The Right’s War on Statistics

‘Zero Hedge’ is in a flap about poll ‘oversampling’ here http://www.zerohedge.com/news/2016-10-23/new-podesta-email-exposes-dem-playbook-rigging-polls-through-oversamples

Gasp!

It even includes a hack email from John Podesta which discusses ways of ensuring that the Democrats own polling over samples minority groups. Again, gasp!

Except. Well over sampling a smaller demographic group is the right thing to do. When I say ‘right’, I don’t mean for opinion polls but for collecting statistics on a population in general.

Say you have a representative sample of a population consisting of a thousand people. Now, of that thousand people you are particularly interested in a sub group that represents 1% of the population. If your sample is exactly proportionate, then it should have 10 people belonging to that sub group. Unfortunately 10 is a shitty sample, if you are unlucky to get 2 odd people with unusual views they then form 20% of your sub-sample.

Sample size is a dark art but the easiest issue to understand is it that magnitude matters. A good sample size is less about the proportion of the whole population in your sample and more about the raw number of people. More is better, but ‘more’ is subject to diminishing returns.

Over sampling means you can get a better picture of the sub group. However, because you end up with more of group X than you should have, their response are then weighted proportionally when looked at the statistics overall.

Are polls manipulated! Well, if by ‘manipulated’ you mean ‘use statistics’ then yes.

Advertisements

23 comments

  1. Aaron

    There is the further issue that the “smoking gun” e-mail that Trump is complaining about is from 2008, and it is clear that it is referring to the campaign’s internal polling. Not the public polling done that shows up in the newspapers, but in the campaign’s own polling that they do to understand how specific segments of the electorate are leaning. As in, information that generally never becomes public.

    Like

  2. thephantom182

    Speaking of guns smoking and whatnot, more dead people were at the polls in Colorado this week. No links for you, go look it up. It’s on Drudge right now.

    Dead people voting. Caught doing so as well. They found a dozen or so. How many did they miss? How big a problem is that? We shouldn’t be asking, right? All in a good cause, right?

    Podesta-quiddick: I wondered what you would make of the oversampling issue. Democrats, colluding with major media, caught RED HANDED jiggering polls to suit themselves, and you’re saying “nothing to see here.”

    Of course you are. You’re even pretending the pollsters applied the proper analysis to compensate for the oversampling.

    Speaking of credulous idiots, I saw a media analysis the other day. 91% of coverage is not anti-Trump, it is savagely anti-Trump. This according to the article. As well, Google was caught screwing with their search suggestions again, this time by Ezra Levant of Rebel media in Canada.

    Sorry for interrupting your day with an alternate opinion, Camestros. As you were.

    Like

  3. thephantom182

    Yes, it would appear I am completely deluded… oh wait, here’s some more in Chicago…

    “In all, the analysis showed 119 dead people have voted a total of 229 times in Chicago in the last decade.”

    Behold: a whole cemetery, voting. Twice.

    And no, no link for you. Drudge Report links to a TV news article. Look it up.

    This could get to be a record year for zombie voting.

    Like

    • Aaron

      We know why you won’t provide links – because you’re hoping no one goes and checks up on your pearl clutching claims and make you look like the fool you are.

      119 voters who might have voted after they died – but that’s not actually proven in the report. The number was generated by comparing the Chicago voter rolls with the death rolls from the Social security Administration. The problem is, that doesn’t necessarily show that someone who is dead voted, just that someone with the same name did, and it may surprise you to learn that there are multiple people who live in Chicago who have the same name. Also, in the same time frame that the report talks about, 60,000 people have been purged from the Chicago voter rolls, meaning that even if all 119 people voted after they died, that’s a failure rate of purging of 0.2%. That’s some conspiracy you’ve uncovered there – the Chicago elections board has an accuracy rate of 99.8%. They sure are rigging those elections.

      Chicago voters cast just over two million votes in the 2012 election, a similar number in 2008, and roughly half that in 2010 and 2014. That’s six million votes in that time frame. Dead people allegedly voted a total of 229 times in that time frame. That’s .003% of the votes cast. That sure is some grand conspiracy you’ve uncovered there.

      And you wonder why you’re a laughingstock.

      Like

      • thephantom182

        “We know why you won’t provide links – because you’re hoping no one goes and checks up…”

        No Aaron, I don’t provide links because you specifically don’t deserve any of my time and effort. The article exists, it says what it says, and since I posted above there’s another one about the same thing in Philly. Please, go and check.

        And while you’re at it, check how many hanging chads it took to decide the 2000 election. The one where Bush was selected not elected, remember that? Couple dozen zombie voters would have sewed that right up for Algore.

        “And you wonder why you’re a laughingstock.”

        No, not really. You’re a partisan, you will say pretty much anything to boost your side. At this point I’m wondering how hard your head is going to explode if Robot Pantsuit loses. I’m thinking thirty yard splash radius. I have my umbrella ready, just in case.

        Like

      • camestrosfelapton

        Phantom – evidence of dead people on electoral rolls, we have already. Evidence that those dead people are then used in someway to actually VOTE in any kind of substantial numbers is what you seem to be struggling to provide.

        Like

      • thephantom182

        Camestros, you asked, I answered: “In all, the analysis showed 119 dead people have voted a total of 229 times in Chicago in the last decade.”

        That is not a struggle, that is a whole cemetery. Maybe not Woodlawn or Mount Pleasant, but a tidy rural church yard. As of now we have Chicago, Philly, Colorado, California and Texas all with the Voting Dead. Like the walking dead, but with ballots.

        There they are: dead, and voting. Hundreds that we know of. Otherwise known as Vote Fraud. That is the facts. That you don’t -like- the facts, that I can’t help you with.

        If you peruse Drudge right now there’s a bunch of these. My favorite is the guy facing 40 years in jail for inventing voters in Virginia, couldn’t even be bothered to dig them up. How many more like that guy are there?

        This is just the dead we’re talking about. You have yet to mention the illegal vote. That’s a thing too, apparently.

        Like

      • camestrosfelapton

        Sorry but no. One list of names and another list of names – both lists necessarily have errors and then the matching process necessarily has errors. Indeed we’ve seen with attempts to use this technique to remove ineligible voters that mismatches are frequent.

        So what you have is 229 in a DECADE as being an UPPER limit on the possible number of dead voting in *Chicago* (assuming they weren’t just using a sample of the records).

        Now, looking at the Chicago Board of Election Commissioners I count 39 elections in the past decade. So you’re graveyard amounts to about *SIX* votes per election in CHICAGO of all places! And some of those will be false positives.
        This is the shittiest zombie apocalypse ever.

        //If you peruse Drudge right now there’s a bunch of these.//

        Yeah, I know. Already discussed the multiple examples that get thrown around.

        The story is always the same: here is an example! Look! Look! Just imagine how many more there must be!
        The rational person goes, well how many more are there actually and goes looking and finds…well not that many considering the size of the US and the frequency of elections and many of those aren’t what they are claimed to be.

        Like

      • Aaron

        No Aaron, I don’t provide links because you specifically don’t deserve any of my time and effort.

        You don’t provide links because you know your claims are laughable. And yet you keep trying, screaming “199 voters!” in a voting population of more than three million. Yelling “229 times!” in a time period in which at least six million votes were cast. And you apparently never seem to notice the parts of the articles where it is pointed out that these numbers aren’t actually indicative of anything substantial – there are reasons why voters might be missed, and reason to believe some of these “examples” are actually false positives, and so on. But you don’t actually care about the truth. You just have an ideological axe to grind, no matter how ridiculously silly it makes you look.

        Like

  4. thephantom182

    “So what you have is 229 in a DECADE as being an UPPER limit on the possible number of dead voting in *Chicago*”

    Why is that the upper limit? That’s what a cheesy TV news station found without trying very hard. A Chicago station at that, in the tank hard for the DemocRats.

    Note that there’s no investigation and nobody has been charged. That’s the important part, Camestros. 229 proven crimes, zero charges.

    In other news, this is my new all-time favorite video from the campaign. Background is: homeless bag lady defends Trump’s star on the walk of fame from vandals.

    Enjoy, as your fellow travelers show us what they’re made of. The comments by the dork who filmed it are priceless.

    This here is my all-time favorite still picture of the campaign.

    This is the sweetest pic -EVER!- I love it. So moving, so fulfilling, its almost pr0nz-like in its perfect beauty! Instant karma, baby!

    By the way, it looks like Anthony Weiner’s wiener may be the straw that breaks Hillary’s back, to coin a phrase. FBI is re-opening the investigation, Huma “Wiener Girl” Abedin’s wandering hubby was sexting a 15 year old and classified Hillary sh1t got on his phone/ipad/computer. Talk about an October surprise, eh? You think Director Comey will be able to pretend this is all nothing, again, before Wednesday? There’s a US Army General going to jail for this, y’know. He didn’t even have a private server.

    Imagine, elected President in November by the Voting Dead, chucked in jail by December because of Weiner’s wiener, blows -another- cerebral artery and becomes vegetative instead of Vegan by January. If somebody wrote that in a movie script I’d walk out screaming for my money back, for breaking my suspension of disbelief.

    Like

    • Aaron

      Why is that the upper limit? That’s what a cheesy TV news station found without trying very hard. A Chicago station at that, in the tank hard for the DemocRats.

      Oh look, a baseless claim that a TV station is “in the tank hard for the Democrats”. That’s convincing. It doesn’t make you look like a lying weasel at all to make things up like that.

      And that’s the upper limit because that’s the upper limit of what has been found. That’s the upper limit because we deal in evidence and not the wild-eyed and baseless speculation you indulge in. That’s the upper limit because there are many reasons why those votes could have been validly cast – all of which is pointed out in the article that you didn’t link to. Given that the article completely eviscerates so many of the claims you make, one really doesn’t have to wonder why you didn’t link it.

      Note that there’s no investigation and nobody has been charged. That’s the important part, Camestros. 229 proven crimes, zero charges.

      There are 229 examples of people whose names on the voting registration rolls matched names of people who are listed as dead on the Social Security rolls. That’s not “229 proven crimes”. That’s 229 times the two lists matched. As I noted before, and this may come as a surprise to you, but there are people in Chicago who have the same names as other people. There are times when the Social Security rolls are wrong. And so on. As usual, facts don’t matter to you, only the ideological axe you are trying to grind.

      By the way, it looks like Anthony Weiner’s wiener may be the straw that breaks Hillary’s back, to coin a phrase. FBI is re-opening the investigation, Huma “Wiener Girl” Abedin’s wandering hubby was sexting a 15 year old and classified Hillary sh1t got on his phone/ipad/computer.

      That story is dead already. Comey is now trying to cover his ass over the fact that the “revelations” have turned out to be nothing at all (and it also seems that he may be trying to backpedal so he doesn’t get nailed for a Hatch Act violation). One might also note that the FBI never said they were reopening the investigation. That’s the spin Chafitz put on the letter, and it isn’t supported by anything Comey actually said or wrote.

      Like

      • Aaron

        There’s a US Army General going to jail for this, y’know.

        He was a Marine. You should at least try to get your facts kind of straight. And he was convicted of intentionally giving out classified information and lying to FBI investigators. Comey has stated already that Clinton did neither of those things. You’re going to have to do better with your conspiracy theories, because you’re so bad at these that you’re getting really dull at this point.

        Like

      • thephantom182

        Aaron, you will not be interested to know that “FBI Emails” are -not- trending on twitter/farcebook/snapchat/etc. right now, despite everything with the Weiner’s wiener problem.

        Incidentally we already know Hillary lied to the FBI. They said she did. Huma did too. Like a Persian carpet.

        Also ABC/WaPo are doing ethnic oversampling on their polls, have admitted same, in writing, and Hillary’s 12 point “lead” in their polls was due to that. And also now gone, thanks to the “FBI Emails” that are not trending on social media. The bosses of which have their own tables reserved in the White House commissary. This is my not-shocked face. 😐

        It’s only a theory for paid Dem internet police, Aaron. Why do you think that blowhard reality TV dork isn’t at 10% in the polls?

        91% media negative on Trump, 91% media positive on Hillary, gushing praise from every side, Dems outspending Trump by 10-1 on ads, rape allegations against him, election is on Wednesday: crooked polls with ethnic oversampling put Robot Pantsuit at 47%, Aaron.

        That’s forty seven f-ing percent. With the biggest propaganda push in history. Goebbels would have wet himself at the thought of it. What’s wrong with that picture Aaron?

        Bring on the voting dead.

        Like

  5. Aaron

    Aaron, you will not be interested to know that “FBI Emails” are -not- trending on twitter/farcebook/snapchat/etc. right now, despite everything with the Weiner’s wiener problem.

    of course, social meadia “trending” is what determines the validity and importance of a story. They are likely currently “trending” because it has turned out that the e-mails aren’t from Hillary, or to Hillary, or on Hillary’s server, and the FBI doesn’t even know what is in the e-mails, because they haven’t gotten a search warrant, and as a result, have not read them. The fact that the Trumpistas have latched on to this story is an indication of just how desperate they are.

    Also, it seems likely that Comey violated the Hatch Act with his letter. So there is evidence of some illegality, just not by anyone connected to the Clinton campaign.

    Incidentally we already know Hillary lied to the FBI. They said she did.

    That’s funny. Comey said the exact opposite. Oh right, here’s the line from Comey’s testimony before Congress: “We have no basis to conclude she lied to the FBI.” You might want to check your sources again.

    Also ABC/WaPo are doing ethnic oversampling on their polls, have admitted same, in writing, and Hillary’s 12 point “lead” in their polls was due to that.

    You really don’t understand polling, do you? They haven’t admitted to “oversampling” Democrats, because they don’t select for party affiliation – when they poll they ask as part of the poll who the person being polled favors. The fact that this number tends to favor Democrats is not surprising, since there are more Democrats in the country than Republicans. This is a demographic reality that you seem to not be able to understand: If you poll 1,000 random people in the U.S. and ask them their political affiliation, you will get more Democrats than Republicans. What ABC and the Washington Post are doing isn’t oversampling Democrats, it is accurately reporting the results of randomized polling.

    Like

    • thephantom182

      “So there is evidence of some illegality, just not by anyone connected to the Clinton campaign.”

      And you call -me- a liar. Wow dude, projection much?

      “If you poll 1,000 random people in the U.S. and ask them their political affiliation, you will get more Democrats than Republicans.”

      Why is the Electoral map divided into red and blue, Aaron? There are plenty of areas in the USA where you won’t find many DemocRats. Minnesota, for example, is sharply divided. Once you get out of Minneapolis, there’s zero Dem-majority districts other than Indian reservations. Interesting to look at that map some time, it really is quite revealing.

      “What ABC and the Washington Post are doing isn’t oversampling Democrats, it is accurately reporting the results of randomized polling.”

      What ABC/WaPo is doing is -oversampling- in high minority population geographic locations. Minority populations have historically voted Dem, although this year is seems a lot of them are thinking twice about it. If you oversample in Minneapolis, you get a skewed result for all of Minnesota. That’s what they’re doing.

      Also, the polling isn’t even faintly random. It is telephone polling, and respondents self-select. That’s a huge bias right off the top.

      And with all the bias baked in, Hillary is not winning. The rodeo clown is dead even with her. I bet that’s really torquing your head bolts, Aaron. ~:D

      Like

  6. Aaron

    And you call -me- a liar. Wow dude, projection much?

    Since you spend all your time lying, it isn’t projection. Thus far, Comey’s announcement has amounted to e-mails, not on a device that Clinton used, or which were sent by her, and which the FBI has not looked at because they don’t have a warrant, exist. On the other hand, we have Comey violating the Hatch Act.

    Why is the Electoral map divided into red and blue, Aaron? There are plenty of areas in the USA where you won’t find many DemocRats. Minnesota, for example, is sharply divided. Once you get out of Minneapolis, there’s zero Dem-majority districts other than Indian reservations.

    And? You don’t seem to understand demographics. Minnesota has a population of roughly 5.5 million. The Minneapolis-St. Paul metropolitan area has a population of about 3.3. million. You seem to think that those large areas of red are meaningful, but they are mostly empty in comparison with the urban centers. Polling evenly across the state requires that one more heavily in Minneapolis-St. Paul, because that’s where roughly sixty percent of the people in Minnesota live.

    The U.S. map looks red if you just look at it by area, but vast swathes of the country are comparatively empty – even if you combine all of the electoral votes from reliably Idaho, Wyoming, Montana, North Dakota, South Dakota, Kansas, Oklahoma, Arkansas, Missouri, and Alaska and add all five of Nebraska’s electoral votes, you still won’t equal the total of reliably blue California. The country isn’t evenly divided between Republicans and Democrats. Years of data on this subject has confirmed that there are more people who identify as Democrats than people who identify as Republicans in the country.

    You would think after the fiasco with the “unskewed” polls from the last election cycle in which conservatives made the same baseless claims about how the polls were biased, and when election day rolled around, the polls turned out to actually have been accurate all along, much to Karl Rove’s dismay. The polls weren’t “skewed” then, and they aren’t biased now. Trump really is losing this election. I predict right now that Clinton will garner more than 300 electoral votes. But just wait a little over a week. You’ll learn the hard way that all the wishful thinking you’ve been doing about how the polls are biased has been the result of your not understanding how polls work.

    Like