Saturday, February 12, 2011

Analysis and Rebuttal of the Center for Michigan Redistricting Study

The Center for Michigan, a liberal, pseudo-centrist think tank in Ann Arbor, recently published a 'study' attacking the redistricting process in Michigan.

How redistricting predetermines many election results and often leaves voters without competitive choices at the polls

The report is cowritten by Susan Demas and John Bebow. Demas is a liberal reporter and blogger.

The 'study' argues that
1. Michigan legislative elections are uncompetitive.
2. This is due to redistricting, particularly the 2001 plan passed by Republicans.
3. This lack of competitiveness 'disenfranchises' Michigan voters.
4. Michigan's legislative districts should be drawn to create more 'competitive' districts.


3. The third point is the easiest to deal with. The authors are misusing the word disenfranchise, which means to take away the ability to vote. Losing an election does not mean you were disenfranchised.

Note however that if we were to accept this false definition of disenfranchisement, that competitive districts would 'disenfranchise' more voters than noncompetitive ones, since in a competitive district, almost half the voters will see their candidate lose each time. Most voters get their way when districts are not competitive.

1. Are Michigan legislative elections competitive? It depends what you mean by competitive. It is certainly true that in many elections, the outcome is not really in doubt. However, the standard used by the 'study' is dubious. They switch between competitive, by which they seem to mean 'has switched parties in the last decade' and 'swing district', meaning 'they either regularly changed hands between the parties or averaged a 3-percent margin or less over the past decade'.

They point out that 25 of 110 state house seats switched parties in the past decade. That sounds pretty competitive to me, given the relatively small number of voters who switch parties even between wave elections. How many districts should have switched? 50? 100? The 'study' doesn't say. If the map is so gerrymandered, why did any seats switch?

Their standard is flawed because it does not account for the fact that which districts are competitive changes from election to election. A seat that may be perfectly competitive on paper may not be so in a given year when it has a popular incumbent running for reelection. Which seats are competitive also depends on swings in public sentiment.

For example, the 61st house district had highly competitive elections in 2006 and 2008, narrowly won by Republicans. In 2010, the public swung to the right and the Republican won by 25 points. In the 20th senate district, there have been three hotly contested elections with the last two seeing a million dollars or so spent. All three were won by the Republican (2006 very narrowly). Both districts are listed as not competitive by CFM. Someone should have told the candidates.

The report doesn't make clear whether their standard for swing districts averages margins or vote percentages. For example, if a district swings from 55%R-45%D to 45%R-55%D, is the average 10% or 0%?

2. Is the lack of competitiveness, such as it is, due to redistricting? If voters were randomly distributed geographically, we would expect to see most areas nearly even in political preferences. But voters are not randomly distributed, they choose where they live. People tend to live near people who are demographically similar to themselves. This is true for income, class, race, and to a lesser extent age. Political preferences are correlated with all these factors to some extent. Thus the distribution of voters is not random. Evenly divided areas are the exception, not the rule.

It is true that gerrymandering can exacerbate lack of competitiveness due to geography. For example, only one of California's 53 congressional districts switched parties in the past decade. This is due to an extreme gerrymandering plan. But this is nothing like what exists in Michigan. The number of swing districts is hardly out of line with geography in Michigan.

2001 State House Districts - MICHIGAN S 110 HOUSE DISTRICTS

Consider the state house. Does the map appear gerrymandered? (No.) The districts are compact and follow county and city lines when possible. If the map were so heavily gerrymandered, why did democrats win the state house in 2006 and 2008? Now, there is a slight Republican lean to the map, but it would take a political expert to demonstrate this.

There were five state house elections over the past decade, so even in the era of term limits, most districts had two or three representatives. Since incumbents rarely lose, there weren't that many chances for districts to flip.

2001 State Senate Districts -

The state senate districts also are compact and follow county and city lines as much as possible. The strangely shaped districts are in Detroit, which are designed to create black-majority districts, not for partisan advantage. The map leans more Republican than the state house map, but again the effect is subtle.

There were only three state senate elections over the past decade and each district has had two senators over that time. Republicans had good years when there were many open seats in 2002 and 2010, and were fortunate to have many incumbents in the bad year of 2006.

In support of their argument, the authors cite the fact that the percentage of seats won by Republicans exceeds the percentage of the aggregate vote that they receive. They blame this on redistricting. But there is a simple geographic explanation for this: Detroit. Most districts in Detroit are 95% democrat. This has the effect of concentrating many democrats in these districts and diluting them elsewhere. There is no comparable concentration of Republicans in the state. The most Republican district in the state (Holland area) is probably around 80% Republican, and few are more than 70% Republican.

There is a simple test for the authors' claim that redistricting causes lack of competitiveness. Seats that cannot be redrawn ought to be significantly more competitive. Consider statewide elections.

President. Michigan has gone democrat each of the last five presidential elections (it went Republican each of the five before that). Most of these elections were not close.

Governor. The governorship changed parties in 2002 and 2010. The 2002 election was fairly close (4%) while 2006 and 2010 were blowouts for opposite parties. (The 1994 and 1998 elections were also blowouts.)

Senate. Carl Levin has been elected to six consecutive terms (36 years), usually by large margins. Debbie Stabenow has been elected to two terms (12 years), succeeding a Republican.

Secretary of State. Republicans have won the last five elections by significant margins. Democrat Richard Austin held the seat for the preceding 24 years.

Attorney General. Republicans have won the last three elections (2002 was close). Before democrat Jennifer Granholm's one term, the seat was held by 'Eternal General' Frank Kelley for 38 years.

None of these seats, except possibly the governorship, would be considered competitive by the standard employed by the Center for Michigan. Apparently the state of Michigan is simultaneously safe for both democrats and Republicans depending on the seat in question!

Similar results hold on the county level. For example, Kalamazoo County is usually considered to be a swing area. But democrats have won the last five presidential races in the county. Democrats won governorship in the county in 2002 and 2006 and lost in 2010. Republicans won the last five secretary of state races and last three attorney general races in the county. The state senate district including Kalamazoo County has voted Republican for 50+ years. The prosecutor, clerk, treasurer, and surveyor have similarly been held by the GOP for many decades. A similar string in the Sheriff's office was broken in 2008. Only the drain commissioner could be considered competitive by the CFM's standard, switching parties each of the last three elections.

Obviously gerrymandering cannot account for this. Incumbency and the particular nature of these offices provide better explanations.

4. The report is somewhat vague with regard to recommendations. It suggests that there be more transparency. But as the report correctly notes, average voters don't care about redistricting, so this is unlikely to have much impact.

The report also speaks favorably of an 'independent commission' to draw districts. But such commissions cannot remove politics from an inherently political process. Such commissions have sometimes produced highly partisan gerrymandered maps, as in New Jersey. Others consistently deadlock and send the issue to the courts. Such commissions are accountable to absolutely nobody for their decisions.

The report also seems to suggest drawing a map to create more competitive seats. But as we have seen, geography guarantees that most areas are not politically evenly divided. To artificially create more 'competitive' seats would require extreme gerrymandering, ignoring county and city boundaries.

And why should 'competitive' districts be favored, anyways? Are substantive outcomes better for Michigan citizens. The report includes one quote that basically undermines its entire premise.

Brittany Galisdorfer, Earhart fellow with the nonpartisan Citizens Research Council, notes that academic research has been inconclusive about the effects of more competitive seats.

"The general theory is that competitive districts mean less extreme political candidates, less political partisanship and ensure basic fairness," she said. "But there’s no consensus that those things actually occur."

There is no question that redistricting is an inherently political process and that the sausage-making process is not always a pretty sight. But the Center for Michigan's critique of the redistricting process is fatally flawed. It uses a skewed standard for competitiveness for that virtually guarantees its conclusion. This standard leads to the absurd result that different seats in the same jurisdiction can simultaneously be considered 'safe' for opposite parties.

It incorrectly blames lack of competitiveness on redistricting. Instead, geography is the biggest explanation for lack of inherent competitiveness and the role of redistricting is exaggerated. The biggest reason for lack of competitiveness in specific elections is incumbency. Most seats that flip do so when they are open, and this most often happens due to term limits. Ironically, the Center for Michigan opposes current term limits.

It blatantly misuses the term 'disenfranchise' to sensationalize the issue. The 'reforms' that it favorable recommends would not help and could easily make things worse.

Whether the Center for Michigan's report is sincere or merely an attempt to impede Republicans in the redistricting process, its poor reasoning and misleading use of statistics do not deserve to influence public debate.

No comments: