This diary is the second installment of my presentation, analysis and discussion of the 2010 election for United States Representative in MN-08, resulting in the defeat of long-time DFL incumbent Jim Oberstar to Tea Partier Lite challenger Raymond "Chip" Cravaack. While polling data in the weeks prior to the election indicated that the race was going to be close, certainly closer than any challenge to Oberstar since his election to the House in 1974, Cravaack’s victory (48.18% to 46.59%) was nevertheless a surprise, and a disappointment.
In Part I, posted on 12 November, I discussed several aspects of the district-wide and county-level voting data. In this diary I will discuss the precinct-level data.
First, however, I feel it necessary to reply to the constructive criticism offered in two comments in the discussion from the first diary in this series, MN-08: Anatomy of Oberstar’s Defeat, Part I, and atone for two lacunae in that presentation.
LACUNA I: COUNTY-SWING, 2008 TO 2010
In a comment to Part I, Kossack decisivemoment noted correctly that I had omitted to document the county-level swing in support for Oberstar from 2008 to 2010, and suggested that these data would temper my observations regarding the election, particularly in my home territory of St. Louis County:
The swing against Oberstar in St. Louis County was bigger than it was in most of the "red" counties. That's what makes this such a strong political earthquake. I think if you break it down further you'll find that the biggest swing against Oberstar occurred consistently outside the Republican friendly territory of central Minnesota, but rather up on the Iron Range.
In tabular form, here are the values for the percentage of the vote won by Oberstar in 2008 and 2010 by county, and the swing of support from 2008 to 2010, with the values bolded for the bloc-like "red" counties of the south and west of MN-08:
County | 2008 | 2010 | Swing |
---|
Aitkin | 63.79% | 43.64% | -20.15% |
Beltrami | 59.58% | 42.36% | -17.22% |
Carlton | 75.22% | 53.79% | -21.43% |
Cass | 59.38% | 38.65% | -20.73% |
Chisago | 59.39% | 38.26% | -21.13% |
Cook | 72.18% | 55.20% | -16.98% |
Crow Wing | 60.46% | 38.53% | -21.93% |
Hubbard | 57.29% | 38.72% | -18.57% |
Isanti | 58.57% | 36.80% | -21.77% |
Itasca | 67.94% | 48.24% | -19.70% |
Kanabec | 60.27% | 39.26% | -21.07% |
Koochiching | 72.52% | 52.54% | -19.98% |
Lake | 75.25% | 54.92% | -20.33% |
Mille Lacs | 61.25% | 40.05% | -21.20% |
Morrison | 61.61% | 35.79% | -25.82% |
Pine | 66.59% | 43.37% | -23.22% |
St. Louis | 77.15% | 56.92% | -20.23% |
Wadena | 55.17% | 37.97% | -17.20% |
The swing in support for Oberstar in St. Louis County, a heartbreaking –20.23%, is therefore not greater than in most of the "red" counties. Neither does the swing in St. Louis County vary significantly from the mean decline of 20.48% for all counties within the district, nor from the mean decline of 19.78% for the six "blue" counties. In short, while I absolutely believe that the pattern of voting in St. Louis County was of fundamental importance to the outcome of the 2010 election, I do not find county-level swing from 2008 to 2010, in and of itself, to be particularly revelatory. As to decisivemoment’s second observation, that the swing on the Iron Range was significantly greater than in the "red" areas of the district, there is much of merit here, and a great amount to be said in the following discussion of the precinct-level data.
LACUNA II: NORMALIZING VOTER INTENSITY
In another comment to Part I, Kossack mbayrob noted, also correctly, that the presentation of average county-level values is less useful, may perhaps even be misleading, than the presentation of values normalized by each county’s percentage of the total district-wide vote. I concur, and I am presenting here a table of county-level measures for voter intensity (vi is the total number of votes divided by total number of registered voters; a proxy for voter turnout), normalized by each county’s contribution to the total number of votes within the district and represented as the number of voters per thousand (vpk) within MN-08. The last two columns detail the difference in vpk from 2008 to 2010 and the difference as a percentage.
County | 2008.vi | 2010.vi | diff.vi | 2008.vpk | 2010.vpk | diff.vpk | diff.(%) |
---|
Aitkin | 80.70% | 69.71% | -10.99% | 25.61 | 26.59 | 0.98 | 103.83% |
Beltrami | 79.64% | 70.47% | -9.17% | 17.60 | 18.81 | 1.21 | 106.88% |
Carlton | 79.02% | 65.95% | -13.07% | 50.18 | 49.23 | -0.95 | 98.11% |
Cass | 78.45% | 69.27% | -9.18% | 44.38 | 46.84 | 2.46 | 105.54% |
Chisago | 79.10% | 68.56% | -10.54% | 79.65 | 82.13 | 2.48 | 103.11% |
Cook | 85.83% | 78.88% | -6.95% | 9.10 | 10.21 | 1.11 | 112.20% |
Crow Wing | 79.98% | 68.03% | -11.95% | 95.60 | 98.81 | 3.21 | 103.36% |
Hubbard | 79.26% | 72.18% | -7.08% | 31.70 | 35.09 | 3.39 | 110.69% |
Isanti | 79.61% | 64.52% | -15.09% | 54.50 | 53.13 | -1.37 | 97.49% |
Itasca | 81.19% | 70.69% | -10.50% | 66.38 | 70.86 | 4.48 | 106.75% |
Kanabec | 79.81% | 67.85% | -11.96% | 23.12 | 23.55 | 0.43 | 101.86% |
Koochiching | 79.45% | 65.91% | -13.54% | 18.51 | 18.29 | -0.22 | 98.81% |
Lake | 80.77% | 71.26% | -9.51% | 18.93 | 20.05 | 1.12 | 105.92% |
Mille Lacs | 78.64% | 65.91% | -12.73% | 36.86 | 36.76 | -0.10 | 99.73% |
Morrison | 78.66% | 67.33% | -11.33% | 45.63 | 46.11 | 0.48 | 101.05% |
Pine | 73.99% | 65.88% | -8.11% | 39.23 | 39.28 | 0.05 | 100.13% |
St. Louis | 77.21% | 62.56% | -14.65% | 323.43 | 304.71 | -18.72 | 94.21% |
Wadena | 80.53% | 66.61% | -13.92% | 19.59 | 19.55 | -0.04 | 99.80% |
As one would expect, even in perennial turnout-leading Minnesota, the measure of voter intensity declined in every county between the presidential election of 2008 and the midterm election of 2010. In Part I of this diary-series, I noted the to-my-mind vexing fact that the measure for voter intensity for 2010 in St. Louis County was the lowest among all eighteen counties in the district (62.56%), that the difference in voter intensity in St. Louis County from 2008 to 2010 was the second lowest (-14.65%), and I placed a great degree of emphasis on the need to explain the relatively low voter intensity in the county, particularly in my hometown of Duluth in southeastern St. Louis County. Those values for voter intensity were, as mbayrob noted, not normalized in any way to document the weight of county-level measures in relation to the entire district, a problem corrected by the vpk statistic. My initial observation on low voter intensity and turnout in St. Louis County nevertheless still holds, as vpk decreased by 18.72, the largest decline by far for any county within MN-08, representing that the strength and influence of the county’s voters diminished to 94.21% of their 2008 value.
What I had not anticipated, given my initial theory that voter intensity in 2008 diminished throughout the bloc of six "blue" counties in northeast MN-08 (Carlton, Cook, Itasca, Koochiching, Lake and St. Louis), was that the vpk values actually increased in three (Cook, Itasca and Lake). However, the net difference in vpk for this six-county bloc remains at -13.18. Put another way, the net gain for the bloc of twelve "red" counties was 13.18 votes per thousand, or 1.318%. In an election decided by a slim 1.59% of the vote, these regional shifts in voter intensity and turnout are indeed significant.
The constructive criticism, much appreciated, from both decisivemoment and mbayrob point to a general problem of spatial analysis: namely, the Modifiable Areal Unit Problem (MAUP). In short, precinct-level data are aggregated into larger areal units, into counties which are essentially arbitrary with respect to the phenomenon under observation across the entire district. Reporting data at a higher level of aggregation (the county, or the district) can produce misleading statistics, and certainly works to smooth over the interesting, and significant, variability within data for the lowest-level of analysis available, the precinct. So, without further delay, let us now consider the precinct-level data for the 2010 vote for United States Representative in MN-08.
PRECINCT-LEVEL METHODOLOGY
In 2010, there were 804 voting precincts in MN-08. I have chosen to present overall maps of the precinct-level data for the district, and then delve into discussion and analysis of a selection of regions in which the precinct-level data seem, to my mind, most significant for posing explanations for the outcome of the 2010 vote. This presentation addresses the following seven questions:
- was the precinct won by Oberstar (DFL) or Cravaack (GOP) in 2010?
- did Oberstar perform significantly better or worse than his percentage of the total vote in 2010? "Significant," in this context, refers to whether the percentage of the vote won by Oberstar in a given precinct falls outside, above or below, a range representing one standard deviation around his district-wide percentage of 46.59%.
- was the precinct won by Oberstar (DFL) or Cummings (GOP) in 2008?
- did Oberstar perform significantly better or worse than his percentage of the total vote in 2008?
- which precincts went from "blue" to "red" in 2010, and what was the swing in support for Oberstar at the precinct-level from 2008 to 2010?
- what was the change in voter intensity in the precinct from 2008 to 2010? Voter intensity is measured here by vpk, the number of votes in the precinct normalized as that precinct’s contribution to every thousand votes across the entire district.
- is there a distinction among the voting patterns of urban versus rural precincts? Now, to be fair, "urban" hardly has the same meaning in northeastern Minnesota as it does in other parts of the country. Nor are the United States Census classifications of "city," "township" and "unorganized" particularly useful for "city" precincts which, while they may include a small population center, also include a large tract of more sparsely settled land. Therefore, I have reclassified the voting precincts on the basis of population density into four categories for analysis: "dense," "moderate," "low" and "sparse."
PRECINCTS: WINS AND LOSSES, 2010
The following image shows the precinct-level geography of the 2010 vote with precincts won by Oberstar represented in blue (347 precincts), those won by Cravaack represented in red (433 precincts) and those resulting in a tie represented in grey (24 precincts).
These data confirm the previously discussed formation of a GOP bloc in the south of the district, adjacent to Michele Bachmann’s MN-06, and the west, adjacent to Blue Dog Collin Peterson’s MN-07. These data also justify, to some degree, the frequently expressed opinion that support for Oberstar diminished as a function of distance from Duluth, the most populous and ostensibly most liberal population center within MN-08. Perhaps most significantly, these data allow us to examine the "blueness" or "redness" of county-level data in more detail. Here I reproduce from Part I the county-level results of the 2010 election with the margin-of-victory indicated below the name of the county.
Within five of the "red" counties of the south (Morrison, Mille Lacs, Kanabec, Isanti, Chisago) the precinct-level support for Cravaack was nearly universal. It is perhaps worth reiterating that Cravaack is himself from Lindstrom, in Chisago County. Oberstar won only four of forty-eight precincts in Morrison, five of twenty-five precincts in Mille Lacs, one of twenty in Kanabec, none of eighteen in Isanti and two of twenty-four in Chisago. Of the twelve precincts won by Oberstar in these five counties, the majority were small and relatively densely populated precincts such as Onamia and Isle.
Oberstar performed slightly better within the more northern counties (Crow Wing, Aitkin, Pine) of this southern GOP bloc, winning eleven of sixty-two precincts in Crow Wing, nineteen of fifty-four in Aitkin and seventeen of forty-seven in Pine. Support for Oberstar among these forty-seven precincts represents a mix of both more densely populated centers and more sparsely populated rural precincts, and as such is less readily identifiable to a single class of population density. The exception is the urban precincts of the western Cuyuna Range cities, from Brainerd to Cuyuna. Support for Oberstar in these three counties seems to be more a function of proximity and affiliation to Duluth and the Iron Range cities, the traditional Democratic core of MN-08, as distinct from proximity and affiliation to the exurban communities of the southernmost counties, which have more in common demographically with MN-06 than with Duluth.
Among the western counties (Beltrami, Hubbard, Wadena, Cass) Oberstar fared slightly better than in the extreme southeast. He won four of sixteen in Beltrami, five of thirty-four in Hubbard, five of twenty-four in Wadena and seventeen of seventy-two in Cass. Support for Oberstar in this western bloc of "red" counties was fairly evenly split among relatively densely populated towns and rural precincts.
At the level of the county, Oberstar’s support in the northwest (Koochiching, Itasca) was the lowest among the six counties forming the northern "blue" bloc within MN-08. Oberstar won only fifteen of thirty-six precincts in Koochiching, but he swept the most densely populated precincts in and adjacent to International Falls. In Itasca, Oberstar won thirty-seven of seventy-seven precincts, notably the urban and adjacent precincts of the western Mesabi Range, from Grand Rapids to Keewatin.
The far northeastern counties of Lake and Cook were both won at the county-level by Oberstar, by margins of 13.11% and 15.15% respectively. While Oberstar won all twelve precincts in Cook (Grand Marais on the coast, the Gunflint Range in the north), he won only eleven of eighteen precincts in Lake, notably the coastal population centers of Two Harbors, Beaver Bay and Silver Bay.
St. Louis county is of profound interest to this discussion, because the results of the voting at the 178 precincts show a stark divide among: 1) the northern sparsely populated precincts, a majority of which were won by Cravaack as in Koochiching and Lake; 2) the north-central precincts of the Vermilion Range cities, between Tower and Ely, and the eastern Mesabi Range cities, between Hibbing and Biwabik, which were won by Oberstar, just as in the Range cities of Itasca County; 3) the south-central rural precincts between the Mesabi Range and Duluth, of which Cravaack won approximately 20%; and 4) the city of Duluth in the southeastern corner of the county, in which Oberstar won every precinct by an average margin of 16.14%. Let’s face it... when a county is approximately four times the size of Delaware, or approximately the size of Connecticut and Rhode Island put together, homogeneity is the exception rather than the rule.
Lastly is Carlton County, south of and adjacent to St. Louis. Oberstar won thirty-six of thirty-nine precincts, losing two low density precincts to Cravaack while tied in a third. The "blue" towns of northeastern Carlton County, Cloquet in particular, have strong affiliation to Duluth and southern St. Louis County.
In short, while the fossilization of "red" blocs in the extreme south and west of MN-08 is documented at both the precinct-level and county-level, the northern "blue" bloc is not quite as evident. Between these blocs, in the center of the district, there are more nuanced gradations of "blueness" and "redness" which we shall explore in more detail below.
PRECINCTS: SIGNIFICANT WINS AND LOSSES, 2010
Of course noting the precincts won by Oberstar, or those won by Cravaack, is not the whole story, as if all precincts were created equal and the margin of victory were also equal. There are "red" and "blue," and then there are the many hues along the spectrum from deep red to deep blue. One of the analytical methods I find instructive in documenting the geography of the 2010 election in MN-08 is mapping the precincts according to whether Oberstar’s percentage of the vote in that precinct was "normal" in relation to his percentage of the total vote across the district (46.59%), or significantly better or significantly worse, with "significant" defined as falling outside of a one standard deviation range from 40.53% to 52.65%. The following image shows precincts won by Oberstar in 2010 by a normal percentage in light blue, those won by Oberstar by a significant percentage in darker blue, those lost by Oberstar by a normal percentage in pink, those lost by Oberstar by a significant percentage in red, and those in which the vote was tied in white.
Essentially, this representation of the data documents the intensity of the support for Oberstar and Cravaack. One could, alternatively, represent the margin of victory along a color-ramp, but I find that this somewhat simplified image suffices to address the regional variation of DFL and GOP support within MN-08.
Here are some general district-wide numbers. Of the 433 precincts won by Cravaack, 133 were normal wins and 300 were significant wins. As one might expect, the great majority of those significant wins are concentrated in the south and the west. Somewhat unexpectedly, however, there is a marked concentration of significantly "red" precincts in eastern Koochiching County, south of International Falls, and in northern St. Louis County. Of the 347 precincts won by Oberstar, 117 were normal wins and 230 were significant wins. Again, as one might expect from previous discussions, Oberstar’s significant wins are concentrated in Duluth (every urban precinct but one was a significant win) and southern St. Louis County, Carlton County, the more densely populated precints of the Mesabi Range, from Grand Rapids to Biwabik, and the urban precincts of the northern "blue" bloc of counties.
More than anything, this representation of significant wins and losses emphasizes the "gravity model" for understanding DFL support in relation to proximity to Duluth. The problem—and I believe this is the fundamental point of decisivemoment’s comment from Part I, to which I responded above—is that even in precincts in which Oberstar’s victory might be considered decisive, these wins were effectively neutralized by the decline (swing) in DFL support from 2008 to 2010 and the decline in voter intensity.
PRECINCTS: WINS AND LOSSES, 2008
For comparative purposes, and in order to document the sea-change that Oberstar’s defeat signals for party-politics in MN-08, it is instructive to consider the precinct-level differences between the 2010 election and the 2008 election, when Oberstar received 67.69% of the district-wide vote against GOP challenger Michael Cummins. That election, two years ago, was a time when dKosopedia’s description of MN-08 as "one of the safest Democratic seats in Congress" (last edit: 13 December 2006) seemed wholly apt. The following image shows the precinct-level geography of the results of the 2008 vote.
Cummins won only fifty-three precincts, with just a few significant concentrations of precincts in the western counties. Interestingly, none of these precincts are Cummins’ home-precinct. Like Cravaack, Cummins was from the southeast portion of the district, specifically southern Pine County. Unlike Cravaack, Cummins was fundamentally unelectable, committed to a reactionary Tea Party ideology before such views were acceptable in Minnesota political discourse outside of the narrow confines of MN-06. Indeed, Cummins’ views were, and remain, extreme to the point that he was not even successful in running in 2010 for Minnesota State Senator for District 8 (south-central St. Louis, Carlton, Pine and Kanabec counties), the election-cycle when extremity seems to have been a desideratum.
PRECINCTS: SIGNIFICANT WINS AND LOSSES, 2008
As above for the 2010 election, I find it instructive to model the percentage of the precinct-level vote won by Oberstar in relation to a "normal" range of one standard deviation around his district-wide percentage of 67.69%. In this instance, I’ve chosen to represent precincts won by Oberstar yet by a percentage significantly less than the district-wide percentage in light blue, precincts won by Oberstar by a normal percentage in blue, precincts won by Oberstar by a percentage significantly higher than the district-wide percentage in dark blue, precincts won by Cummins in red (at the risk of sounding cruel, that he won any precincts at all is significant), and the ten precincts in which the vote was tied in white.
The takeaway from these data, in short, is that Oberstar over-performed in Duluth, Carlton County, the Iron Range and the population centers of the northern counties. While Oberstar won the precincts represented in light blue, they do nevertheless represent relative under-performance in the precincts of the southern counties, the western counties and rural precincts of Itasca, Koochiching and northern St. Louis counties. These areas of under-performance correlate strongly to those regions where the vote went from "blue" to "red" in the 2010 election, as we shall see below.
PRECINCTS: SWING, 2008 TO 2010
I wish to present the data for precinct-level changes from the 2008 election to the 2010 election in two ways: first, as a relatively straightforward map of the status of precincts from 2008 to 2010 (stays blue, blue to red, etc.); second, as a map on which the percentage-point swing against Oberstar is classified on a color-ramp. Both methods are useful for documenting the nature of a profoundly altered political landscape. The following image represents the status of each precinct from 2008 to 2010, according to the following values:
Color | 2008 | 2010 | # precincts |
---|
Blue | Oberstar | Oberstar | 344 |
Red | Cummins | Cravaack | 51 |
Yellow | Oberstar | Cravaack | 377 |
Green | Cummins | Oberstar | 2 |
Grey | Oberstar | tie | 20 |
Brown | tie | Cravaack | 5 |
Purple | tie | Oberstar | 1 |
White | tie | tie | 4 |
To the eyes of this progressive Duluthian, this is the cold and brutal reality of the 2010 election in MN-08. The great majority of those precincts won by Oberstar in 2008 and held in 2010 are from several narrowly defined regions: the urban precincts of Duluth; Carlton County, southwest of Duluth; the Mesabi and Vermilion Ranges; and Cook County in the far northeast. Beyond these regions, Oberstar held the principal urban centers of Koochiching and Lake Counties and of the Cuyuna Range, as well as a smattering of population centers in the south and west. The two precincts going from "red" to "blue" had only twelve and nineteen voters in 2010, and their outcomes were skewed by greater than average votes for third-party candidates. Cravaack, in addition to holding almost all of the precincts won by Cummins in 2008, picked up almost all of the precincts identified previously as having gone for Oberstar in 2008 but by a significantly low margin, and picked up as well five of the six in-play precincts in which there was a tie in 2008 (the four precincts tied in both 2008 and 2010 had zero voters in both election). Quite simply, this was a sweeping indictment of our long-time DFL incumbent.
Even in precincts won by Oberstar, Cravaack met with considerably more success than did Cummins in 2008, as is evident in the following image documenting the swing in the percentage of the vote won by Oberstar from 2008 to 2010.
Precincts in blue represent precincts where Oberstar increased his percentage; of these five precincts, all had less than thirty voters and four were skewed by unexpectedly high percentages for third-party candidates. The five precincts in which the percentage stayed the same (four had no voters in either election; one had Oberstar winning 80% of five voters in both elections) are shown in white. The 794 precincts in which Oberstar’s percentage of the vote declined from 2008 to 2010 are shown on a red-scale, with light pink indicating relatively low decreases through dark brown indicating the greatest decreases.
There are five clear concentrations of precincts in which support for Oberstar swung the most from 2008 to 2010: first, central and southern Pine County, roughly from Finlayson to Pine City along the Interstate 35 corridor; second, the rural precincts of central Aitkin county, along State Highway 65; third, Morrison County and southern Crow Wing County, inclusive of rural and urban precincts, notably the densely populated precincts of Little Falls in Morrison; fourth, the rural precincts of north-central Cass County around Leech Lake; fifth, the precincts of northern St. Louis County, northwest of the Vermilion Range city of Tower; and sixth, the urban and rural precincts of the eastern Mesabi Range of central St. Louis County, from Mountain Iron through Biwabik and Aurora to the Lake County line. It is these last two regions in St. Louis County to which Kossack decisivemoment made reference in his/her comment to Part I.
Of course, any swing at all in Oberstar’s precinct-level percentage of the vote from 2008 to 2010 is a bad thing for Northland DFLers. That said, it was simply not realistic for Oberstar to maintain in 2010 the overwhelming percentages he received in some of these regions in 2008 when, frankly, my extremely cranky cat would have polled well against Cummins. So "yes," Oberstar’s swing in northern and central St. Louis County in particular was dramatic, having won those precincts in 2008 with an average of 76.73% of the precinct-level votes. Indeed, in spite of a swing of more than twenty percentage-points, Oberstar nevertheless won many of these precincts in 2010, but won them with an average of only 54.09% of the precinct-level vote. Wins? Yes. Overwhelming wins in line with the 2008 election? No. Anticipated? Honestly, yes. Disappointing and frustrating nonetheless? Absolutely. In many ways, and in light of the concentrated support for Cravaack in the southern and western counties of the district, Oberstar’s wins in northeastern precincts amounted to Pyrrhic victories.
PRECINCTS: VOTER INTENSITY, 2008 TO 2010
In Part I, I noted my personal frustration with the low voter intensity in northeastern MN-08, especially in the city of Duluth. In that diary, I had employed as a measure of voter intensity (and as a proxy for voter turnout) the ratio of the total number of registered voters in the county (early registration and same-day registration) to the total number of votes cast. Here, in light of Kossack mbayrob’s constructive criticism of that section of the discussion in Part I, I have recalculated voter intensity as each precinct’s contribution to every 1,000 votes across the district (votes per thousand, or vpk). These normalized values were calculated at the precinct-level for the 2008 and 2010 elections, as was the difference from 2008 to 2010. The precinct-level differences are represented in the following image, where dark purple indicates precincts in which vpk decreased significantly, pink shows precincts in which vpk decreased, white is for precincts in which vpk was unchanged, light green represents those precincts where vpk increased, and dark green shows those precincts where vpk increased significantly. The inset is a detail of the difference in vpk from 2008 to 2010 for the precincts of Duluth.
In the most general description, vpk decreased in 273 of 804 precincts, and decreased significantly in 50 of those 273. The value for vpk was unchanged from 2008 to 2010 in five precincts, four of which had zero voters in both elections. Vpk increased in 526 precincts, and increased significantly in 52 of those 526. We can chart the change in status from 2008 to 2010 in relation to the outcome of the 2010 vote within the precinct.
status | Oberstar | Cravaack | tie |
---|
sig. decrease | 42 (12.11%) | 8 (1.84%) | 0 (0.00%) |
decrease | 102 (29.39%) | 114 (26.33%) | 7 (29.17%) |
no change | 0 (0.00%) | 1 (0.23%) | 4 (16.66%) |
increase | 189 (54.47%) | 272 (62.82%) | 13 (54.17%) |
sig. increase | 14 (4.03%) | 38 (8.78%) | 0 (0.00%) |
These data document at least in part the reality of, and the effects of, the "enthusiasm gap" in MN-08 in 2010. Precincts won by Oberstar tended to fall in traditional DFL stronholds and tended to decrease in their vpk, while precincts won by Cravaack tended to increase in their vpk. Bluntly, DFLers stayed home and the GOPers went to the polls.
Nowhere is the problem of low voter intensity more apparent than in the city of Duluth, shown in the inset of the image above. While the average percentage of the 2010 vote won by Oberstar in the thirty-six Duluth precincts was 62.73%, the nature of these victories must be qualified by context: first, the average swing from 2008 to 2010 against Oberstar in these precincts was 16.59%; second, the net loss in vpk was 21.84 or, in other words, voter intensity in Duluth in 2010 was such that it represented only 84.21% of its district-wide impact in 2008.
Perhaps the most effective means for representing the precinct-level change in vpk from 2008 to 2010 is through the following cartogram, a map of the 2010 precincts in which the area of each precinct has been transformed in direct relation to the increase, or decrease, in vpk as a percentage of the vpk in 2008. The yellow lines indicate county boundaries.
Sobering, no?
PRECINCTS: URBAN AND RURAL, 2010
It is fortuitious that Meteor Blades posted an Open Thread for Night Owls last evening concerning the national pattern of GOP gains in rural House districts in 2010, with "rural districts" defined somewhat loosely, yet productively, as those "dominated by small towns." With the exception of Duluth (population circa 85,000), this characterization of a rural district describes MN-08 quite well and, to my chagrin, MN-08 is among those thirty-nine rural districts shifting from Democratic to Republican control in 2010.
Rather than limit myself to the problematic categories of "rural" and "urban" employed by the United State Census (they are effectively out of date, they do not account for the recent growth in emergent exurban communities in southern MN-08 in particular, and the classificatory schema doesn’t really make sense to describe the nature of population centers in the Northland), I have chosen to calculate population density for each precinct and then classify precincts according to whether this population density is "dense," "moderate," "low" or "sparse." In lieu of a map, it seems to me more productive to provide a table relating the class of population density to the outcome of the 2010 election and to voter intensity. For present purposes, I am omitting districts in which the 2010 outcome was a tie.
type | DFL-DFL | GOP-GOP | DFL-GOP | GOP-DFL | 2008.vpk | 2010.vpk | diff.vpk |
---|
dense (n=80) | 56 | 0 | 24 | 0 | 258.45 | 233.39 | -25.06 |
moderate (n=223) | 109 | 6 | 103 | 0 | 419.12 | 423.40 | +4.28 |
low (n=194) | 56 | 13 | 122 | 0 | 222.99 | 235.54 | +12.55 |
sparse (n=307) | 123 | 32 | 128 | 2 | 99.44 | 107.67 | +8.23 |
The fundamental takeaway from these data, in my opinion, is the significant decrease in voter intensity among the densely populated precincts in the district, a loss of 25.06 votes per thousand, and the consequent increase in voter intensity in the less densely populated precincts, particularly among those classified as "low" (gain of 12.55 per thousand) and "sparse" (gain of 8.23 per thousand). In the end, it is not only interesting that MN-08 is a rural district that shifted from "blue" in 2008 to "red" in 2010, fitting the pattern noted in Meteor Blades’ Open Thread last night, but it is also of profound significance that the voting population in 2010 became increasingly rural.
CONCLUSIONS
The raison d’être of this two-diary series on Congressman Oberstar’s loss is my depression, frustration and general disbelief that the good citizens of MN-08 failed so miserably and managed to elect Republican (Tea Party Lite) Chip Cravaack. That said—and I do feel better now, thanks for asking!—it was cathartic for me and I hope useful for others to attempt to come to terms with how precisely this dreadful fate has befallen us Northland DFLers. In answering this question I believe we need to address two aspects of the 2010 election: where did Cravaack win the election? ...and where did Oberstar lose the election?
Cravaack benefited from the steady fossilization of a Republican bloc of precincts and counties in the south and west of the district. These areas are going to prove difficult for DFL success in the future, and it is difficult to imagine that this weakness will be remedied in the redistricting process. Cravaack also benefited from the less geographically discrete yet increasingly significant support of rural precincts (sparse to low population density). Perhaps above all, Cravaack benefited from the simple fact that he isn’t Michael Cummins. The anti-incumbent mood of the nation was clearly demonstrated in MN-08, and this succeeded for three reasons: Cravaack’s ability to position himself as a moderate in comparison to many of the Tea Partiers on the national stage; his ability to draw off a certain amount of traditional DFL support through his union credentials (he was a member of the Airline Pilots Association and shop-steward for Northwest’s Minneapolis-based Flight Engineers), especially in the Iron Range; and finally through what amounted to a shrewd campaign of "respectful disagreement" with Oberstar, a campaign that played very well among "Minnesota-nice" Northlanders even if the rhetoric belied the uncivil tone of many of his supporters in the south of the district, who were on full display at the debate between the candidates held in Duluth on 19 October.
Oberstar lost the election due in large part to the simple fact of his incumbency and, paradoxically, his record of success over his decades in office. His demonstrated ability to bring funding home to MN-08 was turned into an irremediable liability. He was ultimately doomed by low voter intensity, by the "enthusiasm gap," notably among urban DFLers, through an as yet unquantifiable combination of staying home and... it must be said... voting for Cravaack. The onus is, in my opinion, borne most heavily by DFLers in Duluth.