Skip to main content

A poll worker hands a sticker to a voter at a polling place in Charlotte, North Carolina October 27, 2012. REUTERS/Chris Keane
Voted? Yes. For a Republican? Ha ha sure.
Oof. In an article on the GOP's ill-fated efforts to win support among black Americans, we get this laugher:
RNC officials stress that they’re investing in a long-term strategy and that the results won’t always be immediately visible, but they do have one favorite statistic they like to pass along to reporters: From January 2013 to January 2014, the Republican Party’s favorability among black voters increased from 12% to 16%, according to a Wall Street Journal–NBC poll.
Talk about cherry-picking! But as far as cherry-picking is concerned, it's not even good cherry picking.

Here's the January 2014 poll. I don't see any crosstabs on Republican Party favorability and the African-American vote, but I assume they somehow got the raw data and crunched the numbers themselves.

In that poll, 12 percent of respondents were black, or 96 of the 800 surveyed. Want to know the margin of error for that crosstab? 10 percent.

Yup. TEN percent. In other words, it renders the numbers essentially moot, and that four-point "improvement" is well within the poll's margin of error.

What's more, here are the Republican Party's favorabilities:

January 2013:
Very Positive: 6
Somewhat Positive: 20

January 2014:
Very Positive: 5
Somewhat Positive: 19

So the GOP became more unpopular in that year. Now the changes were within the poll's 3.46 percent margin of error among all adults, so it's likely just float. But since the GOP doesn't understand margin of error, they are essentially suggesting that the GOP is becoming more unpopular among non-black Americans.

It's a zero-sum game: If the GOP is gaining among the 12 percent African-American population, and yet it is losing ground among all Americans, then it is becoming even more unpopular elsewhere. And given their popularity numbers are already rock-bottom among Latinos and Asians, that only leaves whites.

So are Republicans arguing that they have become less popular among whites in the past year? I'll let their unskewers try and figure that one out.

Originally posted to kos on Mon May 19, 2014 at 10:45 AM PDT.

Also republished by Black Kos community and Daily Kos.

Your Email has been sent.
You must add at least one tag to this diary before publishing it.

Add keywords that describe this diary. Separate multiple keywords with commas.
Tagging tips - Search For Tags - Browse For Tags


More Tagging tips:

A tag is a way to search for this diary. If someone is searching for "Barack Obama," is this a diary they'd be trying to find?

Use a person's full name, without any title. Senator Obama may become President Obama, and Michelle Obama might run for office.

If your diary covers an election or elected official, use election tags, which are generally the state abbreviation followed by the office. CA-01 is the first district House seat. CA-Sen covers both senate races. NY-GOV covers the New York governor's race.

Tags do not compound: that is, "education reform" is a completely different tag from "education". A tag like "reform" alone is probably not meaningful.

Consider if one or more of these tags fits your diary: Civil Rights, Community, Congress, Culture, Economy, Education, Elections, Energy, Environment, Health Care, International, Labor, Law, Media, Meta, National Security, Science, Transportation, or White House. If your diary is specific to a state, consider adding the state (California, Texas, etc). Keep in mind, though, that there are many wonderful and important diaries that don't fit in any of these tags. Don't worry if yours doesn't.

You can add a private note to this diary when hotlisting it:
Are you sure you want to remove this diary from your hotlist?
Are you sure you want to remove your recommendation? You can only recommend a diary once, so you will not be able to re-recommend it afterwards.
Rescue this diary, and add a note:
Are you sure you want to remove this diary from Rescue?
Choose where to republish this diary. The diary will be added to the queue for that group. Publish it from the queue to make it appear.

You must be a member of a group to use this feature.

Add a quick update to your diary without changing the diary itself:
Are you sure you want to remove this diary?
(The diary will be removed from the site and returned to your drafts for further editing.)
(The diary will be removed.)
Are you sure you want to save these changes to the published diary?

Comment Preferences

  •  For Republicans 'math' really is just (22+ / 0-)

    a four-letter word. ;D

    GOP 2014 strategy -- Hire clowns, elephants, and a ringmaster and say "a media circus" has emerged and blame Democrats for lack of progress. Have pundits agree that "both sides are to blame" and hope the public will stay home on election day.

    by ontheleftcoast on Mon May 19, 2014 at 10:47:37 AM PDT

  •  You know, all they have to do (22+ / 0-)

    is have Dr Ben Carson as their candidate and watch AAs rush to the polls to elect him.  ROTFLM(Blah)AO!

    Maya Angelou: "Without courage, we cannot practice any other virtue with consistency. We can't be kind, true, merciful, generous, or honest."

    by JoanMar on Mon May 19, 2014 at 11:04:22 AM PDT

    •  Be careful... (5+ / 0-)'ll laugh all the blah off yerself!

      "Ronald Reagan is DEAD! His policies live on but we're doing something about THAT!"

      by leftykook on Mon May 19, 2014 at 11:14:20 AM PDT

      [ Parent ]

      •  Very little chance of that; but if it could (5+ / 0-)

        happen, this and the spectacular failure of Operation American Spring would have put me in serious danger.

        “It’s a very dismal turnout,” said Jackie Milton, 61, a Jacksboro, Texas, resident and the head of Texans for Operation American Spring, to The Washington Times. He said hopes were high when he arrived in Alexandria, Va., a day or so ago and found motels and hotels were sold out for 30 miles around.
        Read more:


        Maya Angelou: "Without courage, we cannot practice any other virtue with consistency. We can't be kind, true, merciful, generous, or honest."

        by JoanMar on Mon May 19, 2014 at 11:33:10 AM PDT

        [ Parent ]

    •  I doubt the GOP leadership really believes the (5+ / 0-)

      false "black voters always support their own kind" meme that they've been pushing. I suspect their real goal is to portray African Americans as irresponsible voters who engage in knee-jerk group loyalty rather than thinking about the candidates and the issues, perhaps as part of an effort to justify the party's voter suppression strategy to members of their base who might find it distasteful.

    •  You mean Romney-Carson 2016. Mitt has been in the (5+ / 0-)

      media prepping his second run for months now. He just needs a token Black guy to inoculate himself from charges that he's a rich, White, out-of-touch asshat, who grew up believing Blacks were inferior, until a new magic prophesy appeared one day saying maybe they weren't...or at least that they shouldn't say that in public anymore.

      Just doing my part to piss off right wing nuts, one smart ass comment at a time.

      by tekno2600 on Mon May 19, 2014 at 11:50:02 AM PDT

      [ Parent ]

    •  Or Ron Christie (4+ / 0-)
      Recommended by:
      a2nite, Naniboujou, JoanMar, nocynicism
      Ron Christie, a black Republican strategist and Fox News contributor, said that even if Republicans haven’t landed on a unified message, people of color recognize the party is taking an interest in their vote for the first time in many years.

      “I will tell you, from my television work, I have noticed in people of color who have stopped me on the street in recent months have said, ‘You know, I still don’t agree with you guys but you’re really trying to get black voters to listen to you,’” Christie said. “The biggest mistake the party could make now is to say, ‘We increased our number of black votes from 7% to 10% and now we’re done! Mission accomplished.’”


      We view "The Handmaid's Tale" as cautionary. The GOP views it as an instruction book.

      by Vita Brevis on Mon May 19, 2014 at 12:04:07 PM PDT

      [ Parent ]

  •  Newest trick: We're the droids you're looking for! (8+ / 0-)

    I live under the bridge to the 21st Century.

    by Crashing Vor on Mon May 19, 2014 at 11:08:05 AM PDT

  •  Oh, you eggheads and your so-called "numbers" (6+ / 0-)

    It's well known that you can use "facts" to prove anything that's even remotely true. But you know what Republicans use to prove stuff? That's right, their feelings. And they trust their gut to tell them not only whether that burrito in the back of the freezer might have gone over, but also whether Mitt Romney really won Ohio. All the book learnin' in the world isn't going to trump that!

    And if 96 randomly-selected black people are surveyed, changing three responses should show up a whopper of a difference! So there, you libruls! You're trying to disprove a whopper, and Republicans won't stand for that.

    •  Even with their unskewing ... (0+ / 0-)
      From January 2013 to January 2014, the Republican Party’s favorability among black voters increased from 12% to 16%, according to a Wall Street Journal–NBC poll.
      It's funny that even with their strongest efforts to find something positive in the numbers, the best the Republican unskewers can come up with is still basically:

      In the last year our favorability among black voters has risen from "really, really pathetically sucks" to "still really, really pathetically sucks."

  •  This seems salient... (1+ / 0-)
    Recommended by:

    The margin of error for 800 interviews among Adults is ±3.46%

  •  I think the RNC is referring ... (0+ / 0-)

    to the 'other' African Americans. You know, that invisible, and often forgotten minority that Reince has invited into his tent.

    If music be the food of love, then laughter is it's queen, and likewise, if behind is in front, then dirt in truth is clean.

    by glb3 on Mon May 19, 2014 at 11:18:15 AM PDT

  •  from that article, to the ears of their donars (0+ / 0-)

    zero interest in any other readers...

  •  Somehow, Republicanism seems to bring with it (2+ / 0-)
    Recommended by:
    ontheleftcoast, T J Lewis

    an immunity to cognitive dissonance.

    If I ran this circus, things would be DIFFERENT!

    by CwV on Mon May 19, 2014 at 11:18:34 AM PDT

  •  As long as the modern GOP (4+ / 0-)

    is the party of white resentment, they are not going to get many black voters.

    “If the misery of the poor be caused not by the laws of nature, but by our institutions, great is our sin.” Charles Darwin

    by ivorybill on Mon May 19, 2014 at 11:23:32 AM PDT

    •  Story Often Told Of LBJ Signing Civil Rights Bill (5+ / 0-)

      He turned to someone and mentioned that he had lost the south to the Democratic party for generations.

      I think when Ronald Reagan first decided to use the story of the Chicago "Welfare Queen" on the campaign trail he should have turned to someone and said "we are associating the GOP now with white resentment. It will pay off with working class whites. We will have "Reagan Democrats. But the party will now always have a problem in getting support from African-Americans out of the single digits"

      •  Reagan kicked off his campaign in (0+ / 0-)

        Philadelphia, Mississippi where 3. Civil rights workers were
        Google... Prof. Ian Haney-Lopez...Dog whistle politics.

        We are not powerless!! "Activism is the rent I pay for living on this planet."– Alice Walker

        by nocynicism on Sat May 24, 2014 at 01:39:22 PM PDT

        [ Parent ]

  •  Can someone explain to me (1+ / 0-)
    Recommended by:

    how the margin of error is determined for any given sample size?

    I always see(ok, not always) the margin of error listed. I understand what a MOE is...but how you arrive at it? I have no idea.

    Follow-up question: Does a poll of 1000 people always have the same MOE? I'm gonna guess it doesn't, so the next question is: why not?

    Non-math Fella

    The better I know people, the more I like my dog.

    by Thinking Fella on Mon May 19, 2014 at 11:27:06 AM PDT

    •  Size of the MoE depends... (2+ / 0-)
      Recommended by:
      Thinking Fella, suzq

      Not only on the size of the sample, but the normalizing done to the sample.  For any statistical analysis to truly work, you have to normalize the sample as much as possible, to remove variables that you're not interested that may cause some noise.  Generally, the larger the sample, the less normalizing needed.

      However, when dealing with subjective questions, where results can be altered by the way the questions themselves are asked, this is more difficult.

      I don't know the specifics myself unfortunately (it's been a long time since I took statistics), but that's the basic gist of your second question.

    •  It has to do with confidence (2+ / 0-)
      Recommended by:
      Thinking Fella, nocynicism

      MOE is indirectly proportional to the sample size.

      There's a lot of statistical mumbo jumbo involved: There's a good wiki article here

      We've been spelling it wrong all these years. It's actually: PRO-GOP-ANDA

      by Patriot4peace on Mon May 19, 2014 at 11:34:55 AM PDT

      [ Parent ]

    •  There are three things that affect the MoE (3+ / 0-)

      The first is the sample size in proportion to the size of the population (which is the set of all possible respondents). As the sample size approaches the size of the population, the MoE goes down.

      For instance, if you were going to estimate the President's popularity among US voters, it's easy to see how a sample size of 1 would have a huge MoE. As the sample size gets larger, though, the MoE gets smaller until, if the sample size were the same as the population size (you had asked all voters in the US) the MoE would be zero because you would no longer be estimating from a sample, but would have the actual proportions.

      The second thing that affects the MoE is the variability of the possible responses that are out there. If everyone in the US held the same opinion of the President, the MoE would be zero because there would be no chance you were missing a possible response or miscalculating it's proportion, no matter how small your sample. On the other hand, if there were thousands of possible responses, all held by equal numbers of citizens, you would need a very large sample to accurately capture all of those different sentiments.

      The final thing that affects the MoE is how confident you need to be before you say that the diference is significant. Generally, that is set somewhat arbitarily at 95%. In this case, that means that we can be 95% sure that the true support for the Republican Party is somewhere between 6% and 26%. In other words, it's hard to say that 4% increase is real.

      Despite that, it is possible that the Republicans have seen an uptick in popularity. We just can't say that will 95% certainty. This last point is something that lots of people get wrong. They assume because a result is just inside the MoE that they should automatically ignore it completely. That's not the case. It's entirely possible in many cases, that if we had chosen a confidence level of 80% or 90% that the result would be significant. That said, this result is far enough in the MoE that the confidence low is probably extremely low, and we should ignore the increase.

      To believe that markets determine value is to believe that milk comes from plastic bottles. Bromley (1985)

      by sneakers563 on Mon May 19, 2014 at 11:55:07 AM PDT

      [ Parent ]

    •  Mathematician Karl Friedrich Gauss (1+ / 0-)
      Recommended by:
      Thinking Fella

      worked out what became these elementary statistical measures early in the 19th century. Wikipedia has formulas and tables along with explanations of the concepts in its article on Margin of Error.

      Basically, the margin of error is a range where the correct answer should be found a certain percentage of the time. If the percentage is not stated, it should be 95%, but when people are misusing statistics, one should never assume that without checking.

      That confidence level means that the result will be outside the specified range one time in 20. So if you look at more than 20 estimates, you can expect to find one that is out of range. Any given poll will have more than 20 cross-tabs, that is numbers for subsets of those polled by age, gender, ethnicity, and so on.

      Selecting from only those most extreme results the ones favorable to your cause, and ignoring results from any other polls, is called cherry-picking. It is the relatively moderate form of unskewing, where the extreme case is to claim that all of the polls are biased or fraudulent, and further, if that claim fails, that some other Conspiracy Theory explains the observed results.

      Like blaming ACORN for stealing the 2012 election, long after it has ceased to exist. Extreme unskewing is the result of Cognitive Dissonance, as explained in When Prophecy Fails, by Festinger et al. For The True Believer (see the book of that name by Eric Hoffer) it is simply unacceptable that the world does not agree with them, that there is no longer, in their terms, a Moral Majority.

      Whereas we note with satisfaction that there is a moral, Progressive majority, held back only by gerrymanders, the Senate filibuster and Blue Slip traditions, and the Supreme Court. Assuming that we GOTV, because when we vote, we win. When we next take both houses of Congress we will be able to deal with all three of those obstacles.

      Back off, man. I'm a logician.—GOPBusters™

      by Mokurai on Mon May 19, 2014 at 12:01:07 PM PDT

      [ Parent ]

      •  thanks. (0+ / 0-)

        I guess my question was more along the lines of "How did you arrive at a MOE of 3.145 rather than a MOE of 4.123?" What numbers did you use to arrive at such a precise MOE? Can't those numbers be just as wrong as the MOE in any given poll?"

        I appreciate your explanation. Being a math failure, I wonder how accurate ANY polling is. People seem to keep hiring polling firms who end up way, way off from final results so I figure there is quite a lot about polling I simply don't understand. Also., math

        The better I know people, the more I like my dog.

        by Thinking Fella on Mon May 19, 2014 at 12:28:47 PM PDT

        [ Parent ]

        •  A key point is that estimates for a proportion (1+ / 0-)
          Recommended by:
          Thinking Fella

          (which this is) have a worst-case margin of error that can be reliably computed. The only things that can go wrong are that (1) the stated results do not actually correspond to the result measured (wrong variable, biased or leading question, confusing language, ...) or (2) the sample was far from either representative or random.

          For an average value, a great deal more can go wrong, starting with the possibility that what you're measuring doesn't really have a mean or, if it does, doesn't have a standard deviation (measure of the spread of the data). The latter is likely to have been one of the problems in the mathematical models for risk that led to the Great Recession--even assuming that those models were honestly derived and honestly represented.

          •  AIG is known to have had an accurate estimate (0+ / 0-)

            of its risks, computed by its actuaries, that management refused to pay any attention to, because in their estimate the company had no real risk under Too Big to Fail.

            That actuarial calculation is known as the Ruin Function. It gives the probability that an insurance company will have claims amounting to more than it can pay.

            Back off, man. I'm a logician.—GOPBusters™

            by Mokurai on Tue May 20, 2014 at 02:44:26 PM PDT

            [ Parent ]

            •  But the entire calculation (0+ / 0-)

              Is invalid (and completely useless) if the standard deviation of the loss is unbounded.

              •  In that case you don't get a numeric result (0+ / 0-)

                but you can tell that the company is going under if nothing else is in play. That is not an invalid result.

                Back off, man. I'm a logician.—GOPBusters™

                by Mokurai on Tue May 20, 2014 at 07:53:01 PM PDT

                [ Parent ]

                •  Yes, but (0+ / 0-)

                  even that's useless, since it only says "given an unbounded amount of time, ...".

                  •  No, the actuaries were pointing out that AIG (0+ / 0-)

                    had a high probability of going under within a year because all of the insured mortgages were going to go underwater at the same time, as they did. The probability and magnitude of the prospective catastrophe went up with each succeeding year that the bubble could manage not to burst, while the probability of the bubble bursting also went up strongly each year.

                    The point of standard Ruin Theory is to give the probability of disaster over time into the unlimited future, assuming that policy claims are independently distributed. The AIG case is far worse. Individual defaults are mostly independent, but a crash is 100% correlated.

                    Back off, man. I'm a logician.—GOPBusters™

                    by Mokurai on Wed May 21, 2014 at 07:55:24 AM PDT

                    [ Parent ]

                    •  That was a different modeling failure (0+ / 0-)

                      Failing to account for correlations between variables.

                      Look, I'm not disputing that the problem could have been, and actually was, predicted. I'm looking at the various failures and invalid assumptions in the model that was used to justify packaging low-quality mortgages.

    •  The MOE is also affected by the answer (1+ / 0-)
      Recommended by:
      Thinking Fella

      If you are comparing Romney and Obama then the answer is going to be close to 50% each and a 200 sample size can give you good results.

      If you are trying to estimate the support for the libertarians (which is likely in the 5% range) then you need a bigger sample because with 200 sample size 5% = 10 people and if you are even one person out then that has a big effect on your result.

      Polls routinely record the race, age etc. of the people polled. This is so they can do a double check that the sample isn't skewed (e.g. too many old people because they were more willing to stop and voice their opinion). These numbers can be used by the pollsters to 'unskew' the results = massage them to get the result they would have got if the sample reflected the population. This can miss out bias in the sample which wasn't recorded (e.g. interviewers preferring to interview young afro american males wearing shirts and ties over those wearing hoodies).

      Once you have those numbers of course there is the temptation to publish poll numbers for each segment of the population but the number of people in the black/white/evangelical etc. segment of the sample will be much smaller than the total sample size so the MOE when estimating their opinions will be bigger. When you put them all together you assume that random chance means that you are high in some segments and low in others so the MOE for the whole sample will be less than the MOE for any one segment.

      Hope that helps.

  •  despite the bogus stats (4% improvement! LOL) (1+ / 0-)
    Recommended by:

    if Ben Carson gets on the ticket, however unlikely, more skewing could occur in critical regions,

    Warning - some snark may be above‽ (-9.50; -7.03)‽ eState4Column5©2013 "If we appear to seek the unattainable, then let it be known that we do so to avoid the unimaginable." (@eState4Column5)

    by annieli on Mon May 19, 2014 at 11:27:23 AM PDT

  •  They called Clarence Thomas (2+ / 0-)
    Recommended by:
    chicago minx, Rosalie907

    then they called Alex Jones.

    So they got two positive responses for Republicans, which gave them a 2% bump with zero margin of error.

    We've been spelling it wrong all these years. It's actually: PRO-GOP-ANDA

    by Patriot4peace on Mon May 19, 2014 at 11:27:31 AM PDT

  •  i cant see how recent events (1+ / 0-)
    Recommended by:

    like those in NV and in sports are going to help the GOP

    like it or not bringing up race just reminds many people that racism still exists and that the SCOTUS and the GOP cant wish it away.

    gop is still playing a white man's game-and apparently loving it

  •  You know... (2+ / 0-)
    Recommended by:
    T J Lewis, Aquarius40

    If your poll has a 10% margin of error, it's ok to just say you have no clue.

  •  Delusional (1+ / 0-)
    Recommended by:

    is the best word to describe the GOP.  They just believe what they want to, no matter how what happens or what the facts show.  

  •  Math has a liberal bias. (nt) (5+ / 0-)

    Once you put convenient, lethal force in the mix, liberty becomes a zero sum game. -- DIgby on open carry.

    by Rikon Snow on Mon May 19, 2014 at 11:37:41 AM PDT

  •  Interesting that they say "favorabilty" increased (1+ / 0-)
    Recommended by:

    It's not quite clear what that means. Ordinarily, you'd have to assume that means increases in the "Very Positive" or "Somewhat Positive" categories on the poll, since there is no category that is actually called "favorability." However, given that these are Republicans we are talking about, who increasingly call alternative reality their home, one wonders if they are also counting other things as an increases in favorability: e.g. a higher number of Black voters who say they are "Neutral or Unsure" (hey, that's not necessarily unfavorable, so in the Republican mind that must be favorable, right?); or maybe it's just a lower number of negative views than previous polls (which, as kos pointed out, is not quite the same as a higher number of positive views...except in Republicanland).

    Anyway, color me skeptical for thinking that all the vicious, personal, and racially undertoned attacks on our first African American President have made the Republicans increasingly popular with African Americans and less popular with Whites.  

    Just doing my part to piss off right wing nuts, one smart ass comment at a time.

    by tekno2600 on Mon May 19, 2014 at 11:45:02 AM PDT

  •  Margin of error (0+ / 0-)

    The margin of error -- is the amount by which one poll may report the percentage of the population wrong with a low probability if the answer is 50%. It isn't a zero probability -- you can get every person with one opinion in the population in one sample of one hundred (though that is incredibly unlikely).  

    It is too large if the answer is higher than 80%.

    It is roughly 70% of the difference between two polls that would yield that probability.

    It may be the single term which is least understood in the entire study of polling.

  •  At some point... (0+ / 0-)

    the GOP will reach a crossroads. They can embrace the reality that this isn't 1950 anymore, and they'll have to moderate their views to gain acceptance with the diverse citizenry, at which point their intransigent members will splinter. Or continue on their current trajectory, scaring the hell out of the obtuse, and emotionally fragile citizens, hoping that they vote for the GOP and against their own interests.
    One of these courses has a future. We'll see.

    •  Neither course has a future (0+ / 0-)

      We already have a Democratic Party that considers the interests of real people, as opposed to corporate persons and their owners. You know, there is a word for owning people, and there are a lot of other words for people-owners.

      It is not possible to run a Party on actual Republican principles, specifically ever-lower taxes, union-busting, and deregulation. The historical parties of big business (Federalists, Whigs, and Republicans) have had to support other issues that would bring in other voters without harming business profits. Opposing slavery worked for the original Republicans, until the freed slaves wanted economic rights, too.

      Actually having a Republican platform for actually fixing the economy would get people to support Republicans, but actually fixing the economy requires regulations, union rights, and higher taxes. (See the Progressive Caucus Better-Off Budget.) This does not satisfy the rich and powerful who want to increase their margins of being richer and more powerful than everybody else, because letting the non-rich have any political power scares them to death.

      Back off, man. I'm a logician.—GOPBusters™

      by Mokurai on Mon May 19, 2014 at 12:15:40 PM PDT

      [ Parent ]

  •  could we get a statistician in here for an opinion (0+ / 0-)

    I don't think MOE is so simple. For instance, if Republicans can show an improvement from 10% to 16% with an MOE of 10%, it makes the data suspect, but does not mean, for instance, that the 10% popularity is wrong.

    But I always hates Stats, who doesn't

    Those who quote Santayana are condemned to repeat him. Me

    by Mark B on Mon May 19, 2014 at 11:56:33 AM PDT

    •  Somewhat (0+ / 0-)

      Polls always cite the MOE for 95% confidence, that is, the range of results that would cover 95% of random samples of the population. You can calc the MOE for other confidence levels, so maybe the 10% to 16% move would be significant at a 70% confidence level - still likely to be true, but a pretty good chance of it being wrong.

      "It's the (expletive) 21st century man. Get over it." - David Ortiz

      by grubber on Mon May 19, 2014 at 12:11:14 PM PDT

      [ Parent ]

    •  What you are really asking for is a brief, clear, (2+ / 0-)
      Recommended by:
      davelf2, GrumpyOldGeek

      and correct tutorial on Bayesian inference, which asks the question, what is the probability of a hypotheses H if we have evidence E? What, for example, might the real state of public opinion be, or how might a future election turn out, if all we have is polling numbers?

      There is no way to explain Bayesian inference in detail without a detailed explanation and numerous examples of the formula for Bayes's Rule.

      P(H|E) = P(E|H)*P(H)/P(E)

      where P(H|E) is the probability of H, given E, and so on. You have to start by turning the original question around, and finding the probability of the evidence given the hypothesis. Then you correct for the likelihood of the hypothesis independent of the evidence, and the evidence independent of the hypothesis.

      In science, we try to construct a hypothesis that predicts the known evidence with certainty. Then we look for predictions that are impossible on the previous theory, and run the experiment to find out, that is, to drive the probabilities supposed by other scientists from 0 confidence in the new theory to as close to 1 as we can manage. Quacks and cranks in science, religion, and politics refuse to abide by this method.

      I can be entirely correct on this. However, if I am to be clear, I cannot be very brief, and if I am to be brief at all, I cannot be entirely clear. But everybody understands some of the basic applications. For example, the rule that extraordinary claims require extraordinary evidence, and the Sherlock Holmes rule: When you have eliminated the impossible, then whatever remains, however improbable, must be the truth.

      For example, imagine that I have various hypotheses about the nature of a newborn baby of a friend, including:
      • H1: the baby is a brown-haired boy.
      • H2: the baby is a blond-haired girl.
      • H3: the baby is a dog.

      Then consider two scenarios:
      1. I'm presented with evidence in the form of a picture of a blond-haired baby girl. I find this evidence supports H2 and opposes H1 and H3.
      2. I'm presented with evidence in the form of a picture of a baby dog. Although this evidence, treated in isolation, supports H3, my prior belief in this hypothesis (that a human can give birth to a dog) is extremely small, so the posterior probability is nevertheless small.
      Where Republicans gaining support among Blacks is the unlikely dog.

      If I could make one change to high school math teaching, getting students to understand (not merely calculate) Bayes's Rule would be it.

      What is the probability that neutrinos can move faster than light, given that we have a measurement that says they did? There is no hypothesis that gives a probability above 0 for this, so we can easily conclude that there was a measurement error, which turned out to be the case.

      Cold fusion, ditto. Deuterium fusion starts to occur at measurable levels when the nuclei get within a certain distance of each other, many times smaller than the spaces between atoms in the palladium electrode where it was claimed to be taking place.

      What is the probability that Global Warming is not caused by humans, given the existence of the Greenhouse Effect and the correlations between burning carbon, measured atmospheric CO2, and increasing temperature, with all of its observed consequences? The probability of the observed CO2 levels, carbon burning, and temperature increases are all 1, because we can see them and measure them. The probability of those temperature increases from any cause other than greenhouse gases has been conclusively ruled out.

      What is the probability of Young-Earth Creationism, given its physical impossibility? That depends on the prior probability of an omniscient, omnipotent, and omnibenevolent God unknown to the Bible, but invented in the 16th and 17th centuries, and now invoked as an excuse because of the supposed improbability of evolution. You do the math.

      Now, for many believers in YEC, Adam has to be historical, or else Jesus can't save us.  Given that belief, you can calculate a chain of Bayesian probabilities with the results that any science to the contrary has a probability of 0, and it is certain that all such scientists are lying. Adding more facts makes it worse. The only remedy is to break down the theology, which millions of the children of Creationists do every year, for that and other reasons, so that their churches become ever smaller and ever purer at the same time, and thus louder and nastier.

      What is the probability that Trickle-Down Economics will ever work, given that it never has? Well, if it were effective, then this unbroken run of failure would be impossible, so the answer is 0. But wait! It works perfectly as designed, for enriching the already rich, and never mind the rest.

      What, then, is the probability that a Market Fundamentalist Economist or science denialist is lying or deluded, given that his salary is paid by those who benefit from what he says?

      It is difficult to get a man to understand something, when his salary depends on his not understanding it.
      Upton Sinclair, who thus turns out to be an excellent informal Bayesian analyst.

      Turning it around, what is the probability that a highly visible Denialist is being paid to do it? Low for deluded medical denialists such as anti-vaxxer Robert Kennedy, Jr. or AIDS denier Kary Mullis, also a Nobel Prize winner in an unrelated area of biology, and close to 1 for industries such as lead, tobacco, and fossil carbon.

      Back off, man. I'm a logician.—GOPBusters™

      by Mokurai on Mon May 19, 2014 at 01:37:30 PM PDT

      [ Parent ]

  •  They are like a deluded Sally Field: (1+ / 0-)
    Recommended by:

    "They like us!  They REALLY LIKE us!"

    Rush — the quivering rage heap who is apparently desperately trying to extinguish any remaining molecule of humanity that might still reside in the Chernobyl-esque Superfund cleanup site that was his soul. -- Jon Stewart

    by Its the Supreme Court Stupid on Mon May 19, 2014 at 12:01:30 PM PDT

  •  Until the Rcons give up the white supremacist he- (1+ / 0-)
    Recommended by:
    Ian S

    man woman haters; well listen politely. Voting for them is another matter.

    I voted Tuesday, May 6, 2014 because it is my right, my responsibility and because my parents moved from Alabama to Ohio to vote. Unfortunately, the republicons want to turn Ohio into Alabama.

    by a2nite on Mon May 19, 2014 at 12:12:06 PM PDT

  •  If you can't make it now, you'll never make it (0+ / 0-)

    I'd love to hear a rendition of New York where someone body sings, if I can't make it here, I'll never make it...any...where....referring to constant victum or slave mentality that can't make it in the land of opportunity. Sad.

  •  I think the more (0+ / 0-)

    interesting parts of that article are the parts where Sean Spicer talks about the "strategy" behind picking fights with black journalists:

    Spicer said part of the purpose of all this goading is to illustrate how the political biases of community gatekeepers often prevent Republicans from reaching voters of color.
    That's right. I and most other black people don't see, hear and read the same media whites see, hear, and read. We only peruse what the "gatekeepers" tell us we can peruse.

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site