« Come hear my talk next Tuesday | Main | Sometimes you expose the holes »



I'me not certain how they got their estimates, they reference a paper by Poole, but it isn't in the bibliography. What you get for using Word. Simon Jackman for voting intentions in the US government using data pooled over time uses a latent variable method, essentially item response theory (IRT), and I assume they use this or something that is effectively the same.

The problem with it, is that you need items spread across the range of difficulty, otherwise results can be dominated by a small number of items. In this case given that only 3 nations have voted in favour of the US action on Cuba this makes it a very difficult item. So getting it "correct", unless there are items of similar difficulty, would virtually define the value of the latent variable, which is what the right axis represents. Some sort of sensitivity analysis needs to be done without these unusual items. It is quite probable that they don't fit the IRT model anyway.

The comments to this entry are closed.

Kaiser Fung. Business analytics and data visualization expert. Author and Speaker.
Visit my website. Follow my Twitter. See my articles at Daily Beast, 538, HBR.

See my Youtube and Flickr.

Book Blog

Link to junkcharts

Graphics design by Amanda Lee

The Read

Keep in Touch

follow me on Twitter