political analysis

Just to make things clear, most political reporters are morons, nearly as bad as sports reporters. Mostly ugly cheerleaders for their side, rather than analysts. Uninteresting.

how to analyze polls:

Who ever is ahead in the polls at the time of election is extremely likely to win. Talk about how Candidate X would have a ‘difficult path to 270 electoral votes’ when he’s up 2 points (for example), is pretty much horseshit. There are second-order considerations: you get more oomph per voter when the voter is in a small state, and you also want your votes distributed fairly evenly, so that you win states giving you a majority of electoral votes by a little rather than winning states giving you a minority of electoral votes by huge margins. Not that a candidate can do much about this, of course.

When you hear someone say that it’s really 50 state contests [ more if you think about Maine and Nebraska] , so you should pay attention to the state polls, not the national polls: also horseshit. In some sense, it is true – but when your national polls go up, so do your state polls – almost all of them, in practice. On election day, or just before, you want to consider national polls rather than state polls, because they are almost always more recent, therefore more accurate.

When should you trust an outlier poll, rather than the average: when you want to be wrong.

Money doesn’t help much. Political consultants will tell you that it does, but then they get 15% of ad buys.

A decent political reporter would actually go out and talk to people that aren’t exactly like him. Apparently this no longer happens.

All of these rules have exceptions – but if you understand those [rare] exceptions and can apply them, you’re paying too much attention to politics.

Advertisements
This entry was posted in Uncategorized. Bookmark the permalink.

55 Responses to political analysis

  1. dearieme says:

    “Who ever is ahead in the polls at the time of election is extremely likely to win.” Not in the British General Election in 2015, nor in the Brexit referendum in 2016. American exceptionalism?

  2. pyrrhus says:

    Polls, like all statistical measures, require a representative sample. And a reasonably correct notion of turnout for the various segments of the population. That’s become increasingly difficult to obtain, with many voters having only cellphones, widespread disinterest in answering polls, and candidates challenging the 2 party status quo. Furthermore, when people know that they might be fired for supporting the “wrong” candidate, they are not going to respond….truthfully. That’s how Bernie was 22 points behind in MN the day before the election and won….My wife, an expert in this area, comments only that this would be an extremely expensive election to poll accurately….

    • dearieme says:

      There may also be a shortage of honesty within the polling companies. Who knows?

      • gcochran9 says:

        Look at the track record and see.

      • The poll results only get compared to reality when the election occurs. During the campaign, pollsters can safely bias their results to promote a candidate, e.g., by fiddling with the likely voter guesses. I think some do that. When different polls produce conflicting results — which has been common so far — they can’t all be accurate.

        • gcochran9 says:

          People who don’t like the answer often talk about rigged polls. It could happen. I’ve never seen a case that looked very likely to me, in terms of voting for X or Y. In terms of distorting a question in an attempt to elicit the desired answer, common.

          Different polls use different turnout models.

    • Pincher Martin says:

      That’s how Bernie was 22 points behind in MN the day before the election and won….

      MI (Michigan), not MN (Minnesota).

      There’s a greater chance of these large deviations from polling averages in the state elections because those races are polled far less frequently and by far fewer polling organizations.

      But Michigan was a whopper by any standard. Nate Silver described it as “among the greatest polling errors in primary history.” Hillary was up by more than 21 points in the polling average and lost by 1.5 points.

      The U.S. presidential election, however, is polled more often than any race in the world and by nearly two dozen polling organizations, some of which occasionally use different methodologies. Under those circumstances, I don’t think that kind of unexpected result that we saw in Michigan is possible.

      • Pincher Martin says:

        And of course that’s all the more reason to not pay as much attention to the state polls as the national polling.

      • stalin says:

        Abbreviating the “M” states is problematic and not intuitive
        Maine ME.
        Massachusetts MA.
        Maryland MD.
        Mississippi MS.
        Missouri MO.
        Michigan MI.
        Minnesota MN.
        Montana MT.

      • pithom says:

        I live in MI and naturally thought it would go to Sanders during the primary, since the state didn’t seem to be Hillary country, and the people were desiring a cleaner, fresher politician. There’s a reasonably good correlation between Dwight Eisenhower’s general election and Sanders’s primary maps. Then, I saw the polls and thought Her would win, though not anywhere by the amount the polls said. Then Sanders won anyway. The polls would have been accurate for MI’s most populated county. But MI isn’t 50% Black.

        Polling was less reliable in this year’s Democratic primary than a good demographic model.

        • Pincher Martin says:

          There’s a reasonably good correlation between Dwight Eisenhower’s general election and Sanders’s primary maps.

          I don’t think that’s true. Clinton won many states in the West, New England, and mid-Atlantic. Adlai Stevenson lost pretty much everywhere except the south. He even lost Illinois, his home state.

  3. spottedtoad says:

    I’d agree overall, though there is a difference between the national polling margin and the margin that would deliver an electoral college win; Sam Wang’s meta-margin (which seems to show the latter reasonably accurately http://election.princeton.edu/todays-electoral-vote-histogram/ ) differs in a consistent direction from the national poll average from RCP or Pollster, etc.

    One of the interesting questions that this election will shed light on is how much Get Out the Vote and state-level organization make a difference; I thought they would make a difference in the primaries and they didn’t, but they might well in the general.

    • albatross says:

      Looking at the electoral college map tells you a lot more about how plausible various things are (like whether candidate X can come back from his deficit against candidate Y) than just looking at popular votes. But it’s actually very rare for the electoral college vote to come out differently than the popular vote, so most of the time, the national popular vote numbers should give you a reasonably good prediction.

      I suspect this year’s election is much less predictable than previous years’ elections, because Trump is a really unusual candidate, and the election up until now has looked very different from most other presidential elections.

  4. ohwilleke says:

    State polls are absolutely more accurate. They directly measure the outcome in question and a set of battleground state polls are based on a much larger and much more likely to be representative sample size. Averages of polls in all cases are better. Yes, still polls aren’t independent of each other and vary with national trends, but their variation is not so strongly a function of national polling (which is inherently more error prone) that national polls are most trustworthy. Look at the technical notes to the 538 polling model for more details.

  5. Greying Wanderer says:

    TV news isn’t news, it’s people herding.

  6. Tim says:

    The only exception I wonder about – and this is probably wishful thinking – is when a candidate truly gets new voters to turn out that the pollsters haven’t factored in.

    For example, the Brexit vote.

    • whyteablog says:

      Republican primary turnout was way bigger than in previous years, and our boy Candidate X pretty consistently outdid the polls. He also did better in “open primary” states where non-Republicans could vote in the Republican primary. Hm…

      By most indicators, he’s likely enough to win but doesn’t totally have it in the bag. But I’m wondering if the monster vote might come in out of nowhere, bigger than I expect, slamming Hillary into the dustbin of irrelevance. Perhaps wishful thinking.

  7. HelenaHandbasket says:

    And there is likely to be what Scott Adams calls a Shy Trump effect. There are plenty of noisy bombastic trumpeters (as you’d expect). But there are also a lot of people who just want things to change and know (but feel embarassed to say even to pollsters) that they dont much care how this happens. This also occured with Brexit

    • albatross says:

      Maybe. Both candidates also are widely disliked by voters from their own parties, so you can imagine all kinds of weird rarely-seen things happening in this election–lots of Clinton or Trump voters staying home (screwing over all the down-ticket members of their party) or voting third-party, for example. And Trump seems to appeal to a group that usually isn’t as involved in politics, so he might get voters who would show up in the pollsters’ models as unlikely to vote.

  8. albatross says:

    Does money matter in other races, like for congress? I keep reading about the obscene amounts of money being spent on races for congress or state governor, but I don’t really know how well it tracks. One obvious confounding variable here: if you’re going to be governor, I’d probably like to have contributed to your campaign to be on your good side. So you can see how you could have a positive correlation between dollars spent and outcome, without having any causation at all.

    • Thagomizer says:

      On a political campaign money is communication. Political strategists talk about earned media (the press covering you for free) and paid media.

      If the press doesn’t care and you have no money nobody hears about you. If you can get press coverage for free then you can get your message out with less money.

      So money matters more in smaller races and primaries. The candidate with more money usually wins nomination races.

      Things get more interesting with republicans supporting amnesty because they usually have much more money than popularity. Eric Cantor spent $5,000,000 vs $200,000 and lost. Paul Ryan spent $7,000,000 vs $700,000 and won (by a large margin).

      For perspective, Trump spent $30,000,000 nationally in August. Those were house nominations.

  9. cold anagram says:

    Greg I’ve noticed you’re not a fan of the neoreactionaries, but I was wondering what your opinions on democracy are. Surely knowing what we know about population genetics we can’t take democracy as a viable political system, can we?

    The reactionaries may be crazy, as we also know that being above average in intelligence doesn’t stop one from taking all kinds of bullshit seriously (Marxism, psychoanalysis, etc.). But maybe they aren’t wrong about democracy being crap. Gauss was a staunch supporter of the monarchy after all.

    • ziel says:

      Well I’m not Greg obviously, but it seems really stupid/pointless. First of all, it’s not like there aren’t crazy/stupid/ineffective kings. How did Kaiser Bill work out? Czar Nicholas II? Did the killing of Archduke Franz-Ferdinand cause any problems? And that was just in the 20th century – not even considering all the other little mess-ups throughout history like the Hundred Years War.

      But even if we wanted to democratize, how would we go about doing it? You’re going to remove people’s voting rights? We can’t even implement voter ID’s – how are we going to actually directly disenfranchise adult citizens?

      And if we could, who are these enlightened souls who will lead us into rational policies? The country already pretty much is run by people in the rightmost 10% of the bell curve – how they doing? The State Dept, DOJ – run by smart people by any objective means of measuring intelligence – but idiots nevertheless.

      I think what these people really want is a military coup so generals who really care about the country could take over – but do they actually exist? I have my doubts that the process of selecting military leaders is any better today than any other such process we have – that whatever process got them selected to be generals had anything to do with them being intelligent, forward-thinking, patriotic Americans.

    • iffen says:

      As it turns out, liberal democracy seems to be incompatible with reality over the long term.

  10. “Who ever is ahead in the polls is extremely likely to win”

    Absolutely true and i have profited on it in the past and will profit on this in the future. A few days before the Obama/Romney election the betting markets had Obama a 4 to 1 favorite and the blog 538 had him a 10 to 1 favorite. I max bet the amount allowed at 5dimes on Obama and made 25% on my money. Not long ago between the Wisconsin primaries and the New York primaries Trump was only a slight favorite to beat Cruz for the republican nomination. I Max bet that and nearly doubled my money. Next betting opportunity will come before the November election and I am hoping that Clinton is a four point favorite so that the moron analysts proclaim the election is up for grabs and I can make a max bet on Clinton. I don’t think I will bet on Trump again, the demographics are weighed too heavily against him. If he is up in the polls I think I won’t bet on anybody.

    I won’t make a bet months ahead of time, too risky, too many unknowns. But right before the election, if it is close and there aren’t too many screwy variables, I plan on betting on the favorite in national elections. As Cochran has said, the Tv pundits are morons, and the losing candidate always proclaim loud and long that the polls showing themselves losing are wrong. Max bets at 5dimes on elections are only 1000 dollars so it is not like I am getting rich. I suppose i could find somewhere else to place my bets for more money but I don’t trust any of those shady suckers.

    • pithom says:

      “Not long ago between the Wisconsin primaries and the New York primaries Trump was only a slight favorite to beat Cruz for the republican nomination.”

      -Yeah; that was both weird and stupid. Didn’t everybody know New York was Trump’s home state? The primary was over when Trump won Florida.

      “the demographics are weighed too heavily against him.”

      -Romney would have lost even if demographics were copy-and-pasted over from 2004. Trump’s no Rmoney; he’s beating Rmoney’s very best days ever during his losing campaign in all the FiveThirtyEight models today. It’s not so much Trump’s overperforming Mitt, as Clinton is underperforming O due to Her unlikeability. I’ve predicted a Trump win for quite a while; and it’s looking more plausible with every passing week.

      • If Trump is a 4 point favorite in the polls with days to go and the stupid betting market gives me a great betting opportunity, hell I will take it and bet Trump. I don’t think that will happen but opinions are like assholes, everybody has one.

  11. The fourth doorman of the apocalypse says:

    Uncle Joe knows that it’s not who votes that counts.

  12. MawBTS says:

    Who ever is ahead in the polls at the time of election is extremely likely to win. Talk about how Candidate X would have a ‘difficult path to 270 electoral votes’ when he’s up 2 points (for example), is pretty much horseshit.

    The basic problem is this: the media is only necessary in conditions of uncertainty. As soon as you know what’s happening (or what’s going to happen), you turn off the TV. Therefore, they have an incentive to make one-sided battles look like they’re close-fought.

    If you’ve ever watched sports you’ve seen this. The Anytown Bumfucks will be up thirty points over the Podunk Shitsplats, minutes left on the clock, and the sportscasters will STILL be talking about how it could turn at any moment. “Boy, those Shitsplats look like they’re reaching deep into themselves and getting real angry. I think we’ll see a late-game turnaround. It’s not over ’till it’s over!” They’ve got to keep you watching until the ad break.

    Much has been written lately about how the media has covered Clinton vs Trump. At the start, it seemed like they were pretty uniformly negative on Trump. But as his approval ratings cratered (538 gave Clinton 9-1 odds in mid August), the press really started kicking Clinton. Not just my impression: STUDY: Cable News Devotes 13 Times As Much Coverage To Clinton Health As Trump Foundation

    Presumably if Trump starts doing too well, they’ll go back to stumping for Clinton. They don’t want this to become garbage time. They’re useless if it does.

    • Ziel says:

      That’s silly. The Trump Foundation has been around the whole campaign with available public records. There’s no reason it should be news on September 12. But a presidential candidate collapsing at an event at 10 in the morning 2 months before election day is unprecedented and as big a political news event as can be imagined, short of assassination.

      • dearieme says:

        Whatcha mean “short of assassination”? Obviously it was an attempted assassination by those vile Soviets, sorry Russians, acting for their puppetmaster Trump. Or maybe he’s the puppet; it’s so hard to keep up with loony conspiracy theories.

        • ziel says:

          Trump is the puppet. They will rig the election for him and pay off his massive debts, in exchange for which he will, once he becomes Commander in Chief of the most powerful military force in the world, do what they tell him to, because he promised.

  13. The Z Blog says:

    I’ve always liked to think of elections as fitting certain well known models. 1) The challenger versus incumbent in good times. 2) The challenger versus incumbent in bad times. 3) Two new faces in good times. 4) Two new faces in bad times. 5) Change versus status quo in changing times.

    The first two are familiar and the most common. If you are the incumbent in good times, you just run the clock and you most likely win. If it is bad times then you paint the challenger as a reckless idiot. The next two scenarios are pretty much the same, except in good times the goal is promise change, but not really mean it.

    The funny thing is the last option is the rarest, yet the mass media pretends it is the most common. As a result, most people try to model the election this way. The 2008 election was Option #2, but everyone wanted to believe it was Option #5.

    In all probability, this election is #2 with Clinton a stand in for Obama and Trump the challenger promising to do the same, but better.

  14. dearieme says:

    “most political reporters are morons, nearly as bad as sports reporters”: don’t forget that sports reporters virtually always get the result right, and almost always get the score right.

  15. Julian says:

    I’ll be interested to see if Professor Norpoth’s model holds up.

    “In a match-up between the Republican primary winner and each of the Democratic contenders, Donald Trump is predicted to defeat Hillary Clinton by 52.5% to 47.5% of the two-party vote. He would defeat Bernie Sanders by 57.7% to 42.3%.

    For the record, the PRIMARY MODEL, with slight modifications, has correctly predicted the winner of the popular vote in all five presidential elections since it was introduced in 1996. In recent elections the forecast has been issued as early as January of the election year.

    Presidential elections going back as far as 1912 are used to estimate the weight of primary performance. It was in 1912 that presidential primaries were introduced. That year the candidate who won his party’s primary vote, Woodrow Wilson, went on to defeat the candidate who lost his party’s primary vote, William Howard Taft. As a rule, the candidate with the stronger primary performance wins against the candidate with the weaker primary performance. For elections from 1912 to 2012 the PRIMARY MODEL picks the winner, albeit retroactively, every time except in 1960.”

    http://primarymodel.com

    • gcochran9 says:

      Those models are of course trash. A metaphorical cee-gar to whoever bothers to explain why.

      • whyteablog says:

        They’re done in hindsight. You can tweak em and tweak em until they “predict” things that already happened.

        Of course, you can’t do that for things that haven’t happened yet.

      • albatross says:

        Anyone want to buy my lucky penny, which has correctly predicted the direction of the stock market twenty days in a row? I’m telling you, it’s one in a million!

    • tautology123 says:

      There was an ctopus in germany who correctly predicted winners in soccer. Some people wonder how that worked. Selection effects, not the natural ones.

  16. Mark F. says:

    He predicted Trump would win the popular vote. He was wrong this year.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s