A reader writes:
I’m interested in exploring the SC Dem Party Chair’s claim that the winner’s ballot placement was responsible for Tuesday’s outcome. I know of some research that supports that claim (Brockman 2003; Koppell & Steen 2004—gated pdfs), but I haven’t seen anything that could support Ms. Fowler’s claim that ballot position resulted in Mr. Green’s double digit victory. Even in an extremely low information election, the research I’ve seen shows no more than a low single digit effect.
Here’s a little background on Greene, and here’s another piece by Kosuke Imai and Daniel Ho on ballot order, which finds similar effects (1-3 points) in primary elections.
All I have to go on here is some guesswork, but it seems plausible to me that ballot order could be an important factor here. This was a very low information race, it would seem. Greene’s opponent, Vic Rawl, only raised $186,000 for his campaign, which isn’t much money for a statewide race. The low salience of the race is also evident in the roll-off: 169,542 voted in this race vs. 188,576 in the Democratic primary for governor.
And I’m not sure that the potential ballot order effect is implausibly large. Assume for the moment that voters were essentially choosing at random between the candidates. That would imply a 50-50 outcome. The actual outcome was 58-41, which only implies that 8-9% of voters were influenced by ballot margin.
Another question is whether there was any information on the ballot that might have cued voters to choose Greene over Rawl. I wondered whether SC voters might have inferred the candidates’ racial background from the names of the candidates. I looked to see whether there was any relationship between Green’s percent of the vote in each county (data here) and the percent black in that county from the 2000 Census (data here).
There is a modest positive relationship, although it is not statistically significant. Ecological inference problems make this analysis suggestive at best, but I don’t see much happening.
I welcome other theories in comments.
UPDATE: See also Tom Schaller’s post at 538.
{ 21 comments }
Other theory: Florida can only be the “crazy” state for so long.
You can break it down further thanks to South Carolina’s great election site. To precinct level, and even get each precinct’s registration white vs. non-white.
I grabbed a couple of big counties to test it on the precinct level (I know it obviously would create a urban/rural divide in the districts I am looking at). I came up with a much stronger relationship than yours above when looking at non-linear relationship. Logarithmic and Power R2s of around .25.
Of course, both of our methods have one bigger problem, African Americans are more likely to be Democrats than Whites are. In 2008, African Americans made up 25% of the general election voters according to the exit poll but in the Presidential primary they accounted for 55% of the voters.
Using a probably methodologically unsound re-expression of African American % in each precinct, the Linear R2 went from .1 to .2.
Race may be a factor, but it seems like it is not the only one and it is difficult to isolate.
John,
How strong is the relationship between percent urban and percent black? I’ve been looking at these numbers too (looking for irregularities) and have been focusing on each county’s share of the statewide turnout (in the Democratic primary). Of those counties that gained the most over 2008, 7 of the top 10 gainers were among the top 10 most populous counties. The same pattern did not hold in the 2008 over 2004 numbers.* The variation across counties was much more random over those two elections (at least in terms of population).
*There was no Senate seat up in South Carolina in 2006.
Chris: Can you post those graphs somewhere and put a link in the comments? A link to some precinct-level data would also be helpful.
Josh: The short answer is that I don’t know. I don’t have extensive SC data, just what I gathered quickly for that post. Let me know what you find.
Voting machines. South Carolina uses touch screen voter machines that have been shown to be easily tampered with.
It would be interesting to see if the trend is more or less present for day-of voters (excluding absentee ballots). Voters filling out absentee ballots would have more information about the candidates, whereas voters at polling places would be more likely to take cues from ballot order or assumed race of the candidates.
Here are the graphs:
http://i45.tinypic.com/2jc8y0g.jpg
http://i47.tinypic.com/icmhz4.jpg
Here is the link for the precinct level results:
http://www.enr-scvotes.org/SC/16117/27822/en/select-county.html
And here is the link for precinct level enrollment:
http://www.scvotes.org/statistics/by_counties_and_precincts
The only problem on those you have to choose county by county.
On my charts I used the following counties, Aiken, Beaufort, Charleston, Greenville, Horry, Richland, Spartanburg and York.
This was a very low information race,
YES! I live in SC and let me tell you there was virtually no coverage of this race.
There was no other information on the ballot – just the candidates’ names. And Greene’s was first. I knew Rawl and voted for him. But I suspect the majority of voters hadn’t heard of either candidate. Unfortunately. I like Rawl, but he did very little campaigning, at least not in my part of the state.
Vicki, it seems unlikely to me that anyone would tamper with the voting machines to throw just this one race, instead of the primary to nominate the Democratic candidate for governor. I think that the Democrat Sheheen, who won the Dem gubernatorial primary, is going to give the GOP candidate a harder run than Vic Rawl would have given DeMint.
John,
Hi. As a voter in South Carolina, I figured I’d offer some on the ground details about the election. It definitely was a low information race. While Greene did no campaigning at all, Rawl did little campaigning either. Granted, $186,000 isn’t a lot of money raised, but he ran no TV ads, had no signs (in the Columbia area at least), and didn’t have much of a campaign presence in general. It wouldn’t surprise me if 80% of the voters in the Democratic primary hadn’t heard of either candidate.
The candidates were indeed listed in alphabetical order for each office. There was no information on the electronic ballot other than the candidates’ names. Most of the attention on the Democratic side focused on the gubernatorial election, and in Columbia, on a local solicitor’s race.
Also, regarding Gabriel’s question, I heard on the local news that Rawl won in the absentee voting while Greene won in the election day voting. I don’t see the breakdown between absentee and election day voting in the data on the South Carolina election website, however. Maybe someone else can look into that question if they find the data.
Did Alvin Greene Win Because of Ballot Order? Because of Race?
Everyone seems to be glossing completely over the point. Study after study keep coming out pointing to the lack of knowledge of liberal (read democrat) voters. Just last week there was another one. (I don’t make the studies, I just read them.) They all say the same thing. Liberals are the least likely to actually know what the hell they’re talking about and the loudest to say it. And now, a man no one knows, who doesn’t know what TARP was, who spent less than $2000 campaigning (read didn’t actually campaign), wins the SC democrat party nomination for U.S. Senate. That he is a man who is also charged with a felony only sweetens the deal. I’m sorry everyone is irked, but they got exactly what they deserved. Way to go Alvin!
Aside from the fact that I don’t live in South Carolina, the fun part of this whole thing is watching reporters think of ways to describe Greene without using the phrase “grossly under qualified” and to describe the situation without using the phrase “We haven’t a clue.”
I don’t make the studies…
You also don’t link to them, mention the name of the person who did conduct the study, or in any other way provide information.
You said “which only implies that 8-9% of voters were influenced by ballot margin”. But isn’t the correct figure just under 18%? If 18% voted for Greene because his name was first and the remaining 82% split their votes evenly, the result would be 59-41.
MH, it’s very reminiscent of how they covered obama’s lack of credentials, qualifications and capabilities. Hell, Alvin could be the next president!
It was a WSJ article about a Zogby poll. I don’t subscribe so I can’t give you the link. Sorry.
The first thing to note is that these results were recorded on paperless (DRE) voting machines, meaning that they are incapable of being either recounted or audited. No one has ever invented a voting system as subject to undetectable fraud and/or error as the DRE system. That’s why most jurisdictions that bought them are now replacing them with paper ballots counted by optical scanners.
That said, it is plausible that this unexpected result was solely due to ballot order.(Placing the names in alphabetical order in every precinct in the state rather than alternating the order from precinct to precinct as good election management practice requires.)
But it is even more plausible that the results were manipulated (hacked) with the intent of bringing about this result. The motivation for this would have been to make DeMint’s reelection so certain, by nominating the weakest possible opponent, that no resources would have to be expended in his campaign, and thus more resources could be expended in the more competitive gubernatorial race.
Since the primary election results cannot be recounted–there being no record of the votes other than the numbers on the machine registers–the best evidence of fraud would come from reviewing the relationship between the Senate results and the gubernatorial results, precinct by precinct. If there are precincts where the Senate results are noticeably different from the statewide results–e.g., Greene got 25% more votes than his opponent while the gubernatorial results were consistent with the statewide results–that would be grounds for suspicion that they had been subject to hacking, especially if the dropoff (rolloff) results in those precincts systematically differed from those in the rest of the precincts.
The bottom line is: between the hypothesis that this was the result of ballot order vs. the hypothesis that it was the result of hacking, the latter is the more plausible, imo.
South Carolina’s e-ballot, paperless voting system is wide-open to undectable, unaccountable vote fraud. Why wouldn’t an insider (technician or election official) take advantage of the unaccountable, inauditable method of tallying votes to rig the election?
An alternative possibility is that someone innocently misprogrammed the ballot definitions to count votes for candidate A for candidate B and vice-versa.
Only the elections industry would use an inauditable method of determining who is elected to control budgets and contracts worth millions to trillions of dollars.
TP: You are right! My math was faulty. Thanks.
“It was a WSJ article about a Zogby poll.”
I believe you must be referring to the Klein and Buturovic study that purported to show that liberals are less “economically enlightened.”
It’d be hard to tell this from reading the WSJ, but this study was a complete piece of trash. As Nate Silver pointed out, they used Zogby internet polls, which are the worst polls available, and they asked many questions whose answers are by no means a matter of consensus among economists. Furthermore, if you read the study the authors themselves explicitly admit that they asked no questions that challenged conservative views (that is, where they would’ve considered the conservative answer to be wrong), and they more or less explicitly admit that their motivation was animus to progressivism (“We put much stock in Hayek’s theory… that the social-democratic ethos is an atavistic reassertion of the ethos and mentality of the primordial paleolithic band”). Essentially, their method was to devise a quiz in which they marked liberal answers as wrong, and then report that liberals were more likely to give liberal answers. It may also be worth noting that Klein is the editor-in-chief of the journal in which the original study appeared.
The real moral is that the WSJ op-ed page will lie to you.
I have a question – what else, if anything, was on the ballot? To assume that only 8-9 percent of voters swung the election because his name was first, or because it was African American, you have to also assume that everyone was selecting randomly. If they were selecting randomly in a primary election, why did they show up to vote in the first place? Was there anything else on the ballot that would have drawn them to the polls, and then stayed to vote for the Senate candidate with the most familiar name?
Princeton University touch screen voting machine study – Video:
http://itpolicy.princeton.edu/voting/videos.html
Comments on this entry are closed.
{ 1 trackback }