Saturday, April 9, 2011

Mark your calendars

Photo credit Brian Kelly

Folks in Texas have a saying about the summer. Summer lasts from Memorial Day to October, and it is best enjoyed in Colorado. Now, since this year will be my first summer in Texas, Susan and I have decided to avoid it for a bit longer by heading to Michigan for Memorial Day week. Colorado will have to wait another year.

Every time we visit west Michigan is wonderful, but we often are so caught up meeting with family that we miss our friends. Or, getting together with friends is a last minute event that is very hard to fit in everyone's schedule. So, this time I'm going to do things a bit differently. If you are interested in hanging out for dinner, or drinks, or just catching up, level a comment on this post, or give me an email at peterbratt at It would be great to see you all again, and I hope that you feel the same way.

Tuesday, April 5, 2011

2010 Recap Part III: A Polarized Electorate

While Michigan was among the last states to have its 2010 Census data released, the wait was well worth it. The story that the media covered was Detroit’s 26% population loss since the 2000 Census. As I mentioned in an earlier post, in 1950 Detroit was home to 29% of the state’s population, a figure that has since dropped to 7% by the 2010 Census. The story of Michigan since then end of the era of “Grand Expectations” between 1945 and 1974 is one of painful economic restructuring from a industrial to a service-based economy, population dispersal (whether from core cities to suburbs or to other states), and growing income/economic inequity within the state.

Michigan’s story since 1974 also must be understood as one with continued voter polarization by race and location. As always, this story is best told with a mixture of data and narrative, and this post try to provide both. Tables One and Two, shows the Michigan’s population by current state house and state senate districts in both 2000 and 2010. As you can see, in 2000 Michigan had 9,938,444 residents, with 80% of the state’s population White, 14% African American, 3% Hispanic, 2% Asian, and 2% Other. By 2010 the state’s population declined to 9,883,640, and the White share of the population dropped to 77%, the African American share remained around 14%, the Hispanic percentage increased to 4%, while the Asian share of the population remained at 2%. To put the story in numbers, the state lost 54,804 residents over the past decade. The White population dropped by almost 400,000 (396,114 residents to be precise), 28,986 African Americans were lost, while the Hispanic and Asian population increased by 112,481 and 59,980 respectively.

Table Three shows that this population decline was not evenly dispersed between amongst the state’s legislative districts. Among both the State House and Senate seats, districts experiencing the greatest population loss were concentrated within the city of Detroit, as well as urban core districts throughout the state. In contrast, districts representing outlying portions of metropolitan Detroit and western Michigan experienced growth. At the same time, the racial composition of these districts changed over the decade. For example, the Hispanic population grew by more than 7,000 in the 77th State House seat (Wyoming and Byron Township), so that now it is has the third largest Hispanic population percentage in the state (training districts 76 (Grand Rapids) and 12 (southeast Detroit). At the same time, the dispersal of African Americans from the City of Detroit shows in its growing population share in various suburban districts. District 1 (Grosse Points, Harper Woods, Far East Detroit) had its African American share of the population increase from 21% to 35% in a decade, which somewhat explains the seat’s trending to the Democratic column. Likewise, in District 12 (southeast Detroit), the White share of the population dropped from 44% in 2000 to 20% in 2010, while the Hispanic share jumped from 42% to 55%.

Table Four is perhaps the table of most immediate interest for mapmakers and strategists within the state. This table shows the Voting Age Population (VAP), which consists of residents over the age of 18, for the current state house and senate districts. A careful review of the data shows that the VAP tends to be slightly more White than the total population, largely because the Hispanic population is younger than White and African Americans.

Back in November I did a bit of regression analysis to see what impact race had the 2010 elections. I looked at 55 state house seats, and found that the Democratic baseline average from 2004-2008 largely predicted the 2010 result. Since I have the average Democratic baseline for the past four election cycles by Senate and House district, I decided to run another regression test to if a district’s Democratic baseline could be a predictor for the same district’s racial breakdown, and expanded it to the state senate seats as well.

The result is shown in Tables Five and Tables Six(all in the same Google Spreadsheet linked above) shows that racial voting polarization is alive and well in Michigan. In state house seats, the White VAP% and the Democratic baseline explained 77% of the variation, and is significant well beyond the .05 level, while the African American VAP% had a r-squared score of .751, and was likewise significant beyond the .05 level. Both tests also had significant T-scores. In looking at the data, with just one exception, every state house or senate district with an African American population share of greater than 10% received a Democratic baseline percentage above 50%. Similarly, one a district’s White VAP dropped below 74% of the total population, the Democratic baseline never fell below 53%. In contrast, Table Seven shows that the Hispanic VAP% does not have significance on the Democratic baseline percentage.

The state senate results were similar to the state house data. Table Eight shows the White VAP% and the Democratic baseline with a r-squared score of .79 and is significant well beyond the .05 level. The African VAP% in Table Nine has a r-square score of .81 and was likewise significant beyond the .05 level. The Hispanic VAP% in Table Ten did not have significance in the state senate. Like the state house, any district with a White VAP% below 72% or an African American VAP% above 19% gave a Democratic baseline score above 51%.

What does this all mean as 2012 nears? While elections are about candidates, political parties, and voter mobilization, the way the 38 State Senate and 110 State House seats are drawn to protect the current GOP majorities in each chamber will need to also take into consideration the continual dispersal of African Americans into suburban communities. Likewise, the growth of Hispanic communities in various districts (Senate Senate District 29 and State House Districts 77 and 90) might cause future problems for the GOP if the Democratic Party can ever get around to mobilizing these voters. However, perhaps the most interesting story is whether another GOP drawn map can withstand long-term demographic trends in Michigan. The 2001 map, which was perhaps the best drawn GOP plan outside of Texas, collapsed in 2006 on the state level, and fell on the national level in 2008. While the GOP gained back all these losses and some in 2010, the sheer number of GOP legislatures might force the map makers in each chamber to draw aggressive maps to protect their enlarged ranks. That could spell trouble in a bad election year.