# Hard News by Russell Brown

Read Post

## 389 Responses

First ←Older Page 1 12 13 14 15 16 Newer→ Last

• Maybe this will help with visualizing the 92% thing as a distribution. I just noticed on the right hand side of the methodology page there is a graph called Electoral Vote Distribution. It shows the probability that Obama receives a certain number of Electoral College votes. Sorry I don’t know how to easily imbed that graphic here, you have to go and look for yourself.

There are three obvious peaks, the highest at about 20% is 330-ish (it’s such a small graph it’s hard to read). That’s your distribution. I presume it’s so lumpy because of the effect of FPP in each state (except 2). Yet not too lumpy because different states get different numbers of votes.

The to left of 270 are all the outcomes in red which elect Romney, to the right are all the blue outcomes which elect Obama. Add up all the probabilities on the red side and you would get about 8%, and on the blue side gives you about 92%. Minus something for “exact tie” and because we’ve been rounding along the way.

I am of course embarrassed to say I read all the words on the left and didn’t look at all the interesting graphics on the right. I used to be such a proponent of graphical data analysis, yet I got stuck into the words. Too many tea leaves for me.

sunny mt albert • Since Jan 2007 • 116 posts Report Reply

• BenWilson, in reply to David Hood,

So in some respects, the 92% is doind a similar job to expressing a confidence interval, but it is a different thing in the finer level of detail.

Heh, yes, it's a degree of trust that you can't trust.

Auckland • Since Nov 2006 • 10657 posts Report Reply

• BenWilson, in reply to steve black,

Add up all the probabilities on the red side and you would get about 8%, and on the blue side gives you about 92%.

Yup. That would, in fact, be how the number 92% was arrived at. Given that, how is it not accurate to say that the simulations are saying that is the actual probability of an Obama win? Whatever biases, deviations, trends, the whole lot, are captured in the simulation. Which is the point of doing the simulation.

Is this reasoning correct? I understand that problems in the underlying sims cascade into the final result. Do they do that in a way such that any ring fence you put around the underlying problem has an unknown effect in the final? If so, why even bother coming up with the final? It's turned into a much less meaningful number. A lower bound maybe, if you knew there was a deliberate Romney bias?

Auckland • Since Nov 2006 • 10657 posts Report Reply

• David Hood, in reply to BenWilson,

I understand that problems in the underlying sims cascade into the final result.

As I understand it, a greater degree of uncertainty in the poll analysis and aggregation would, in general, produce a wider spread of potential results (this might move the 270+ percentage a bit). A mistake in the model in the way that the states results interrelate would tend to produce a massive error in results.

For a long while Nate Silver was listing Obama's chance of getting 270 seats or more as smaller than the other correct polling aggregate sites (though vastly larger than the pundits). This was because 538 was making a deliberately high estimate of uncertainty (in part because he thought for the Republicans to be that confident they might know something he didn't), so his graph of simulated results was wider and flatter. As election day approached, all the polls, and all the polling analysts, converged in their predictions.

As 538 keeps portions of his model private, we can't assess if the way the factors flow together is justified, but the uncertainty in the aggregation process is reflected in the spread of the results in possible electoral college vote outcomes, and the percentage figure is an aggregation of this spread based on the 270 threshold. I would call it a valid measure of the likelihood of Obama getting more than 270 electoral votes if the modelling process is accurate. I suspect Nate Silver would say the accuracy of his model should be judged on the results (in reality some of the open academic models were as accurate, but Nate Silver is much better at communicating with a general audience).

Dunedin • Since May 2007 • 1445 posts Report Reply

• I really enjoyed media3 last night (sunday) however I saw something alarming.
There were two panel people ( not panel beaters ) discussing the election in USA. One obviously a gentle and thinking democrat.............and a guy called Carl.......both americans..............there was good banter but Carl really dissapointed me by among some good stuff............saying stuff that americans say, like what can we know ( read anyone outside america) about american politics...............and yeah Obama may be an empty suit.............but what are the other suits full of..........living in NZ now with his may I say ugly bipartisan views...........maybe he has seen what the suits may just be full of..........out on the farm somewhere. It concerns me that americans come here with such attitudes and spout them on air....................emperors new clothes................we smile and say nothing..........Carl, leave the silly stuff back where it seems happy..................

New Zealand • Since Nov 2012 • 3 posts Report Reply

• Yes - the Republican guy on Media3 was short on facts and a bit long on truthiness for my tastes.

For example, he repeated that old Republican lie about how Obama never made a single vote as a a Senator because he was far too busy preparing to run for President. Completely untrue. You can see more about Obama's voting record in the Senate here (thanks Wikipedia) - especially in the Legislation and Voting Record section.

I don't believe that someone who clearly takes such an interest in US politics and Obama in particular would make a mistake about something as fundamental as a person's voting record, therefore the only conclusion I can come to is that he was deliberately telling an untruth, with the assumption that he wouldn't be called out on it during the interview. When a person with a particular viewpoint is interviewed and flat-out lies in order to strengthen his/her position, I find it hard to take anything else they say very seriously.

I was thinking how great it would be if interviewers could press "pause" during an interview whenever someone makes a statement of "fact" which might not actually be very factual. Pressing "pause" would freeze time while you whizzed off and checked The Google to see if they were telling the truth or not. Now that would be cool.

Wellington • Since Nov 2006 • 332 posts Report Reply

• Sacha, in reply to webweaver,

Quite. The guy didn't seem to offer much credibility from the get-go.

Ak • Since May 2008 • 19740 posts Report Reply

• Ian Dalziel, in reply to webweaver,

great comebacks...

Pressing “pause” would freeze time
while you whizzed off and checked

what you need is a...
Magic Boomerang!

Christchurch • Since Dec 2006 • 7950 posts Report Reply

• Hebe, in reply to Islander,

I wish there was a function wherebye I could put tiny wee pointed teeth inbetween my smiles )

Heh. You beastie you.

Christchurch • Since May 2011 • 2899 posts Report Reply

• ~....~
0....0
\vvvv/

Auckland • Since Nov 2006 • 10657 posts Report Reply

• Fang you very much..

I wish there was a function whereby
I could put tiny wee pointed teeth
in between my smiles

some kind of file protocol maybe?
or an 'authordontist' perhaps...
:- )

Christchurch • Since Dec 2006 • 7950 posts Report Reply

• Eye Eye Cap'n.

Something incisive?

Upper Hutt • Since Jun 2007 • 1590 posts Report Reply

• A good blog entry from Sam Wang (the Princeton poll analyst) about the hissy fit that Gallup have thrown around having their polling accuracy compared.

Dunedin • Since May 2007 • 1445 posts Report Reply

• Craig Ranapia, in reply to webweaver,

For example, he repeated that old Republican lie about how Obama never made a single vote as a a Senator because he was far too busy preparing to run for President. Completely untrue.

Now, you could fairly make the point that due to the nature of the primaries Obama did miss more votes in the 110th Congress that I'd like. 303 out of 657 recorded roll call votes. But you'd also have to note that was more than Clinton (303) and significantly less than McCain (420). Or would that take some work?

North Shore, Auckland • Since Nov 2006 • 12370 posts Report Reply

First ←Older Page 1 12 13 14 15 16 Newer→ Last

## Post your response…

Please sign in using your Public Address credentials…

Login

You may also create an account or retrieve your password.