Skip to main content

Mitt Romney's candidacy has a magical ability to increase its percentage of the votes cast in every voting precinct in the nation.  The amount that his percentage of the vote goes up is directly dependent on the TOTAL NUMBER OF VOTES CAST IN THAT PRECINCT.

This precinct-size dependent slope anomaly becomes especially compelling and conclusive because it is observed across many states and counties and favors the same candidate, regardless of its rank in the official vote count in that state.  
I didn't believe it so I went to the Ohio Secretary of State's website and downloaded the raw precinct-by-precinct datafile.  Yes, it is true, Mitt Romney did proportionally better in each of Ohio's precincts according to the size of the precinct's numbers of votes cast!

From the Comments below:

So I am supposed to believe that, for all states observed, the single most determining factor whether a precinct votes overwhelmingly more republican than the statewide totals is. . .

the total number of votes cast within a precinct?

here is the video with the paper that broke the story.

First, I read this report that shows that Mitt Romney's vote lead in key states increased at a rate that was proportional to the size of the vote precincts.  In ALL cases, he won a proportionally LARGER percentage of the vote if the precinct reported a larger vote total.

Nose Counting
The idea of examining large precinct results came via a link to a report written by Francois Choquette and James Johnson. Choquette became curious about South Carolina primary results in the February Republican contest. There a poll observer noted an unusually big gain of votes for Mitt Romney in larger precincts than in smaller ones. Choquette wanted to know why?

He examined and applied all of the normal statistical markers to see where a variance might occur: income level, population density, race, urban vs. rural, even party registration numbers. He found no correlation to explain why Romney votes trended upward while Paul and Santorum votes trended downward -yet only in large precincts.

So, I decided to read the referenced report.  The report is dated August 13, 2012.

It listed showed vote totals, as a percent of the total vote, listed according to precinct size (like I did, independently from the Ohio data below).  In the report they analyzed the states of. . .

Iowa

--

--

New Hampshire

--

--

Oklahoma, Alabama, Louisiana, Arizona (all showing increases in Romney's percentage proportional to the size of the precinct).

--

--

and

Wisconsin

Well, this just didn't make sense to me!  I mean, there is nothing in the laws of physics or politics that says a specific candidate should, AT A CONSTANT RATE, gather higher and higher percentages of the vote according to the number of votes cast within a precinct.

And then it became obvious, on a national level, the votes are being switched.

It is easier to hide the vote switching when more votes are cast within a precinct.

well, I wanted to be sure it wasn't faked,

here is their Ohio Reported Values

Here is mine, taken from the Ohio Secretary of State's Official Website

so, if the total votes cast in a precinct was 35, Mitt Romney's average percent of the vote total was 32.14 percent.  

As the size of the average precinct increased to 225 votes, Mitt Romney's average percent of the total vote increased to 35.43% of the total vote.

Once the Largest precincts were added to the final state total, Mitt Romney's average went up to 37.964% of the total state vote (just narrowly beating out Rick Santorum).

Here is what happens if you sort the precincts by the largest to the smallest precincts.

In the 500 largest Ohio precincts, Romney won 42.3% of the total republican vote

Once the final totals are made Romney won 37.964% of the total republican vote

In 2008 in the 500 Largest Precincts in Ohio, John McCain had 53.66% of the total votes cast and Obama had only 45%.

9:43 AM PT: Updated with 2008 Ohio President Vote Totals

by request, here is the scatterplot of Ohio 2008 McCain

By request, this is McCain's 2008 average vote take from precincts by quintile (the fifth largest precinct down to the fifth smallest precincts with average take from each of the groups)

2010 democratic incumbent Y axis is percent of precinct vote, x axis is size of votes in precinct

Here is the Republican Challenger who won

2010 strickland vs. Kasich

EMAIL TO A FRIEND X
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

  •  Compelling (3+ / 0-)
    Recommended by:
    New Minas, gfv6800, Lujane

    Recd for all the work this took
    Looking forward to the discussion

  •  Horrifying. n/t (1+ / 0-)
    Recommended by:
    Lujane

    Best. President. Ever.

    by Little Lulu on Sat Oct 27, 2012 at 09:07:13 AM PDT

  •  Did anyone do a control, (2+ / 0-)
    Recommended by:
    librarianman, Lujane

    say in the 2008 Demo primary with Obama against Clinton to see if the same sort of result occured?

    The world is a den of thieves and night is falling. -Ingmar Bergman

    by Pirogue on Sat Oct 27, 2012 at 09:10:19 AM PDT

    •  They say that (1+ / 0-)
      Recommended by:
      Lujane

      they are starting to look at the other elections, better to see the report.

      still, there is absolutely no reason to believe that a single candidate would have his percentage of the vote total go up just because the precinct had more votes cast, unless there was vote switching going on.

      •  you're confounding empirics with causality (4+ / 0-)
        Recommended by:
        elmo, Quicklund, Argyrios, Lujane
        there is absolutely no reason to believe that a single candidate would have his percentage of the vote total go up just because the precinct had more votes cast...
        Sometimes the old saw that "correlation doesn't prove causation" is overused, but it applies perfectly here.

        Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
        Better Know Your Voting System with the Verifier!

        by HudsonValleyMark on Sat Oct 27, 2012 at 10:01:15 AM PDT

        [ Parent ]

        •  Right. I guess that smaller precincts (2+ / 0-)
          Recommended by:
          HudsonValleyMark, Lujane

          would be more rural, and these would tend to support Santorum. Romney just does better in bigger urban areas because he appeals to economic voters more than social conservatives who have given him a pass but don't really like him. That alone would explain the effect.

          For if there is a sin against life, it consists perhaps not so much in despairing of life as in hoping for another life and in eluding the implacable grandeur of this life. - Albert Camus

          by Anne Elk on Sat Oct 27, 2012 at 10:09:58 AM PDT

          [ Parent ]

          •  If that is true then why would (1+ / 0-)
            Recommended by:
            Lujane

            the same effect be seen during the 2008 Presidential Election?

            why would Romney's proportion of the vote total increase in Urban Centers compared to Obama?

            •  I don't really know (2+ / 0-)
              Recommended by:
              HudsonValleyMark, Lujane

              but you are assuming that the large precinct effect in the primaries and in the election are being caused by the same thing. Just because you see the same outcome in 2 events doesn't require us to conclude a common cause.

              For if there is a sin against life, it consists perhaps not so much in despairing of life as in hoping for another life and in eluding the implacable grandeur of this life. - Albert Camus

              by Anne Elk on Sat Oct 27, 2012 at 10:23:26 AM PDT

              [ Parent ]

              •  no, (1+ / 0-)
                Recommended by:
                Lujane

                I am just observing that it DID happen.

                I remember the lines out of Urban precincts and I know that Obama polled much better in Urban centers than McCain.

                •  Well, no, that cannot go unchallenged (4+ / 0-)

                  The person who wrote this title

                  Rigged Elections for Romney
                  and made this claim
                  there is absolutely no reason to believe that a single candidate would have his percentage of the vote total go up just because the precinct had more votes cast, unless there was vote switching going on.
                  cannot honestly make this statement
                  I am just observing that it DID happen.
                  in the same diary.
                  •  you are right (1+ / 0-)
                    Recommended by:
                    Quicklund

                    of course, that is what I believe to be true to the data.  Interesting as well, I found out from this project that the 12 largest precincts in Ohio reported votes for president that were on the order of 800 total votes but the tallies showed that 8000 people had voted in those locations.  

                    Here are the 15 largest precinct vote totals in ohio

                    BALLOTS  TOTAL
                    CAST        COUNTED
                    TOTAL

                    9640    959
                    8772    873
                    8383    836
                    7004    694
                    5748    562
                    5498    541
                    5246    515
                    5140    503
                    4015    395
                    2925    287
                    1684    1666
                    1607    1587
                    1540    1534

          •  here's the thing... (3+ / 0-)
            Recommended by:
            Quicklund, Argyrios, Lujane

            This "effect" occurs within the Des Moines city limits.

            The dynamics you describe may explain that (or maybe not). The obvious urban/rural controls won't work.

            It's not up to us to explain the dynamics. If someone really thinks that Romney stole votes throughout Des Moines in proportions that covaried with the total number of votes cast -- how? why? -- s/he has to make a plausible argument about that.

            I'm almost screaming about this because I am a grizzled veteran of the exit poll wars. A lot of exit poll results simply make no sense. It is not up to me to provide a comprehensive statistical account of why some exit poll results are crazier than others.

            The crucial move in CT is to reverse the burden of proof.

            Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
            Better Know Your Voting System with the Verifier!

            by HudsonValleyMark on Sat Oct 27, 2012 at 10:17:31 AM PDT

            [ Parent ]

            •  without causation assumed (1+ / 0-)
              Recommended by:
              Lujane

              cannot one recognize impropriety?

              In your exit polling, did you find that Obama would fare better in Urban Precincts (the ones that have higher numbers of votes cast on election day)?

              If that is the case, why would McCain experience a 5% boost from the 500 largest precincts when compared to his final state vote tally?

              •  ? (2+ / 0-)
                Recommended by:
                Quicklund, Argyrios
                without causation assumed cannot one recognize impropriety?
                I'm not at all sure what you mean, but isn't "impropriety" in fact an attribution of cause?
                In your exit polling, did you find that Obama would fare better in Urban Precincts (the ones that have higher numbers of votes cast on election day)?

                If that is the case, why would McCain experience a 5% boost from the 500 largest precincts when compared to his final state vote tally?

                Umm, whut?

                (1) I didn't do any exit polling.

                (2) You have offered no warrant for treating "Urban Precincts" and "higher number of votes cast" as synonymous.

                (3) If you think McCain's vote shares in the 500 largest precincts are fishy, have you even taken the time to locate those precincts?

                Once again, the burden of proof is not upon the rest of the world to convince you that your thrilling plots don't evince fraud. With all due respect, the rest of the world doesn't necessarily care what you think. If you think you have evidence of fraud, it's up to you to marshal an argument.

                Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                Better Know Your Voting System with the Verifier!

                by HudsonValleyMark on Sat Oct 27, 2012 at 10:31:01 AM PDT

                [ Parent ]

                •  here ya go (0+ / 0-)

                  highest 20 precincts

                  BUTLER    09-P-AJX    WEST CHESTER TWP WC41
                  LORAIN    47AJP    PRECINCT GRAFTON TWP #1
                  LUCAS    48-P-AVA    WATERVILLE 9
                  RICHLAND    70-P-ADX    SPRINGFIELD TOWNSHIP C
                  SUMMIT    77AHU    PRECINCT AKRON 9-L
                  SUMMIT    77AUG    PRECINCT NFLD CTR TWP E
                  SUMMIT    77ASZ    PRECINCT NEW FRANKLIN 2-B
                  SUMMIT    77ATA    PRECINCT NEW FRANKLIN 2-C
                  SCIOTO    73-ACN    JEFFERSON TWP C
                  ASHTABULA    04030-ABD    CONNEAUT 1-C
                  FRANKLIN    25-BHX    COLS 82-E
                  PIKE    66018-AAR    PRECINCT PEE PEE TOWNSHIP
                  FRANKLIN    25-BCA    COLS 73-H
                  ATHENS    05-P-AAG    ATHENS 2-1
                  STARK    76-P-AGY    JACKSON TWP 26
                  FRANKLIN    25-BHY    COLS 82-F
                  CUYAHOGA    18-P-DBJ    OLMSTED TOWNSHIP-00- G
                  FRANKLIN    25-ARG    DUB 1-A
                  WOOD    87-P-ADG    PERRYSBURG TWP NORTH
                  STARK    76-P-AIG    LAKE TWP 4
                  FRANKLIN    25-BEM    JEFFERSON-D

                  Butler county is Hamilton/Akron area
                  Lorain County is Cleveland
                  Lucas County is Toledo
                  Summit County is Akron
                  Franklin County is Columbus

                  •  well, a core dump is a start, I guess (0+ / 0-)

                    So, is there anything suspicious about McCain's performance in any of those precincts?

                    Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                    Better Know Your Voting System with the Verifier!

                    by HudsonValleyMark on Sat Oct 27, 2012 at 10:52:05 AM PDT

                    [ Parent ]

                    •  Yes (0+ / 0-)

                      This is interesting

                      http://www.dailykos.com/...

                      looks like the 12 largest precincts dropped the "ones" column in the presidential vote totals.

                      when I add a zero in the ones column to the counts president vote counts for these precincts it looks like this:

                      BALLOTS   TOTAL
                      CAST         COUNTED
                      TOTAL   

                      9640    9509
                      8772    8550
                      8383    8270
                      7004    6850
                      5748    5485
                      5498    5275
                      5246    5069
                      5140    4886
                      4015    3923
                      2925    2789
                      1684    1666
                      1607    1587
                      1540    1534

                      --------------

                      After doing this and summing the 25 largest precincts with a total number of votes for president cast of 77,712
                      1.3% of the state total

                      total the breakdown is

                      McCain = 42,231 = 54.3% of the vote
                      Obama = 35,481 = 45.7% of the vote

                      final state total tallies (with the correction above)

                      McCain = 2,708,311 = 47.00% of the vote
                      Obama = 2,964,298 = 51.44% of the vote

                      Total delta (+McCain + -Obama)

                      =13.07% vote swing

                      •  huh? (0+ / 0-)

                        I'm not seeing a lot of zeroes in your ones column. Also, did you look at the registration figures and candidate counts in those precincts? Are you sure that there were over 77,000 votes cast for president in these precincts?

                        Setting that aside, do you know anything about these precincts that would lead you to suspect "vote swing"? I realize that you're just getting started -- except that, at the same time, you've already asserted your conclusions.

                        Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                        Better Know Your Voting System with the Verifier!

                        by HudsonValleyMark on Sat Oct 27, 2012 at 11:34:46 AM PDT

                        [ Parent ]

                        •  Hah! (0+ / 0-)

                          You are right, the problem was with the reported votes, the actual votes counted was ok

                          Here is the new data

                          in the 15 largest precincts McCain had 51.82%
                          and Obama had 47.06%

                          total votes cast was 11,075 for McCain
                          and 10,058 for Obama

                          in the 1,230 largest precincts McCain had a total of 53.3% and Obama had 45.35%.

                          total votes cast was 559,925 for McCain
                          and 476,438 for Obama

                          the top 1,230 largest precincts accounted for 1,050,593 total votes or 18.4% of the state total.

                          •  OK, that looks about right (0+ / 0-)

                            Now, when I split the precincts into quintiles by votes counted for president, I see a monotonic increase in McCain's vote share. Yes?

                            Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                            Better Know Your Voting System with the Verifier!

                            by HudsonValleyMark on Sat Oct 27, 2012 at 11:57:11 AM PDT

                            [ Parent ]

                          •  heck, we don't even need quintiles (0+ / 0-)

                            You stand by your graph, yes? (It looks OK to me, although the choice of scale might be construed as quirky.)

                            So, the relationship under discussion does not appear "only in large precincts," yes?

                            That might have some bearing on your attempts to explain it, if indeed you do attempt to explain it.

                            Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                            Better Know Your Voting System with the Verifier!

                            by HudsonValleyMark on Sat Oct 27, 2012 at 12:17:44 PM PDT

                            [ Parent ]

                          •  quintiles (0+ / 0-)

                            average mccain by precinct size quintile

                            http://i46.tinypic.com/...

                            also on the blog

                          •  It looks like the predictor (0+ / 0-)

                            for higher GOP returns for larger precinct sizes is scalable to the size of the precinct, on average.

                            so, yes not on a precinct by precinct basis but on ALL precincts on Average due to their size.

                            This would be the most effective way to hide a vote flipping scheme when smaller precincts are more susceptible to a recount verification.

                            (theory)

                          •  really? (0+ / 0-)
                            This would be the most effective way to hide a vote flipping scheme when smaller precincts are more susceptible to a recount verification.

                            (theory)

                            That seems pretty close to saying, "If this were a great way to hide fraud, then it would be a great way to hide fraud!"

                            I think it would make some sense if the supposed "swing" were heavily concentrated in a small fraction of large precincts -- at least in states where there is some sort of simple-random-sample audit of the primary results, however many states that may have been. But that isn't at all what we see here. What we see here is a "pattern" that is trivially easy to pick up in correlational analysis, and that confers no obvious advantage to a hacker.

                            Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                            Better Know Your Voting System with the Verifier!

                            by HudsonValleyMark on Sat Oct 27, 2012 at 03:10:14 PM PDT

                            [ Parent ]

                          •  Interesting (0+ / 0-)

                            Regardless of how easy it is to determine this bias of larger precincts toward the GOP, I would wonder what other methods could be used to skew the vote, if one were so inclined.

                            It seems to me that, if there was a way to flip votes, and the goal was to produce a 10% swing in the direction of the favored candidate (flipping 5% of the total votes) then, one would HAVE to focus on the larger precincts.

                            What is the error margin of the random sample?  I would bet that it is 90/5.

                            Does Ohio even perform a random sample check?

                          •  it isn't obvious why one would have to focus (0+ / 0-)

                            Every time one decides not to attack one precinct, one has to flip that many more votes in some other precinct. That may or may not be expeditious depending on what one is up against.

                            Does Ohio even perform a random sample check?
                            As a result of a legal settlement, it presently does in presidential primaries and in even-year general elections including the presidential contest. This year, it's a simple random sample including at least 5% of votes in each county. So, packing the fraud into a small number of precincts could be a good play -- although Ohio also lets candidates ask for partial recounts -- but smearing it up and down the size distribution doesn't seem to confer much advantage. The specifics would depend on the state.

                            Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                            Better Know Your Voting System with the Verifier!

                            by HudsonValleyMark on Sat Oct 27, 2012 at 04:34:14 PM PDT

                            [ Parent ]

                  •  oh, here's something... (0+ / 0-)

                    I would guess that the "ballots cast" figures in many of those precincts is off by a factor of 10, or so.

                    Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                    Better Know Your Voting System with the Verifier!

                    by HudsonValleyMark on Sat Oct 27, 2012 at 11:01:02 AM PDT

                    [ Parent ]

                    •  It turns out that (0+ / 0-)

                      the totals go up even more if you correct it

                      for the 12 highest vote precincts before correction

                      McCain =54.48%   
                      Obama = 43.63%

                      Note: independents had a greater effect before correction

                      after correction

                      McCain = 55.42%       
                      Obama = 44.39%

                      •  umm, so what? (1+ / 0-)
                        Recommended by:
                        librarianman

                        I am still waiting for a substantive response to my first comment in this diary: what is the argument, at all?

                        Is there any reason why the 12 largest precincts in Ohio shouldn't differ in partisan composition from the state at large? None is apparent.

                        The "trend" under discussion seems to occur across most of the range of votes cast. Do you have any explanation for it?

                        Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                        Better Know Your Voting System with the Verifier!

                        by HudsonValleyMark on Sat Oct 27, 2012 at 11:43:28 AM PDT

                        [ Parent ]

                        •  The kernal of this diary can be stated thusly: (0+ / 0-)

                          The percentage of votes cast within a precinct is a determining factor in deciding how well the GOP candidate (2008 election) or Mitt Romney (2012 primary) will do.

                          And

                          This determinant has been identified in states all over the country, and can be reasonably isolated from other contributing factors.

                          •  don't try that in an intro stats class (0+ / 0-)
                            The percentage of votes cast within a precinct is a determining factor in deciding how well the GOP candidate (2008 election) or Mitt Romney (2012 primary) will do. [emphasis added]
                            Especially don't try it if you titled the paper, "Rigged Elections for Romney."

                            You have neither demonstrated causation nor even proposed a plausible causal mechanism.

                            Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                            Better Know Your Voting System with the Verifier!

                            by HudsonValleyMark on Sat Oct 27, 2012 at 12:04:40 PM PDT

                            [ Parent ]

        •  You assert therefore that (0+ / 0-)

          So I am supposed to believe that, on a national level, the single most determining factor whether a precinct votes overwhelmingly more republican than the statewide totals is. . .

          the total number of votes cast within a precinct?

    •  Not sure if that would be a control (0+ / 0-)

      but worth looking into.

  •  This was done yesterday, the day before, and... (3+ / 0-)
    Recommended by:
    HudsonValleyMark, elmo, Argyrios

    This diary today does a much better job explaining the claims made. However the topic was covered in a diary yesterday.

    Anyway...

    This is an Reverse-Underpants Gnome Conspiracy.

    Step 1) A correlation is observed

    Step 2) Nothing happens (except)

    Step 3) A conspiracy is declared!

    There is a massive amount of work missing in step two. There are any number of other reasons to explain this circumstance; demographics, TV coverage, ad buys, vote collection methods et al. Much more work is needed before leaping to Step 3.

    So instead of worrying about some massive conspiracy which would require several participants from both parties in each of thousands of county election offices, not one of whom has ever talked, let's keep our noses down and GOTV shall we?

    •  Yesterday's diary link (1+ / 0-)
      Recommended by:
      elmo

      Dang it I intended to link to the comment section of yesterday's diary on the same exact topic.

    •  I just found the trick (3+ / 0-)
      Recommended by:
      elmo, Anne Elk, librarianman

      I thought these graphs plotted the voting proportion versus the number of votes cast per county, after sorting on county size.

      They do not.

      They plot the cumulative vote totals as you add each county one-by-one, starting with the smallest and ending with the biggest.

      This pre-sorting approach is the trick. It masks the effect of small counties and magnifies the effect of large counties. This is simple common sense when you think about it.

      Let's say a small county goes for Romney 55, Santorum 35, and Gingrich 20. That's 55%/35%/20% = 100% for the small county.

      A large county goes 600/330/170 which is 60%/33%/17% - 100% Not too much different but definitely better for Romney.

      The cumulative total for this is R 655 / S 365 / G 190 or 59.5% / 33.2% / 17.3% = 100%.

      Do you see how the cumulative breakdown  is very close to the large county breakdown taken by itself? The cumulative total is dominated by the large counties.

      When you plot the cumulative total in ascending order, that will tend to smooth out the ups and downs. A medium sized country that went for (say) Santorum would not cause the graph line to dip sharply at the point, because it is smoothed out by all the other votes.

      This creates a visual graphic trick of smoothing out what would otherwise be a jagged up-and-down track.

      The key point is this: there is no logical or statistical reason to examine the cumulative vote curve. If one is looking to correlate county size to voting portion, those are the two values which should appear on the plot.

      •  That is incorrect -- see the 2008 data (0+ / 0-)

        which is why I also sorted them Biggest to Smallest.

        in the 500 Biggest precincts (precincts are different from counties) McCain won 53% of the total vote cast and Obama won 46% of the total cast.

        •  Show me 1948 data (0+ / 0-)

          Sorting top to bottom or bottom to top does not change anything except to make a mirror image.

          There's no reason to plot the cumulatives.

          But before you make your case, how about plotting data from election years before the advent of machines. If this data artifact exists in every election that would make for a much more impressive conspiracy theory.

          •  the reason you plot the percent of cumulative (0+ / 0-)

            is to determine if the number of votes cast within a precinct affected how well each candidate would do in that precinct.

            in this case it does.

            and it is consistently for Romney and/or GOP.

            Nationwide.

            •  You can so that with individuals not cumulatives. (0+ / 0-)
              •  sure (0+ / 0-)

                if I wanted to show 9,000 individual graphs

                •  SCATTERPLOT (2+ / 0-)
                  Recommended by:
                  New Minas, Quicklund

                  Hello?!

                  Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                  Better Know Your Voting System with the Verifier!

                  by HudsonValleyMark on Sat Oct 27, 2012 at 10:50:38 AM PDT

                  [ Parent ]

                  •  It is up (0+ / 0-)

                    Link to the Graph
                    http://i50.tinypic.com/...

                    SUMMARY OUTPUT                               

                    Regression Statistics                               
                    Multiple R    0.188666332                           
                    R Square    0.035594985                           
                    Adjusted R Square    0.035508133                           
                    Standard Error    0.180714868                           
                    Observations    11106                           

                    ANOVA                               
                        df    SS    MS    F    Significance F           
                    Regression    1    13.3843282    13.3843282    409.8347753    1.62542E-89           
                    Residual    11104    362.6329177    0.032657864                   
                    Total    11105    376.0172459                       

                        Coefficients    Standard Error    t Stat    P-value    Lower 95%    Upper 95%    Lower 95.0%    Upper 95.0%
                    Intercept    0.34896164    0.005042444    69.20485816    0    0.339077554    0.358845726    0.339077554    0.358845726
                    1684    0.000182195    8.99979E-06    20.24437639    1.62542E-89    0.000164554    0.000199836    0.000164554    0.000199836

                    •  OK (1+ / 0-)
                      Recommended by:
                      Quicklund

                      Smaller plotting characters might help a bit, but that's fine.a I'll let Quicklund reevaluate what he thinks of the cumulative plot in light of this one. However, it's pretty clear that we aren't talking about a trend concentrated in large precincts. In fact, the residuals on the right side of the plot are very disproportionately negative.

                      Just as an aid to understanding, you may want to superimpose a loess line or something like that. It probably wouldn't look radically different except for leveling off at right.

                      Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                      Better Know Your Voting System with the Verifier!

                      by HudsonValleyMark on Sat Oct 27, 2012 at 12:21:42 PM PDT

                      [ Parent ]

                      •  As for the cumlatives (0+ / 0-)

                        The data set includes extremely small precincts. These have little to no bearing on the outcome of the election. It is hard to imagine them as part of a nationwide conspiracy. But when you include them in the original cumulative plots they do establish the X-axis. It's sort of like a forced X-axis intercept in a regression calculation ... is my guesstimate.

                        I suspect the cumulative 'tendlines' might not have the same visual impact were the precincts <100  votes excluded.

                        •  eh, I don't think I would censor them altogether (1+ / 0-)
                          Recommended by:
                          Quicklund

                          I'm reluctant to say that "the whole point of the theory is X" when I don't think there is much of a theory. But rather than drop small precincts that they expect to be exempt from fraud -- and the really tiny ones won't amount to many votes -- it seems reasonable to focus on using consistent scales and measures. C&J actually have some measures, which I didn't try to figure out in detail because I was too alienated by the overclaiming.

                          Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                          Better Know Your Voting System with the Verifier!

                          by HudsonValleyMark on Sat Oct 27, 2012 at 05:48:22 PM PDT

                          [ Parent ]

                    •  Thank you sincerely (0+ / 0-)

                      My first few impressions:

                      1) An R-sq of ~0.04 yells out 'no correlation here'. The data cloud if anything has a vertical structure. A vertical structure is what you'd expect if precinct size played no role in vote share. There might be a bit of a 'McCain bulge' above the 700-800 vote level, but it's not a clear bias. Thus we return to the low R-sq.

                      2) It looks like 6 of the 10 largest precincts went to Obama not McCain. Going to the top 20 it looks like a bigger Obama advantage. That's not consistent with the theory.

                      3) I used the term counties earlier because that is usually where the counting is managed. But the fact this uses precincts is important. And there is something about these data that strikes me as odd.

                      One 'junk' precinct gave McCain 100% of the vote. (And thank you sir.) A few the same for Obama. Other than that we have one outlier for McCain at about a 97% share. And we have a solid blue foot for the data cloud. It's solid blue all the way down. The Y=0 axis itself is solid blue between ~225 and ~575.

                      I wonder that there are scores of precincts of ~500 voters per, each of which is delivering 95% to 100% of their votes to any candidate. This makes me wonder what's been plotted here.

                      4) There are several so-called 'junk' precincts. These are often industrial areas and such with few sometime zero residents. You can see them on or very near the x-axis. Here is a good time to remember out topic before applying our math.

                      If we are looking for vote-rigging, we are not likely to find it in precincts with 10 votes total. Much more worrisome and likely are attempts in the larger areas.

                      Myself, I would exclude the precincts with fewer than 100 votes. (Or rather, I would do those in a separate analysis.) That allows the left flank of our data analysis te rest upon most statistically-significant data points.

                      Bottom line though, this scatterplot if anything tends to discount the vote-rigging method alleged.

                    •  Different scatterplot'd be more imformative (0+ / 0-)

                      With X as Raw total Obama votes and Y as Raw Total McCain Votes. That's eliminate junk 0 vote precincts and be more informative than a 'McCain Share' pre-calculation. (A 45 degree line would serve as a handy 50%/50% visual gauge.)

                      •  That would preclude (1+ / 0-)
                        Recommended by:
                        Quicklund

                        a breakout according to precinct size which is the point.

                        When talking about R2 being low you are assuming that 100% correlation would mean that vote turnout is strictly determined by precinct size.  What I am saying is that there is a slight adjustment to the data which favors McCain if the precincts are larger.  

                        The line shows that the best fit curve for this value is that McCain will, On average, get 60% of the vote if the precinct has 1400 votes and only 40% of the votes if the precinct has 400 voters.  And that 3.5% of the total data points fit the curve with 100% accuracy.

                        which would mean that there is a slight bias toward McCain if the precinct is larger.

                        You simply won't get much more out of this, I am afraid.  I was thinking of correlating the votes to party votes if more registered democrats voted republican in larger districts then that could mean something.

                        •  No it wouldn't (0+ / 0-)

                          Asymetries in the data cloud distributions would indicate if there was something to the proposition.

                          Personally, I suspect the raw data used in this study. There is something wrong when scores or even hundreds of precincts go 100% (or nearly so) for Sen Obama and 2 precincts do so for Sen McCain.

                          •  here ya go (1+ / 0-)
                            Recommended by:
                            Quicklund

                            tell me what you think

                            http://i48.tinypic.com/...

                            p.s. the zero mccain votes in many precincts indicate how some locations have zero mccain support due to race/class isolation.

                          •  Thank you again very much (0+ / 0-)

                            You are a good egg.

                            There's two takeaways from this plot IMO.

                            1) There's something funky about the data. The Y-axix has solid blue, but the data cloud slopes away from the X-axis. It is hard to imagine so many 100% Democratic precincts and no 100% GOP examples. Here we need to look at specific matchups of the data to specific precincts to find out what is going on with these data.

                            2) If there wee no bias we'd expect to see a square distribution. That is roughly the distribution we see here,after one accounts for the Y-axis having roughly twice the resolution per inch of the X-axis. That is, what would be a square appears on this graph as a horizontal rectangle aka a 'stretched' square. There is some deviation from the ideal but not much. So whatever share/size effect there is, if any, does not seem strong.

                            Personally, I would want to clear up the discrepency between the blue Y-axis and the clean X-axis before I used thee data to make any conclusions whatsoever. One cannot trust the results unless one trusts the raw data.

                            Thank you for your responsiveness and willingness to incorporate alternative explanations.

                             

            •  And if it truly a nationwide thing (1+ / 0-)
              Recommended by:
              HudsonValleyMark

              That points to the likely explanation being some artifact of the methodology rather than the aforementioned vast conspiracy requiring thousands of culprits from all parties and walks of life not one of whom has ever screwed up or let an ill-considered word slip.

            •  seriously, have you looked at a scatterplot? (1+ / 0-)
              Recommended by:
              Quicklund

              And, by the way, how would any bivariate method establish the following?

              if the number of votes cast within a precinct affected how well each candidate would do in that precinct
              Did I mention that correlation doesn't prove causation?

              Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
              Better Know Your Voting System with the Verifier!

              by HudsonValleyMark on Sat Oct 27, 2012 at 10:37:12 AM PDT

              [ Parent ]

      •  I wouldn't say it's a trick (1+ / 0-)
        Recommended by:
        elmo

        It's certainly a reason not to obsess over whether a particular curve happens to look straight -- especially when others don't, and you apparently think that they all evince fraud.

        Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
        Better Know Your Voting System with the Verifier!

        by HudsonValleyMark on Sat Oct 27, 2012 at 09:52:01 AM PDT

        [ Parent ]

        •  It is an uneeded step (0+ / 0-)

          Which makes me wonder why it was included. To smooth the curves is my suggestion.

          Look at the 2nd chart, for New Hampshire. The lime green curve shows several shallow valleys. if this curve were not smoothed by the cumulative effect these shallow valleys would plot out as deep reversals of the "steady trendlines" this "proof" relies on to make its case.

          To me that is a trick. The creator of these graphs could not sell the "steady trendlines" theme if the graphs jagged up and down like a Lehman Brothers' Going-Out-Of-Business sale. So a curve-smoothing trick was found to punch up the visual aids.

          If I was employed to crunch numbers for a political outfit I'd be quite willing to look at these data to see if there are lessons to be learned. But if I worked for the DoJ I would not be calling for a subpoena based on this thin evidence. These data are not uninteresting. They just do not rise to proof of election rigging.

          •  I concur in the judgment (1+ / 0-)
            Recommended by:
            Quicklund

            Actually, I would put it more strongly, but you already did that elsewhere.

            Ideally, I think it's reasonable to look both at scatterplots with some sort of smoother overlay and at cumulative plots like these -- where small precincts actually occupy less "space." But that's in a world where we're actually interested in politics, not just in pitching fraud theories we haven't even really formed yet.

            Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
            Better Know Your Voting System with the Verifier!

            by HudsonValleyMark on Sat Oct 27, 2012 at 10:43:17 AM PDT

            [ Parent ]

            •  Better put (1+ / 0-)
              Recommended by:
              HudsonValleyMark

              The cumulative graphs are legit math and as you say, maybe there's a reason to show that particular perspective on the data. But it's sort of like playing around with the resolution of the y-axis. A curve plotted on a scale on 0-100 might look smooth. But plotted at a scale of 35-40 the same curve would appear to zig-zag up and down.

              I guess I'm more into psychology than math here. If one is, hypothetically speaking, someone disinterested in making a good-faith case, one would might publish their data in a form that creates neat visual reinforcement to the desired theme. Those sorts of old presentation tricks is what I see at work here.

              •  yes, and perhaps also some self-deception (1+ / 0-)
                Recommended by:
                Quicklund

                I just looked at the new diary. One interesting thing is that Duniho at least sounds more reasonable than the other people involved -- he isn't making the wild claims of proof. (That makes for an interesting disconnect between what he says and what Denis says.) Another is that at least one of his "flat" plots doesn't seem flat to me at all.

                If you more or less expect some plots to be flat and others not to be -- and if you're using a method that tends to create lots of noise on the left side, so there is always some justification for tweaking the scale -- it is very easy to produce the Right chart, imposing your priors upon the data.

                Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                Better Know Your Voting System with the Verifier!

                by HudsonValleyMark on Sat Oct 27, 2012 at 04:59:49 PM PDT

                [ Parent ]

                •  I think you are both missing the point (0+ / 0-)

                  The author of this diary (me) and the author of the original work (re: New Hampshire plot) is not trying to say that there is a steady trendline.

                  At least not one of strict correlation between total votes cast and the final outcome per precinct.

                  what he (and I) am saying is that, the larger the precinct, there is a statistically significant chance that it will come in higher than the statewide average for either Mitt Romney (2012 Primary) or McCain (2008 presidential) (note:  I also did the 2010 election results with identical findings)

                  AND

                  That the amount that the precinct comes in above the statewide totals is going to correlate to the overall size of that precinct by a significant margin.

                  This is beyond strict scaling.  

                  I have checked, the larger precincts come from URBAN areas, traditionally democratic strongholds.  This correlation is completely against what one would expect.

                  Given that the variance is on the order of 10% for the very large precincts, how would you explain it?

                  •  In other words (0+ / 0-)

                    There is a skew on the vote totals, a small bias toward the GOP which would appear most significantly when one shows the cumulative error per precinct (this error is above what it would normally be)

                    so:

                    if the precinct was large but went for obama, it went less for obama than it would have without the bias, on average by about 5% for the larger precincts.

                  •  This is interesting but quite preliminary (0+ / 0-)

                    I am not sure the premise has been established. But were I in the politics business I'd task my stats team on this to see what could be learned. But much more work remains even to establish the premise a precinct size bias much less a meddling scenario.

                    If I am in charge of a DoJ unit I'd be happy to do the same thing, if I could afford to. I assume my ficticous political alter-ego has a much larger budget. ;)

                  •  I don't really think you've made a point yet (1+ / 0-)
                    Recommended by:
                    Quicklund

                    Quicklund and I have been politely 'debating' how to present the data, but I agree that that is a side point.

                    Your diary did say this:

                    I mean, there is nothing in the laws of physics or politics that says a specific candidate should, AT A CONSTANT RATE, gather higher and higher percentages of the vote according to the number of votes cast within a precinct.

                    And then it became obvious, on a national level, the votes are being switched.

                    Sounded like you were trying to say there was a steady trendline, at least in WI -- although, as I commented hours ago, I don't see why that should matter anyway. It also sounded like at some point, you took an imaginative leap that doesn't have much to do with causal inference.

                    If you think you see Democratic precincts in Ohio going for McCain, you might have a finding. But you haven't actually made that case.

                    Otherwise -- yes, yes, it has been apparent from the first read of your diary that there is a correlation; you're convinced that the correlation is important and in fact evinces fraud; you haven't presented a coherent argument to support your conviction; and you're trying to shift the burden of proof to the rest of us.

                    There is no reason to expect the small precincts to be just like the large precincts; so far, there is nothing to rebut.

                    Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                    Better Know Your Voting System with the Verifier!

                    by HudsonValleyMark on Sat Oct 27, 2012 at 06:04:38 PM PDT

                    [ Parent ]

                    •  Agreed (0+ / 0-)

                      The "constant rate" I was talking about is an average one with decreasing effect as the precinct size decreases.

                      I am looking at the 2010 gubernatorial race for Ohio scatter plots right now, they show a huge bulge at the bottom of the first quintile and the beginning of the second for the republican.

                      I cannot fathom why larger vote turnout precincts would favor the GOP so decidedly.  I always believed that higher turnout precincts were in urban areas (and indeed the counties they are in are all urban) favored the democratic candidates.

                      I will post the scatter plots for your perusal, thanks for the feedback guys.

                      •  don't confuse yourself re: "turnout" (0+ / 0-)

                        Actually, it could potentially be helpful: the fact is that turnout (as a % of registered voters) often is lower in urban areas. That seems to be largely because people are more likely to have moved. But the number of votes counted depends on both the number of registered voters and the turnout among them. It's a highly confounded variable. That is why it is hard to get excited, a priori, about a correlation involving it.

                        You are still stuck at point 1. You may always have believed something about the relationship between urbanism and number of votes per precinct, but your prior belief isn't directly pertinent to empirical research. (And I don't think this has much to do with "urbanism" anyway. A precinct in Columbus ward 70-something isn't likely to be much like a precinct in ward 9.)

                        Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
                        Better Know Your Voting System with the Verifier!

                        by HudsonValleyMark on Sun Oct 28, 2012 at 05:08:27 AM PDT

                        [ Parent ]

              •  for you too (see below) (0+ / 0-)

                The author of this diary (me) and the author of the original work (re: New Hampshire plot) is not trying to say that there is a steady trendline.

                At least not one of strict correlation between total votes cast and the final outcome per precinct.

                what he (and I) am saying is that, the larger the precinct, there is a statistically significant chance that it will come in higher than the statewide average for either Mitt Romney (2012 Primary) or McCain (2008 presidential) (note:  I also did the 2010 election results with identical findings)

                AND

                That the amount that the precinct comes in above the statewide totals is going to correlate to the overall size of that precinct by a significant margin.

                This is beyond strict scaling.  

                I have checked, the larger precincts come from URBAN areas, traditionally democratic strongholds.  This correlation is completely against what one would expect.

                Given that the variance is on the order of 10% for the very large precincts, how would you explain it?

  •  Can we just purge the lunatics out of here? (1+ / 0-)
    Recommended by:
    Argyrios

    We've purged people who weren't nearly as pernicious. Lunacy doesn't become us.

    Romney '12: Berlusconi without the sex and alcohol!

    by Rich in PA on Sat Oct 27, 2012 at 09:35:55 AM PDT

  •  Frightening, but (1+ / 0-)
    Recommended by:
    Quicklund

    if indeed, this is compelling statistical evidence of cheating because of the probability..., then someone should and will go to the courts and get warrants for the machines, more importantly the codes that operate them. Romney has no shame and would do absolutely anything, but this would require too many people being in on the fix to ever keep secret over time. My points are that even with this techonology, evidence will exist about what happened and eventually conspirators talk. As other commentors have said, there are also other possible explanations for the phenomenon.

    You may recall that the online poker cheaters were caught by statistical analysis demonstrating impossible results for which only one explanation could exist for the defiance of the outcomes. If this analysis becomes that tight, I am sure investigation will go on and truth will be known. If not this will continue to be conspiracy theory held by many, especially if O happens to lose (not gonna happen).

    If we don't change, we don't grow. If we don't grow, we aren't really living. - Gail Sheehy

    by itisuptous on Sat Oct 27, 2012 at 09:40:35 AM PDT

  •  It seems to me that both parties have CTs (5+ / 0-)

    On the left, there is a persistent belief that shadowy forces are rigging voting machines.  On the right, there is the persistent belief that "brown folk" who are not entitled to vote nevertheless engage in massive voter fraud.  Both are ludicrous.

    It is extremely unlikely that there is a nationwide conspiracy of hundreds or thousands of poll workers and other election officials who are knowingly engaging in serious criminal activity with no apparent benefit to themselves (it's not like they're going to get wads of secret cash from Romney after the election).  It is also extremely unlikely that so many people could keep such a secret, or that some written evidence (a letter, an email, etc.) would never be disclosed.  This theory, such as it is, also does not take into account differences in voting machines or differences in how those votes are counted, verified, and reported between states and counties.  Sometimes people just need to use a little common sense.

    The pleasure of hating...eats into the heart of religion...[and] makes patriotism an excuse for carrying fire, pestilence, and famine into other lands. - W. Hazlitt

    by rfahey22 on Sat Oct 27, 2012 at 09:41:17 AM PDT

    •  absolutely -- with a caveat (2+ / 0-)
      Recommended by:
      elmo, Quicklund

      Different hacking scenarios imply various scopes of conspiracy. This one, indeed, would be grandiose. I don't know if any of the Iowa caucuses used scanners, but surely most did not. The notion that votes were stolen all over the state without anyone noticing doesn't make much sense. (One precinct did have the wrong results -- and that was noticed.) And that is just one state.

      Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
      Better Know Your Voting System with the Verifier!

      by HudsonValleyMark on Sat Oct 27, 2012 at 09:49:25 AM PDT

      [ Parent ]

  •  what is the argument here, at all? (2+ / 0-)
    Recommended by:
    Quicklund, elmo

    It seems that almost all these plots show more or less monotonic increases -- some more linear than others -- across almost the entire range of cumulative votes, ergo of vote counts.

    So, obviously this doesn't evince vote miscount concentrated in large precincts. If it is miscount, it is miscount scattered indiscriminately in large fractions of precincts in, apparently, 49 different states.

    Somehow the authors have convinced themselves that this is a masterful scheme to avoid detection. I don't see how that makes any sense.

    Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
    Better Know Your Voting System with the Verifier!

    by HudsonValleyMark on Sat Oct 27, 2012 at 09:41:25 AM PDT

  •  Oy (3+ / 0-)
    Recommended by:
    Quicklund, Argyrios, Cat Whisperer

    How many more of these diaries are we going to be seeing between now and election day? Dozens? Hundreds?

  •  If You Are Right Dems to Blame For Not Protecting (1+ / 0-)
    Recommended by:
    New Minas

    If you are correct and the precious day's allegations regarding Kerry's Ohio defeat in 2004; then the Dems are to blame for not protecting us. They have had time to investigate with the Justice Dept etc and even if they couldnt prove it but still thought the Repubs were guilty; they should have exposed it and put the most sophisticated poll watchers in the large precincts and put the Repubs on notice and their vote counting companies; criminal charges would be filed.

    It would seem to me if the Feds took this seriously; it wouldnt be too hard to find someone who was aware or in on the fix and in exchange for immunity to prosecution;had them 'sing'.

  •  McCain vs Obama by precinct graph (0+ / 0-)

    On the McCain vs Obama by precinct graph,  it looks like the axis titles are on the wrong axis.  Just switch the Obama and McCain titles to fix it.

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site