because I’ve been traumatized by their assaults on political science.

Warren’s (D-MA) voting history places her just a little to the left of the median Democratic senator, right next to Richard Blumenthal (CT), Tammy Baldwin (WI) and Ben Cardin (MD).

Boxer is a little to the left of Feinstein, but we’re estimating those ideal points rather imprecisely (we’re not seeing a lot of roll calls that split the Democrats).

PNG version of the distribution of ideal points:

I’ve also plotted ideal points against Obama vote share in the state in the 2012 presidential election:

The separation by party is (as usual) the most compelling feature of the data. As she was in the 112th, Murkowski is decidedly more liberal than we’d expect from a Republican senator representing a state in which Obama got 40% of the vote.

Durbin and Menendez are the 2 biggest “more liberal than expected” residuals on the Democratic side; Kerry and the two DE senators (Carper and Coons) are substantially more conservative than we’d expect, given that Obama won 60% of the vote in their respective states. In Kerry’s case we’ve got a truncated voting history given his elevation to SecState, but it will be interesting to see where the DE senators wind up.

## Monday April 1, 2013

Filed under: politics,statistics — jackman @ 12:18 pm

“I needed to learn how to be persuasive. I needed to learn how to win arguments. And so I did two things.”

“I took a ton of statistics classes…”

“And I enrolled in the Ethics in Society Honors program.”

## Saturday March 30, 2013

Filed under: Australian Politics — jackman @ 6:40 am

Labor’s price is out past $7.00, the Coalition in to under$1.10. The implied probability of a Labor win is under 13%, once you factor out the bookies’ profit margins.

In the 6 years I’ve been watching the national-level betting markets, I think this is as lop-sided a market as I’ve seen. Some state-level markets have exceeded this, from memory (I haven’t been logging those prices), but I think I’m correct in asserting \$7.00+ prices are new territory for the national betting markets.

The 10 highest Labor prices I’ve recorded:

The 10 highest Coalition prices aren’t as lop-sided, and almost all date back to the 2007 election:

July 2011 to the present, implied probability of a Labor win, based on the prices offered by some of the better known betting agencies:

## Wednesday March 20, 2013

Filed under: ANES,politics,statistics — jackman @ 12:17 pm

The Senate just adopted Coburn’s (amended) amendment:

To prohibit the use of funds to carry out the functions of the Political Science Program in the Division of Social and Economic Sciences of the Directorate for Social, Behavioral, and Economic Sciences of the National Science Foundation, except for research projects that the Director of the National Science Foundation certifies as promoting national security or the economic interests of the United States.

Barbara Mikulski accepted the terms of the amendment, let it go through on the voices.

Nicely wedged, nicely played.

A great day for IR. A bad day for the study of American politics, political methodology…etc.

The “national security” or “economic interests” tests will be an interesting thing to see play out.

Or maybe we’re all sociologists now?

## Tuesday March 19, 2013

Filed under: Australian Politics — jackman @ 10:32 pm

A look at Sportbet’s market on the Labor leadership, at least for the four leading contenders…

### why are Labor’s odds improving at Centrebet?

Filed under: Australian Politics — jackman @ 7:29 pm

There is nothing like leadership chatter to help draw some punters back to the Federal election betting market…

In particular, Labor’s price has been improving a little bit over the last 72 hours at Centrebet. At the time of writing, Labor is at 4.80 to the Coalition’s 1.17 (implied probability of ALP winning = 19.6%), in from 5.60/1.13 (16.8%) as recently as 5pm Monday March 18:

Labor had been out on 5.90 to the Coalition’s 1.12 as recently as March 12.

At sportsbet, the Labor leader market has Gillard out to 2.50, Rudd at 1.50. Factoring in the other Labor leader possibilities, this equates to a 44% chance that Rudd is leader at the next election. As recently as March 9, Rudd was at 2.35 (30% probability) and Gillard at 1.60 (44% probability).