Statistics

Wisconsin Election Statistics

Wisconsin Kathy Nickolaus press conferenceMuch is being made of the election between Republican State Supreme Court Justice Prosser and Democrat Kloppenburg. On election night a quick count of votes was made and released immediately to the media. This count showed Kloppenburg ahead by about 200 votes out of some million-and-a-half cast.

The election was of interest because of the contretemps involving left-leaning state employee unions (who say their members aren’t provided sufficient benefits and who want the right to collectively bargain on this point), a right-leaning Governor (who wants to reduce the state’s deficit by not increasing the state employee benefits as fast as has been done historically), and Democrats in the legislature who fearfully fled the state to forestall legislation.

To emphasize their point, some in the unions, or their supporters, threatened (mostly anonymously) various Republicans with death, destruction, desecration, etc. The media chose not to emphasize the violent nature of these threats, doubtless believing they were not representative, and perhaps because the media recalled their behavior in the Gabrielle Giffords Tucson shooting.

The media instead sought to turn the election into what they call a “national referendum,” which to them meant that this local election was indicative of the mood of the rest of the country. So when it first appeared that Kloppenburg won, the media began running stories claiming victory for Kloppenburg and their referendum.

In spite of media hopefulness, some pointed out that a contest with such a narrow margin was not much of a referendum and was instead an indication that the populace was evenly split.

Then came the press conference from Waukesha County Clerk Kathy Nickolaus who explained that an error had been made and that, after correcting it, Justice Prosser was the victor by some 7,000 votes (via HotAir).

Two things are significant. The method of data collection and the behavior of reporters.

I always tell my students and clients that eighty-percent of any data analysis project is spent in data preparation. Almost universally, people believe that their data is “ready to go”, has no errors, or is otherwise unproblematic. The opposite is nearly always true. Errors abound.

In Wisconsin, they gave Excel spreadsheets to precincts to fill out. The precincts were not supposed to meddle with the spreadsheets except to input data, after which they were to return the spreadsheets to headquarters where they would be read into an Access database. The process was done by hand at each level.

Naturally, some people could not help themselves and added extraneous columns and data into the spreadsheets. This always happens. If you have explicit instructions that the data in a cell can only, under penalty of torture, be coded “Y” or “N”, there will always be found someone who will write “Y but only because…” or “Not sure” or God knows what.

The creativity of humans in lousing up instructions is infinite. So it is no surprise, none whatsoever, that errors were made in Wisconsin, especially in the heat of battle and by clerks anxious to release results to the media.

Some in the media took the updated results like a man, but the dejection in many of their voices was evident when asking Nickolaus “How?” and “Are you absolutely sure…?” Earlier, I argued that all journalists should preface their questions by naming their party affiliation or by admitting who they wanted to win, e.g., “Bob Boberts, ABC, Democrat, Kloppenburg hopeful. Do you think…?”

This begins interrogations on a fair footing. Both the questioner and questionee know who is who, and more importantly, so do the audience. It also saves reporters from having to falsely claim objectivity and thus weaken their souls.

But notice what has happened. By insisting on the national-referendum theme and then prematurely touting it, the press are stuck with it. They are now in the position of saying (perhaps tacitly) that Prosser’s victory is important for the country. The pain many reporters are feeling must be extraordinary.

The vote objectively indicates an even split in the temperament of the populace, just as it did when the media thought Kloppenburg won. Yet that story has been eclipsed by the horse-race referendum, a theme many would have missed had the press not focused so much attention on it. I wonder if they cover this subject in J-school?

If you liked this, you’ll also like this: The California Federation Of Teachers Meet: A Play In One Act.

Categories: Statistics

17 replies »

  1. Dr. Briggs,

    I take exception to one statement in your article, “The vote objectively indicates an even split in the temperament of the populace.” I think the indication is the temperament of those voting. We did not get a sample of the temperament of the non-voting population and so not of the whole populace.

  2. Craig,

    True, true. But we have set up our government to function such that voters represent all of us.

  3. The 2009 population of Wisconsin was 5,654,774 and we had 3,000,000 voters. That’s a pretty big sample, eh?

    That 5,654,774 (from http://quickfacts.census.gov/qfd/states/55000.html) is an ESTIMATE ! Like getting an estimate on car repair down to the penny. I wonder what the error bounds are +/-1? +/-100? If the latter how can the estimate have much higher precision? I can’t believe they somehow didn’t let 100 people slip through the cracks.

  4. Absent a 100% turn-out, the election results are most likely to over-represent the opinions of those who hold strong opinions; as they would be most likely to take the trouble to vote. Would that be a form of self-selection bias?
    So, lets take a look at events in Wisconsin over the past few months. Who do you think has strongly held opinions? The folks that occupied the capitol building for weeks or the folks that stayed home and went to work every day?

  5. Dead people always vote Democrat so obviously not enough effort was spent on getting out the dead vote.

  6. All,

    The perpetually nervous are having conniptions over the corrections. Fraud! Conspiracy! Nefariousness! Just for an example, somebody on that site called “Mitch Beales” said in response to the correction. “The ‘better sort’ of Rs are the one’s in the cemeteries.” I take it “Rs” are Republicans.

    Also check out this story, which shows that errors are frequent.

  7. Apropos Ray’s remarks, reports circulated beforehand of an extraordinary “register-and-get-out-the-vote” effort aimed at WI domiciled collegians, regardless of their registration and voting status elsewhere. Thankfully it seemed to equal the effort with cemetarians.

    I believe even voters such as Belvidere’s Craig way over in Buffalo County would be concerned of the potential for mischief if voter eligibility laws become so fluid they enabled fraud. An honest vote needs to count, and a dishonest vote needs to be nullified. IMO this is the true “national interest” factor from this election.

  8. My take on what happened:
    -Dems withheld reporting from a heavily Dem precinct to see if any adjustments would be needed.
    -Repubs, planning for Dem chicanery, withheld media reporting of a large Repub precinct.
    -Dems, thinking all the votes were in, knew how many they needed to put Kloppenburg over the line and “found” enough “overlooked” votes and declared victory.
    -Repubs, now believing the Dems had used all their ammunition, reported the delayed correct votes from their precinct.
    -Repubs win without doing anything unlawful.

  9. I’m generally (and in this case) of the “never-attribute-to-conspiracy-what-can-be-chalked-up-to-incompetence” school.

    I should note that UW law professor and blogger Ann Althouse noted the discrepancy from Waukesha County on election night as she was following the returns. The county updated its status from about 60% of precincts reporting to 100% without adding any votes to either candidate (but it is a heavily Republican county).

    Althouse wrote at the time, “Waukesha is now shown as completely in, but the numbers didn’t change, so I think something may have been misreported. I took the trouble to do a calculation and was going to predict that Prosser would net 40,000 more votes in Waukesha. What happened?”

    http://althouse.blogspot.com/2011/04/watching-election-results.html

  10. By the way, the vote totals show just under 1.5 million votes cast total in this election.

  11. Curt,

    Thanks. For no good reason, I multiplied by 2 (since the vote was 50/50; see how easy it is!).

  12. Excel spreadsheets? By hand? This doesn’t sound like current technology, even for Wisconsin.

    Briggs notes how easy it is to fluff the numbers in a spreadsheet. It happens all the time in processes like sales forecasting to accounting. Finding more mistakes of this variety is inevitable.

  13. Spreadsheets used for serious purposes have ALL cells not intended for input protected. Several times, I’ve been asked to “debug” a spreadsheet only to find that a formula had been overwritten by a value.

    If an input is constrained, one cell is used for input but the actual value used in calculation is filtered into another cell for use in the calculation. If the input is out of bounds, nothing appears in the filter result.

    Regards,
    Bill Drissel
    Grand Prairie, TX

  14. You say “some” threatened, but only one person has been charged. Meanwhile, a Recall Darling petitioner was punched in the stomach with little attention by the media. There are going to be bad actors on both sides, but they shouldn’t be used to represent the movement as a whole. That’s why we can’t get anything done in this country because we keep focusing on the extreme minority of each side instead of focusing on the majority in the middle who are closer on most issues than we want to admit. However, anyone who engages in broad brush painting, such as you, is likely on the extreme end so I’m probably wasting finger power by trying to reason.

    To your greater point, I don’t think there is fraud, but there is clearly gross incompetence due to Ms. Nickolaus’ history of screw ups. She messed up election results in 2006, she refuses to allow an audit of her system, she handles elections differently than every other county clerk in Wisconsin – this is a pattern. She should step down. I don’t know why we elect people to the county clerk position anyway. Holding elected people accountable for doing a piss poor job is so much harder than holding a person accountable who has to apply for the job.

  15. Dan, with respect, I hope you’ll look at what you wrote and realize its as bad as Dems screaming “fraud.” You’re saying both sides were conspiring, but I can’t see anything in your logic as to how anyone withholding election results would be beneficial. Technically, you’re suggesting all county clerks are partisan hacks.

  16. Nickolaus has a long history of this type of questionable, highly suspect, likely fraudulent activity.
    As GWB once said, ” fool me once, shame on — shame on you. Fool me — you can’t get fooled again.”

  17. Jesse, nice try but no cigar. The whole country saw how the union teachers, leftist academics and their students, and union police behaved in Madison. We saw the teacher whose protest sign identified the governor’s son as her student, with school name openly declared. We heard how a note slipped under a legislator’s door proclaimed “the only good Republican is a dead Republican.” We saw Walker depicted as Hitler (and groaned…..LAME). And we saw the vicious protest signs, posters, and the hateful faces of the left as they screamed invective toward those the people of Wisconsin had elected to do the people’s business. “Just one person has been charged” but thousands behaved like barbarians. Thanks to Althouse and Meade and the alternate media, those images are captured for history, and the tide is turning against the liberals/progressives as ordinary citizens see how deeply people are corrupted by being dependent on government.

Leave a Reply

Your email address will not be published. Required fields are marked *