Politico Falls for Phony Stats

As we write, the loser of the South Carolina Democratic primary for US Senate, Vic Rawl, supposedly has a team of volunteer academic experts conducting analysis on the votes that led to his unexpected defeat. His spokesman tells Politico that they may have found something to hang his battered hat on: absentee ballots.

Politico reports that the so-called experts (who don’t want to be named yet) have found “one potential red flag: A significant difference between the results of absentee and election day ballots.”

The story explains:

“The election day ballots all favor Mr. Greene. We don’t know what it means,” [Rawl's campaign manager Walter] Ludwig said in an interview. “We did significantly better on absentees than Election Day, which is according to the mathematicians, quite significant.”

If David Catanese, the writer of the Politico story, had only bothered to check the official results himself, he could have found he was being led down the primrose path.

It is true, as Politico writes, that…

In Lancaster County, Rawl won absentee ballots over Greene by a staggering 84 percent [424 votes] to 16 percent [81 votes] margin; but Greene easily led among Election Day voters by 17 percentage points [863 votes to 602 votes].

But Ludwig’s trying to claim up is down, black is white, in this scenario. If anything, it’s Rawl’s freaky success in the absentee ballots that’s fishy as opposed to the more normal vote disbursement seen in Greene’s victory on election day. Who’s going to believe a relatively unknown political hack is going to exceed the landslide 80%+ vote percentages that only nationally known rock stars Senator Jim DeMint and Congressmen Joe “You Lie” Wilson and Jim Clyburn were able to attain.

After all, Rawl is from Charleston, on the coast in lower SC, while Lancaster is on the north-central border of the state. What reason has Lancaster to be wild for Rawl? Even in his Charleston home territory, he could only achieve 68% of the absentee ballots. He took the election day voting here by a lackluster 53% for an area where he should have had a ready-made GOTV operation staffed by his long-term supporters. Clearly, even where he is best known, he’s not as popular as he’d like to think he is.

Ludwig snookered Politico when he claimed that the absentee/election day ballot totals flip-flop “didn’t happen in any other races on the ballot.” Yet, in Rawl’s very own preferred example county of Lancaster, the Democratic governor’s race shows Robert Ford earning a whopping 72 absentee ballots [13.5%] to Jim Rex’s 36 [6.7%] (and winner Vincent Sheheen’s 426 [79.8%]). But on election day, voters’ 2nd place love for Ford became fickle. Ford garnered a measly 193 votes [12.1%] in comparison to Rex’s 283 [17.8%] and Sheheen’s 1118 [70.1%].

By not checking the data, Politico further failed to discover that the same phenomenon occurred in the Democratic primary for State Superintendent of Education, in which the alphabetically winning candidate Frank Hollerman racked up only 33.5% of the Lancaster County absentee vote, but scored 59.3% at the same county’s polls.

We do agree with Rawl’s campaign on the claim that something’s rotten in Lancaster County, but it involves the ballots that Rawl won on, not lost. Inspectors looking for voter fraud may want to take a look at the interesting numbers presented in the totals for the Democratic candidates for the Lancaster County Council’s District 2. This council district represents five of the county’s 29 precincts. However, a total of 530 out of the countywide 620 absentee ballots come from these precincts. [Out of the 916 total votes cast in the council race, a whopping 57.9% of them were by absentee. That's practically implementing Oregon's by-mail-only election system.]

Contrast the fact that 25.6% [or 505] of the total 1,970 votes cast in Lancaster County’s Democratic US Senate primary were done by absentee, whereas a minuscule 3.2% [or 89] of the total 2,809 votes cast in the corresponding Republican primary were by absentee. In the super-heated Republican gubernatorial race, only 95 absentee ballots were cast in the same county. Perhaps someone should be looking into the massive discrepancy between the absentee voting patterns of the two parties and the implications for potential fraud.

We don’t have enough staff to pore over the election results of every precinct in the state like Rawl’s three national teams. But Ludwig may want to consider hiring us to check the other  counties for him, as it seems that his teams of “experts” are drawn from a pool of SC Democratic voters.

That brings us to the experts’ second argument: “In Spartanburg County, Ludwig said there are 25 precincts in which Greene received more votes than were actually cast and 50 other precincts where votes appeared to be missing from the final count.”

The claim of 25 precincts having over-votes for Greene appears to be completely without basis. It doesn’t take a “national academic expert” to check the facts. Any fool, including us, can go to the SC State Election Commission website to view the certified official vote counts in each race in each precinct in each county. Check the vote totals in the Democratic primary for US Senate in Spartanburg County against the total votes cast on a Democratic ballot in those same precincts. You’ll find that not only do no Greene vote totals exceed the total votes cast, but never do the combined Greene and Rawl totals exceed them either.

As far as the number of votes cast for a Democrat to oppose Jim DeMint rarely–if ever–matching the total number of votes cast, virtually no race on any ballot ever has the exact same number of votes cast as the number of people voting. Voters don’t have to choose a candidate in every race. They are free to show up to help just one candidate and shun all the rest of them if they wish.

An experienced politician like Rawl should know these things. Could it be that Rawl is merely exposing the straight-ticket mentality of Democrats, in that they don’t have to know anything about who they are voting for, just as long as they vote for the candidates they are told to vote for? Perhaps that makes them particularly lost in a primary where everyone has a “D” next to his name, and they have to resort to alphabetical voting in the absense of any further guidance inside the voting booth–unless they’ve snuck in the cheat sheet that their union or ACORN or other group has provided them.

Too bad Catanese opted to just transcribe Ludwig’s junk hypotheses and garbage math instead of doing some simple addition, subtraction and division to test numbers.

Or he might have paid more attention to Ludwig’s description of their campaign tactics to get out the vote, which included 300,000 spammy emails and a quarter million irritating robocalls. That right there seems to explain a lot of votes not for Greene, but against Rawl.

One interesting tidbit to conclude on: The vote totals on the SC State Election Commission website aren’t the only thing that continues to taunt the Rawl campaign. In some county result totals, Rawl is accidentally awarded the “winner” ribbon next to his name.

Unfortunately, the winner ribbon graphic doesn't make it so.

Update I: The Rawl campaign has released more information regarding their vote analysis. They have identified one of their three expert teams, saying:

“One of the teams was Dr. Walter Mebane of the University of Michigan and Dr. Michael Miller of Cornell University. Dr. Mebane is a professor of political science and statistics and a recognized expert in detecting election fraud. As of August 2010, Dr. Miller will be professor of political science at the University of Illinois, Springfield, and specializes in the analysis of election data.  Neither is affiliated with the Rawl campaign.”

The press release says that “Dr. Mebane performed second-digit Benford’s law tests on the precinct returns from the Senate race.” After describing what in the world that is, Dr. Mebane’s findings are summed up thusly: “The results may reflect corrupted vote counts, but they may also reflect the way turnout in the election covaried with the geographic distribution of the candidates’ support,” Mebane said.” In other words, we tried our best to find something here, but at best, it’s inconclusive.

Dr. Miller’s role is described as having “performed additional tests to determine whether there was a significant difference in the percentage of absentee and Election Day votes that each candidate received.” He reiterates the key points of the absentee analysis presented in the Politico story–except that instead of saying the anomaly occurred in no other races, he says it occurred in one other: Robert Ford’s race, but in the opposite direction.

There’s nothing in the release regarding the Lancaster and Spartanburg fairy tales published in the Politico piece. Perhaps that bit of wishful thinking comes from another expert team.

The campaign concludes their analysis-to-date wrap-up with an expression of concern for their non-findings of anything provably manipulated by saying:

“These findings concern the campaign, and should concern all of South Carolina. We do not know that anything was done by anyone to tamper with Tuesday’s election, or whether there may have been innocuous machine malfunctions, and we are promoting no theories about either possibility.”

Update II: Welcome readers from Chris Haire’s blog at the Charleston City Paper, and readers from the comments at FiveThirtyEight.com, which has some superior, super wonky analysis of the Greene/Rawl votes–though the conclusions tend toward the Left. Still, wish we’d come across this treasure trove earlier.

Our reply to Chris Haire: @haireofthedog *semi*personal blog in seedling stage, but w/Columbia U masters in journalism & an extra accounting BAS if that adds any cred

Update III: The SC democrats continue to push the bogus “25 Spartanburg County precincts having overvotes” line. Instead of looking like an accidental mistake, it appears that the Democrats are going with the adage: “If you tell a lie enough times, it will become the truth.” In fact, the phony stats have already taken root in liberal blogs. Has no one else has bothered to check the data that’s easily available online?

Here’s the press release from SC State Senator Phil Leventis that bases their case on the “25 precincts.”

Update IV: Silly us. We called the winning Democratic gubernatorial candidate by the wrong name, identifying him as Jim Sheheen instead of the correct Vincent Sheheen. We have corrected the error. Our apologies to Mr. Sheheen, and our thanks to the commenter that pointed it out to us. [We did get Sheheen's name right, however, in our endorsement of his opponent Robert Ford for governor on the Democratic ticket.]

About these ads
This entry was posted in Politics and tagged , , , , , . Bookmark the permalink.

13 Responses to Politico Falls for Phony Stats

  1. You suck says:

    You are an idiot.

  2. Idiot says:

    You suck.

  3. The Liberty Flag says:

    The best voter analysis I’ve seen on this wacky, only could happen in SC, story. Please keep digging until you hit the real facts—not what the pols think are facts.

  4. The Horned Frog says:

    I am familiar with electronic voting and some systems like NY and PA could be different than SC. In SC, is there a separate screen for each race that you have to touch the skip button if you don’t want to vote for that candidate?

    The biggest anomaly I see is that the undervote % is so low. This race should have a 30-40% undervote % because even in SC nobody knew who these two guys were. In fact I’m sure most democrats in SC didn’t even know that two men were running for Senate. In the old manual ballot, one has to “mark up” their ballot and it is easier to just skip the mark up process in races one knows nothing about and proceed to the races one wants to vote. Hence undervotes are prevalent in contests where the candidates are not well known.

    In computer systems where the screen shows one race at a time, it takes just the same amount of work to skip the race as it does to vote. So might as well touch of a name because it is just the same a hitting the skip button. This would not be a bug of the system but a feature. Greene may have been the victim of an innocuous computer feature.

    I for one have always favored the computer systems that show the whole ballot on one, two or three screens that show multiple races per screen to encourage undervotes where the voter is better served by undervoting than casting a random vote.

    • Horned Frog,
      In SC (at least in Charleston and Berkeley counties), they put as many candidates on a page as possible. One vote per page would indeed be tedious and ripe for problems.

      The Reader

  5. The Horned Frog says:

    The undervotes, or lack thereof, is a serious problem. If there were 200,000 votes for Democratic Governor I would expect 120,000 – 140,000 votes for this race. If the current undervote total is correct it may be the result of our PC generation and all the previous political conventional wisdom could be ripe for updating.

    Did someone or something replace undervotes with votes for Greene? SC needs to test the computer hardware and software.

  6. yarrrrr says:

    Horned, you’re going to have to give a reason why you think the undervote would by 30-40%… this race received fewer votes than any other statewide race… but it’s nowhere near that high…

    The people voting know the importance of the Senate primary… which ironically makes it more likely to cast a vote for it even though they are ignorant of the candidates than if this were a less prestigious position… In my state we elect judges and I often skip those questions because I don’t know, but it is very unlikely that I would ever skip something like the Senator question even if I’m ignorant of the candidates….

  7. Rich Rostrom says:

    There are some funky numbers in this election. I downloaded some results and poked around.

    There were two counties where more votes were recorded for governor than ballots cast (Darlington, 43 over; Marlboro 70 over).

    Darlington reported 215 absentee ballots – and 404 absentee votes.

    Marlboro reported 621 absentee ballots – and 948 absentee votes; of which 684 were cast for Democrat Vincent Sheheen. Also in Marlboro, Adamsville precinct reported 44 ballots and 65 votes – of which 44 were for Democrats.

    Going back to the state numbers: Kershaw County reported exactly 5,000 for Sheheen – 94.9%, when his statewide result was 59.0%. Though it is his home district, so maybe…

    It looks to me like a lot of semi-competent election workers, and maybe some fiddling.

  8. patrick says:

    Quickly: VINCENT Sheheen, not Jim Sheheen.

    Otherwise: Good one.

  9. Pingback: the 2010 Sooper Awards Ceremony! | SOOPERMEXICAN

Leave a Reply

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

WordPress.com Logo

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

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s