Thursday, April 19, 2012

Mr. Strom responds to Dr. Klumpp

Today I present a guest post from Michael Strom,a frequent commenter on this blog. Strom writes in response to a post I put up on April 13 ("Dr. Klumpp reviews countywide judicial results"), prepared from emails exchanged with Albert J. Klumpp, PhD, a research analyst with McDermott Will & Emery LLP -- also a frequent and generous contributor to this blog. In an email, Strom explained that he started this post as a comment to last Friday's post, but it became too long.

Strom practices with the firm of Strom & Associates in Chicago. He currently serves as first vice president of the Decalogue Society of Lawyers.


During my undergraduate days at the University of Illinois, like many of my compatriots, I wound up majoring in Political Science awaiting admission to law school. I quickly learned to my horror that many of the professors were serious about the word "science" in their department. Many of them spun sparse material into endless graphs and charts that allegedly proved things. Those efforts uniformly fell far short of spinning straw into gold. I did not stay around long enough to find out exactly what it was they thought those charts were proving, since I dropped any course where a political science instructor demonstrated a belief that science (in chart or graph form) was actually part of the analysis. My disciplinary training on analysis of numbers and statistics is limited to a misspent youth poring over baseball statistics.

Your opening comments concerning generals preparing to fight the last war and poking through the March 2012 judicial primary ashes seeking patterns helpful in 2014 are spot on. Each primary election season around here is so different from the one before, we must despair for anyone aggregating data over a 35 year period on the purpose of drawing conclusions significant to future contests.

When Avy Meyers started asking me to analyze judicial elections in 2008, since my personal experience in the field consisted of finishing fifth of six candidates in a 2004 primary, I was reduced to mining data for a statistical analysis. I started with all the usual "decisive" criteria: first ballot line, party slating, Tribune endorsement, etc. -- but when the analysis was carried to the next step, it became exceptionally clear that in 2008 gender trumped everything else on the table. For example, considering the fact that the candidate with the top line on a judicial ballot won 12 of 25 contested races, it would be easy enough to cite that as advantageous. However, it gets difficult to make that argument in a year like 2008 where male candidates holding the top line of the ballot only won two of 11 contested races where both male and female candidates were in the race. Since female top line judicial candidates won 10 of 14, it is difficult to take the ballot order as seriously as we often hear. In other words, in a year where one of the cited factors is clearly subordinate to other more decisive factors, aggregating data becomes more misleading than enlightening. Since female judicial candidates had a surprisingly bad year in 2010, that business about looking to the past to prepare for the next war really hits home.

So, I will turn to a few comments relating to Dr. Klumpp's analysis. I am certain that his methodology is exceptionally sound and rigorous in all aspects susceptible to scientific deduction. However, I do question some of the conclusions drawn from his review of what is always a limited pool of data in any given year.

“DILUTION”: One of the common insider strategies deployed where there is fear of a single candidate with a characteristic deemed favorable –- such as gender or an Irish name -– is to put up faux opponents to dilute the advantage. Dr. Klumpp agrees: "If more than one candidate in a race shares one of these advantages -- such as more than one woman or Irish-surnamed candidate in a contest, the statistical advantage is diluted." The only problem with that theory is that I have seen it wrong in two consecutive election cycles. Eight of the 18 female primary winners in 2008 were in races with multiple females. In 2010, there were nine judicial primaries with one female candidate against one or more male candidates (seven races had one female against multiple male candidates). The female candidates won only one of those nine races that year. Just to make things more confusing, in the 10 races with multiple female and multiple male candidates, five winners were male, five female. Go figure. If the point of such analyses is to draw statistical conclusions that will be more frequently accurate than not, we might do better with a Magic 8-ball.

FAITHFUL COMMITTEEMEN: This was the part that got my head spinning.
Klumpp acknowledged "there are instances in which a ward or township organization will deviate from a party slate, but I haven't seen evidence that it's widespread enough to justify adding into a big-picture model like mine." * * * [Klumpp] preferred to state "the slating measure is strictly a countywide value."
Down in the trenches, the candidates know the difference between "slated candidates” and "really slated candidates.” But let's take a limited peek at the variable advantages of slating.

In 2010, there were three contested Illinois Appellate Court races in Cook County. Here is a list of slated candidates, percentage of vote (countywide) and order finished:

EPSTEIN26.88%[1st]
HILL-VEAL19.95%[3d]
PATTI15.91%[3d]

Through some statistically convenient fluke, all three Appellate Court races had five candidates, neutralizing any data-skewing effect of comparing three candidate races against five candidate races. If "slating is slating," how come Judge Epstein received almost twice as many votes as Judge Patti? Say, wouldn't it be great if we could gauge the effect of slating when the same person runs in two consecutive elections, once with slating and once without? No problem, since Judge Hill-Veal was slated in 2010, but ran without slating in 2012 –- both in races for the Illinois Appellate Court. Here are Judge Hill-Veal’s results, targeting three wards where the Democratic Party tends to be more successful, plus the 50th Ward as sort of a control factor -- Avy Meyers would be curious about his home ward anyway. The vote totals are lower across the board since the overall city vote in 2012 was about 24% less than 2010, so the percentages are more significant:

Pamela E. Hill-Veal/2010Pamela E. Hill-Veal/2012
Ward
VotesPercentVotesPercent
113966.45%2896.47%
135887.89%5777.36%
191,72712.19%1,69516.91%
5074815.89%28010.58%

2010 City Total 291,015
2012 City Total 221,038 [-69,977 from '10 = 24%]


Looking at these numbers, it seems pretty apparent that slating did nothing for Judge Hill-Veal. Oddly enough, she did markedly better in the 19th Ward (arguably the best Democratic Party ward in the City) without slating (16.9%, rather than 12.2%). In the final analysis, she finished third in the City both times, improving her City total of 23.29% in 2010 to 25.85% in a six-person race in 2012.

Lest we stir up troubling concerns on party slating for African-American candidates, in 2010 the first countywide race after the appellate contests featured a slated African-American candidate, Judge William Hooks. Judge Hooks carried about 52% citywide. For comparison (albeit in a three-person race), here are Judge Hooks’ totals in the same wards reviewed above:

WardTotal Votes CastVotes for William H. HooksPercentage of Vote
115,7373,50661.11%
137,1244,36261.23%
1912,5777,05256.07%
504,3502,26252.00%


EDUCATED VOTERS: Dr. Klumpp perpetuates the "good government types'" wistful hopes of educating the voters to do a better job on the judicial ballot: "[T]here are civic-minded, responsible voters out there who diligently study judicial elections, there just aren't very many of them. At least not enough to remove poorly rated retention candidates, for instance." He added, "Unless we start doing a better job to educate young people and new voters about the judicial part of the ballot, it's not going to change." Oh the irony of it – the chances of getting the good people of Cook County to scour ratings, endorsements and qualifications for perhaps 90 judicial candidates in 25 or so races per year are so remote that he has entirely ignored the salutary effect of an exceptionally low vote total. Those civic-minded, responsible voters lauded by Dr. Klumpp always vote. In an election like 2012, the people who did not vote are those much more likely to choose on the basis of lyrical Irish names, gender preference, and all the other stuff he hates. So really, the ideal scenario for judicial elections is to alienate as many casual voters as possible so the civic-minded folks can decide the election. If only there was a way to square that with democracy.

FIRST BALLOT POSITION: Dr. Klumpp fears that top ballot position is such a big deal that: "Ballot position was determinative in two of last month's contests. Jesse Reyes would have lost if William Boyd or Ellen Flannigan had been first on the ballot, and Stanley Hill would have defeated Karen O'Malley if Hill had been listed first instead of O'Malley." We have been encouraged to consider a rotating ballot position to neutralize this decisive edge. To which I posit the following query: Huh?

There were three contested Illinois Appellate Court races on the 2012 ballot. In the first one listed, Judge Hill-Veal had top ballot position and lost. In the second one listed, Judge Patrick Sherlock had top ballot position and lost. If we are aggregating Irish names and first ballot position, Judge Sherlock should have won. If we are aggregating gender and first ballot position, Judge Hill-Veal should have won. I am puzzled on Dr. Klumpp's theory of how it is that first ballot position became so compelling for Judge Reyes. Is it a theory of ballot fatigue, where thinking through the Supreme Court race and the first two Appellate Court races led to mental fatigue alleviated only by poking the first name seen in the third race? Is there an emerging theory under which first ballot position is especially advantageous to an Hispanic candidate, notwithstanding the fact that the second ballot position was also held by an Hispanic candidate? Was the certainty of Judge Flannigan's victory (if provided first ballot position) determined by some sort of transitive mathematical property under which we can extrapolate the victory of Karen O'Malley into a similar victory by other female Irish names candidates holding first ballot position? Is this theory contradicted by the victory of Jessica O'Brien from the second ballot position, or would civic-minded, responsible voters have known she was ethnically Asian, thereby neutralizing the aggregated advantages of first ballot position plus her politically golden married name?

No comments: