America’s standoff between the shills and the gamblers…

I’m afraid there will be no #DigitalLunch today due illness (mine) in the kitchen… So I’ve a couple short posts to share that I’ve collected over the last few days.

One of the things that’s been exercising comment in America is the dominance of number cruncher in chief Nate Silver, who’s dominance of the prediction market this time out has been frustrating some of the old schools hacks and shills.

In the Atlantic, Conor Friedersdorf points out that this came to a crescendo when Silver took on his critics by offering to bet on who was going to be right. Something that’s brought him into conflict with the NYT’s readers editor:

Predictions made by the political press are especially suspect. This is partly due to the hack and shill problem, and partly because there is no accountability for inaccurate speculative analysis.

It doesn’t matter that Bill Kristol has amassed a record of inaccurate predictions so long it is comical. He’s still treated by television hosts and fellow conservative writers as a knowledgeable speculative commentator. I don’t know how valuable speculative analysis would be in any case; but any value it has is significantly diminished by the Kristols of the world.

Is it any wonder that I sometimes fantasize about a media landscape where predictions weren’t taken seriously unless the people making them had some personal monetary stake in getting them right? How many pundits would’ve more carefully hedged their Weapons of Mass Destruction predictions?[emphasis added]

, ,

  • OK, if not 538 — which is rather more than just Nate Silver — who?

    The other lot (other than where I’m coming from) might prefer Rasmussen and the magic rolling poll: 500 automated telephone response a day, aggregated by some alchemy every three days. Margin of error: +/-3½%, which is effectively useless in the present context. Moreover, Rasmussen’s rolling poll is a nationwide scoping: we are interested in just seven States.

    Ditto Gallup, which — by general agreement — seems to have lost the plot.

    Then there’s Cillizzi at the WaPo and The Fix. I always check that one out: it is currently being roundly abused for being less Obama in perceived bias. That strikes me as a “good thing”, if one wants balance.

    The Daily Beast (though it’s mainly fisking others)?
    HuffPo (as previous, but even more so: anything between serious thought and amateur hour)?
    Ballotpedia (magnificent for background and demographic background)?
    The Cook Report (very thin — at least for non-subscribers — this cycle)?
    Iowa Electronic Markets (which feeds into Pollyvote — a personal favourite, when Google are not signposting it as virus-infected)?
    The Hill for inside-the-Beltway noos and views (and pointers to some happenings in the lesser reaches — e.g. that Mourdock in Indiana seems to be in the merde)?

    What have I missed?

    Bottom line here: is Nate Silver all-conquering? Only if one must have a quick overview, and some informed essays, all packaged neatly. But there are others which should be checked for distance.

  • Greenflag

    What have I missed?

    Not much . I can see Nate Silver being renamed Nate ‘the Oracle ‘ Silver and achieving near Zeusian status amongst the tribes of numerologiae and statisticiae as hiberno greco latinists might put it 😉 .

    But show me the money as somebody remarked ?

    On the money front here’s an interesting set of maps which show in pics what would take half a book to explain . Fascinating in particular the impact of the Superpacs .

    This animated map explores political ad spending through creative cartography. From 6 dollars a vote in Nevada to 1 cent a vote in California it’s clear that not all votes are equal or as the bould George Orwell would put it ‘Some votes are more equal than others ‘ .

    Best wishes to Mick for a speedy recovery

  • I am an ardent follower of Nate Silver I hope his prediction is correct.Like the Huff Post Pollstar,TPM, this prediction is based on aggregating up to a dozen national polls and showing that Obama has a lead of around 1% in the national poll and a slightly larger lead in the in the Electoral College.Realclear politics still had Romney ahead a few days ago but has now got the candidates level.
    I am concerned that the aggregation could give a misleading result.There are significant differences between the various polling methods and by aggregating the polls we are eliminating these differences.Most polls show Obama with a slight lead but a few particularly Gallup and Rasmussen would indicate a Romney win.They use different methods and make different assumptions.Could their methods be rightand the majority wrong.
    I have some problems with Gallup/Rasmussen in terms of their assumptions,weighting and internal consistancy but I feel it would be wrong to dismiss them out of hand.
    Polling is becoming more difficult ,varied,and the response is declining. This election will be a real test .of traditional polling methods.
    -A recent study of polls carried out by Pew Research centre showed a response rate to landline calls to be 9%and cell phones 7%and some polls have been carried out with a response rate as low as two per cent.Can we assume that the tiny minority who respond are representative of the population as a whole ?
    -Some polls call cell phones while others do not-will this lead to underrepresentation of certain groups e.g young(Democrats?)
    _Some polls distinguish between registered and likely voters.Virtually all polls show Obama with a significant
    lead among registered voters but when this is adjusted to consider only likely voters the lead is often reduced and in some cases eliminated.Gallup is particularly severe and weeds out those who express little interest in the election or who did not vote in the previous election .Perhaps Gallup is correct but such an assumption significantly reduces Democratic support.
    Other assumptions such as differential turnout among various age and ethnic groups.(The Hispanic turn out is particularly important for Democrats in Florida , Colorado and Nevada).Are automated/on line calls as accurate as live phone calls?
    Some of these questions may be answered on Tuesdayand some polls will claim sucess while others will have to review their methods.They all cannot be right.
    I do hope Nate Silver is correct and feel his analysis is convincing but given the divergence in the polls it would wrong to take anything for granted.

  • pauluk

    The result is obvious and a lot of pollsters will end up with egg all over their chin. Just ask these guys!

  • pauluk @ 12:45 am:

    I know: I’ve been ‘had’ that way before. Please assure us that you realise Michael Medved is an ironist.