Wednesday, May 28, 2008

Primaries by The Percentages

Primaries by The Percentages

If you're watching the primary returns next Tuesday night, you may want to keep a calculator handy. Iowa law requires a candidate to win 35 percent of the vote to earn a nomination, but the percentages that get reported aren't always the ones that matter.

If no candidate gets 35 percent, the party has to hold a convention to choose a nominee, and that's a realistic possibility in one Iowa congressional race.

The prospect of a convention was a hot topic in Democratic circles in early 2006, and Ed Fallon's gubernatorial campaign made a big effort to elect delegates at precinct caucuses and county conventions. But the convention talk cooled after Patty Judge left the race to form a ticket with eventual nominee Chet Culver, who won a four-way race (don't forget Sal Mohamed) with 39.1 percent.

Even in a three-way race, a convention is a possibility if the math breaks just right. The Republicans came within one percent of a statewide convention for governor in 2002, when the three candidates all bunched up in the 30s; nominee Doug Gross emerged as the nominee with only 35.6 percent.

That same year, Republicans actually went to a convention for the 5th Congressional District. Four strong candidates competed in the brand-new, solidly Republican district, and they landed between 21 and 31 percent each. Steve King landed at the top of that heap, and then turned that lead into a convention win. The rest is history.

Who attends a convention depends on what level of office is involved. For legislative seats, it's county central committee members, with votes weighted by the size and party strength of each precinct. County offices are nominated by a county convention, made up of the precinct delegates elected at the Jan. 3 caucuses. Congressional district and state conventions are made up of delegates elected at the March county conventions.

The likeliest prospect for a convention is the Democratic race in the 4th Congressional District, with four candidates seeking the nomination. Republicans have two major three-candidate races: the U.S. Senate race and the 2nd Congressional District.

Prospects also exist for conventions in local contests. The definition of percentage is less than straightforward in elections for more than one seat, such as at-large county supervisor contests. To determine the 35 percent threshold in these races, the Code of Iowa specifies a "percentage of votes" method. To get at this number add up all the votes cast for all the candidates, including write-ins. Then divide it by the number of seats available.

For example, look at the 2004 Democratic supervisor race in Johnson County. Eight candidates were running for three seats, and a total of 20,905 votes were cast. Divide that by three, and the base lien for calculating percentage is 6,969.

Johnson County Supervisors, 2004 Democratic Primaryvotespercentage of votespercentage of votersTV Percentage
Terrence Neuzil5,06072.6%54.4%24.2%
Pat Harney4,62766.4%49.7%22.1%
Rod Sullivan3,18745.7%34.3%15.2%
Mike O'Donnell3,06043.9%32.9%14.6%
Rick Dvorak1,99828.7%21.5%9.6%
John Schneider1,33919.2%14.4%6.4%
Julie Gilmere1,13416.3%12.2%5.4%
James Knapp4466.4%4.8%2.1%
write in540.8%0.6%0.3%

This formula does not factor in the common dynamic of under-voting or "bullet voting," in which supporters of one candidate don't use all their votes in a vote for two or vote for three race. Instead, bullet voters vote only for their favorite candidate. You can't use both, or all three, of your votes on that candidate, but you don't have to use all of your votes. It's a common campaign tactic, but one that's usually done sotto voce so as not to alienate supporters of other candidates. If you want to crunch some math, add up the total votes cast for all candidates plus write-ins, then divide it by the number of voters. This will give you a "votes per candidate" number, illustrating how many people cast some sort of bullet vote. In our 2004 sample race, the 20,905 votes were cast by 9,305 voters, meaning the average voter only cast 2.25 votes for supervisor out of a possible three.

Calculating percentage of voters, by dividing the candidate's vote by the total number of voters, produces a lower percentage but provides a more reliable assessment of how many people supported a candidate. Another common, misleading method in a vote for two or three race is to add up all the votes, and divide each candidate's vote by the grand total. This is often seen on local TV, and produces seeming paradoxes like candidates winning with 17 percent.

Another seeming paradox happens in a race where four candidates are competing for three seats. It's possible for a candidate to win support from a majority of voters, yet finish in last place, as in this 2001 Coralville contest. Over half of the voters supported Bream, yet he drew the least support and lost.

Coralville City Council, 2001votespercentage of votespercentage of votersTV Percentage
Henry Herwig74480.0%70.4%26.7%
Dave Jacoby73278.7%69.3%26.2%
Tom Gill68273.3%64.5%24.5%
Bob Bream61466.0%58.1%22.0%
write in171.8%1.6%0.6%

No comments: