Iowa’s presidential caucuses are finally (and mercifully) upon us. Unfortunately, the final results may take a while longer to show up. In states with primary elections, it's pretty simple: County election officials count the number of actual votes, report them, and that's that. With a caucus, though, the media relies on reports from the Democratic Party, so we won't see popular votes—at least not right away. (Popular vote totals will eventually surface; for instance, you can see the 2008 popular votes from the Democratic caucuses in Iowa here, via CNN.) Instead, as results start trickling in, you’ll most likely just see the number of delegates chosen at precinct-level caucuses in each county.
Each county has an assigned number of delegates, which is based on how many people vote Democratic in each county. This also correlates with how populous each county is, but not entirely. For instance, of the 1,401 delegates available this year, by far the most delegates (228) are in Polk County, which is also the county with the largest population.
However further down the list, there are slightly more people overall in Scott County than there are in Johnson County, but Johnson County gets more delegates (92, versus 82 in Scott) because Johnson County is more strongly Democratic (in other words, the total number of people voting for Barack Obama was higher in Johnson County, which went 67 percent for him in 2012, than it was in Scott County, which went 56 percent for him, even though there were more votes total in Scott County). You can see the full list of county delegate counts as part of this excellent explainer from the local blog Bleeding Heartland.
Hold on, though: It’s more complicated than that.
Even though Scott County gets fewer delegates, it’s likely that each voter in Scott County will have more leverage, individually, than each voter in Johnson County. Desmoinesdem at Bleeding Heartland goes on to explain that, for instance, in 2008 there were 134 caucus-goers per delegate in Johnson County, but 88 caucus-goers per delegate in Scott County. The number of delegates per precinct doesn’t go up or down depending on how many people show up, but is based on previous voter turnout levels (including midterms).
That winds up giving a disadvantage to the counties where there are lots of college students (like Johnson County), who tend to not show up for midterms—meaning fewer delegates two years later for the presidential caucus. And that’s why some analysts are saying that even if a huge surge of unexpected voters for Bernie Sanders does show up on Monday, its impact will be somewhat blunted; his support is heavily concentrated in college towns, and with the number of delegates already fixed in place, dramatically expanded turnout won’t get him a dramatically expanded number of delegates in those places.
Further complicating matters, the county delegate numbers you see reported on Monday night are only the first steps in the process, and the final results may turn out to be somewhat different after several levels of conventions. The delegates from the precincts will go to a county convention, and delegates from the county convention will go to a congressional district-level convention and a state convention, where they'll pick the much smaller number of delegates to the Democratic National Convention.
The Green Papers explains those steps in a bit more detail; there will be 44 DNC delegates chosen at the state convention, 9 of whom are at-large and 6 of whom are pledged superdelegates (this contrasts with the unpledged superdelegates, who can pick whoever they want). The remaining delegates are allocated by congressional district: 8 in IA-01 and IA-02, 7 in IA-03, and 6 in IA-04 (the part of the state with the fewest Democratic voters). So they have to figure out how to distill the large numbers of precinct-level delegates down into nice round numbers at the county and state-level conventions.
So, the most important night of the entire Democratic nomination process is unpredictable and largely opaque … which is bizarre, considering how much arbitrary “momentum” is assigned to the caucus’s apparent winner. (I say “apparent" because you may remember that in 2012's Republican caucus in Iowa, what initially looked like a Mitt Romney victory that night turned into a Rick Santorum victory as the days unfolded ... and then, in an often-forgotten detail, it finally turned into a Ron Paul victory by the time the state convention was finished.)
Given how the popular votes don’t translate directly into delegates, and the state-level delegates are even more distantly removed from the reality of popular votes, it seems like a fool’s errand to try and predict final delegate outcomes. However, what I can do is walk you through what the congressional districts and the state’s key counties look like demographically. That may give you at least some advance sense of what’s likely to happen at the most basic level of building blocks.
After 2012 redistricting, Iowa was reduced to only four congressional districts, which correspond neatly with the northwest, southwest, northeast, and southeast corners of the state. (You can see the current map at the top of the story.) The 4th, in the northwest corner, is by far the most conservative part of the state; this is the district represented by Steve King in the House. The other three districts are swing districts in general elections. However, merely looking at how conservative or liberal an area is in a general election doesn’t map directly on to the Democratic primary.
Take a look at the 2008 results, for instance. It’s safe to say that Barack Obama performed better in the eastern part of the state while Hillary Clinton performed better in the western part of the state. (In the map to the right, Clinton won the light blue counties and Obama won the dark blue counties.) However, Clinton put up her best results in the southwestern part of the state, not the most conservative northwestern part. (For that matter, it’s not entirely clear whether Clinton was truly running to the “right” of Obama in 2008. And additionally, John Edwards might be seen as running to the “left” of both of them in the 2008 caucus, yet fared best in the state’s rural south, not its urban areas and college towns.)
So, what’s left for clues about where Clinton and Sanders might be strongest this year? One possible source is the crosstabs of the polls. They can tell us, for instance, that Clinton outperforms Sanders among women: The most recent PPP poll of Iowa tells us that Clinton leads 54-37 among women while Sanders leads 44-40 among men. There isn’t much variation by gender around the state, though, so that’s a topic we won't focus on much. Every district hews very closely to 50-50, and even the county with the most lopsided gender ratio, Story County, is tilted only about 52-48 in favor of men. (Story County is home to Iowa State University, the “cow college” in Iowa’s higher ed system, so perhaps it has more male students there specializing in more stereotypically "male" subjects.)
And you’re most likely aware of large racial differences in the Clinton and Sanders coalitions that show up in national-level polls. That’s not much of an issue in Iowa, though, to the extent that PPP doesn’t even include racial crosstabs. Iowa, as a whole, is 87 percent white, and that doesn't vary much from CD to CD (ranging from 85 percent white in IA-03 to 89 percent white in IA-01).
One real difference maker, though, is age. PPP says that Clinton leads 48-40 among persons age 46-65 and 65-24 among persons over age 65, while Sanders leads 49-36 among person age 30-45, and 52-37 among persons age 18-29. Quinnipiac’s most recent poll sees an even more exaggerated split, with Sanders leading 78-21 among persons 18-44, and Clinton reading 71-21 among persons over 65.
So, the big question would be: Where are there a lot of young people, and where are there a lot of old people? Let's jump into the demographic data, looking at each CD, and at three key counties in each CD. (For the most part, these are the three most populous counties in each CD. Jefferson and Sioux Counties don't quite fit that criterion, though they have some interesting characteristics that we'll get into later. This data comes from the Census Bureau's 2014 American Community Survey, except for Jefferson and Sioux Counties, which, because of their smaller size, use the 2010-14 five-year sample.)
IA-01: Percentage of population that is 20-34 years old: 19.4; Percentage of population that is 65 years or older: 16.5; Median age: 38.6
Linn County: 20-34 y.o.: 20.2; 65+: 14.2; Median age: 37.1
Black Hawk County: 20-34 y.o.: 24.2; 65+: 14.7; Median age: 34.3
Dubuque County: 20-34 y.o.: 20.1; 65+: 16.2; Median age: 39.0
IA-02: 20-34 y.o.: 20.0; 65+: 65+: 15.8; Median age: 38.4
Scott County: 20-34 y.o.: 19.1; 65+: 14.4; Median age: 38.4
Johnson County: 20-34 y.o.: 30.0; 65+: 9.8; Median age: 30.5
Jefferson County: 20-34 y.o.: 21.2; 65+: 16.3; Median age: 42.7
IA-03: 20-34 y.o.: 20.3; 65+: 13.5; Median age: 37.0
Polk County: 20-34 y.o.: 22.3; 65+: 11.7; Median age: 35.1
Dallas County: 20-34 y.o.: 19.5; 65+: 11.0; Median age: 35.3
Pottawattamie County: 20-34 y.o.: 18.9; 65+: 15.5; Median age: 39.2
IA-04: 20-34 y.o.: 19.1; 65+: 17.5; Median age: 39.1
Woodbury County: 20-34 y.o.: 20.3; 65+: 13.6; Median age: 35.9
Story County: 20-34 y.o.: 32.6; 65+: 10.8; Median age: 26.9
Sioux County: 20-34 y.o.: 21.0; 65+: 14.3; Median age: 33.2
If you look only at the congressional districts, you won’t see a lot of difference. IA-04 is a bit older than the other districts, with a higher median age, a lower percentage of millennials, and a higher percentage of senior citizens … but it's still not a big difference. One reason for the age uniformity between the districts is that each district gets one of the state’s four major universities (University of Northern Iowa in Black Hawk County in IA-01, University of Iowa in Johnson County in IA-02, Drake University in Polk County in IA-03, and Iowa State University in Story County in IA-04).
Instead, when you zoom in and look more specifically at the counties, you can see how those four counties contrast with the other large counties around the state. Story County (where Ames is located) is possibly the most extreme example, with a median age of only 26.9. Polk County, where Des Moines is located, is a little closer to the statewide norm than the other three, because it’s the state’s commercial center rather than just a college town. Even then, though, it still contains a lot of 20- and 30-somethings who have graduated and are there to work, so it skews a little younger too.
Look for these four counties (Black Hawk, Johnson, Polk, and Story) to be where the greatest concentration of Sanders votes are found. (Also, bear in mind that these are the counties where the arcane delegate allocation rules that I described earlier are mostly likely to blunt any Sanders surge.) On the other hand, look for the counties with the older populations to be where Clinton performs best. Pottawattamie County (where Council Bluffs is located), for instance, is not only the one of the 12 selected counties with the highest median age and the lowest percentage of millennials, but it’s also the only county of the 12 that Clinton won in the 2008 caucus.
Jefferson County is one weird exception that deserves an extra few words. It has the highest median age of any of the counties, but it’s one of the other counties that seems likeliest to tip to Sanders. That’s because it has an unusual constituency: The Maharishi University of Management is there, meaning that there are thousands of transcendental meditation enthusiasts there. This was one of the few strongholds of the Natural Law Party back in the 1990s, and at this point, Bernie Sanders seems to be the next most counter-cultural option for them, so this small county is pegged to be one of his strongholds.
One other demographic consideration is income. Quinnipiac found that persons earning more money were more likely to support Clinton: Persons in the over $100k bracket supported Clinton 58-34. Persons in the $50-$100k bracket supported Sanders 52-46, and persons in the under $50k bracket supported Sanders 58-37.
That contrasts a bit with a Pew poll of the national primary (Pew doesn’t poll at the state level, but it’s one of the few other polls I’ve seen with income crosstabs). They found Sanders does best in their middle bracket ($40-75k), trailing only 37-33, while Clinton does better among both the most affluent (leading 49-28 in the $75k+ bracket) and the poorest (leading 47-16 in the under $40k bracket). (That may have to do with the difference between Iowa and the nation, though; at the national level, many of the people in the under $40k bracket would be people of color, who, polls so far indicate, predominantly support Clinton. In Iowa, the under $40k bracket, like every other bracket, is going to be mostly white.)
So let’s look at income at the district and county level … and one other related category, the number of people in “creative class” jobs (or what the Census calls “management, business, science, and arts occupations;” urban studies theorist Richard Florida came up with a catchier name).
IA-01: Median household income: $52,189; Percentage of ‘creative class’ occupations: 33.6
Linn County: MHI: $62,984; Creative class: 38.2
Black Hawk County: MHI: $51,685; Creative class: 34.8
Dubuque County: MHI: $56,129; Creative class: 31.5
IA-02: MHI: $51,019; Creative class: 34.9
Scott County: MHI: $54,679; Creative class: 33.6
Johnson County: MHI: $60,254; Creative class: 47.2
Jefferson County: MHI: $44,688; Creative class: 38.7
IA-03: MHI: $59,460; Creative class: 38.2
Polk County: MHI: $60,978; Creative class: 39.4
Dallas County: MHI: $77,412; Creative class: 46.6
Pottawattamie County: MHI: $52,239; Creative class: 29.4
IA-04: MHI: $49,623; Creative class: 33.3
Woodbury County: MHI: $44,721; Creative class: 28.9
Story County: MHI: $46,091; Creative class: 47.5
Sioux County: MHI: $59,272; Creative class: 32.2
So if we’re going to put a lot of stock in the Quinnipiac crosstabs, the counties where people make more money are going to go more strongly for Clinton: Johnson and Polk Counties for instance, and especially Dallas County (which is the exurbs to the west of Des Moines). But something doesn’t seem right with that idea. Johnson County, and to a lesser extent Polk, are the ones that we just predicted would be most likely in Sanders’ column, because they’re full of young people.
Another potential problem is Pottawattamie County, which has a below-average income (and a low level of creative-class occupations, meaning it's mostly blue-collar), which would put it more likely in the Sanders column … but we also know that that was one of Clinton’s strongest counties in the 2008 caucus! So it’s possible that income doesn’t actually have a lot of predictive value here. If the college towns are where Sanders does best, then maybe the flip-side applies and Clinton will do best in the blue-collar towns like Council Bluffs, Dubuque, and Sioux City (which is in Woodbury County).
Or it’s possible that you have to look at how income and type of occupation interact. Story County, which is probably going to be Sanders’ strongest county in the whole state, has both low-income and high levels of creative-class occupation: Taken together, that suggests a lot of students and other people associated with the university. (Barack Obama was dominant in Story County in 2008.) So then the real question might be what happens in places like Dallas County where you have, on the one hand, high incomes, but on the other hand the high levels of education that go with knowledge-sector jobs. It’s possible that Dallas County will be one of the most closely divided counties this time (and that’s what happened in 2008—Obama only narrowly won here that year).
Finally, let’s look at one other demographic category that doesn’t get discussed much, but that can be of key importance on Monday: Religiosity. You can see how that’d be very important on the Republican side of the primary—polls (like the Pew poll I mentioned earlier) have shown Donald Trump doing significantly better with non-religious Republicans, while the most religious are more interested in Ted Cruz (that poll is a few months old, so it had Ben Carson their top choice at the time). That may show up in terms of a geographical divide: Cruz is more likely to do better in the more evangelical western part of the state, Trump is doing better in the eastern part of the state. (If you look at the 2008 maps above, you’ll see a similar geographical pattern with Mike Huckabee and Mitt Romney.)
But if you look at the Democratic crosstabs in the Pew poll, you see a similar phenomenon on our side. Except for perhaps age, the most pronounced difference between Clinton and Sanders supporters was religiosity. Among persons who attend church weekly or more, Clinton led 50-7. Among persons who attend less frequently, Clinton’s lead was only 42-31. (Again, bear in mind that’s a national poll, and that African-American and Hispanic Democrats are more likely to be religiously active than white Democrats. White people who attend church weekly or more are, for the most part, likely to be Republicans instead. So you might see different results in an Iowa-specific poll that asked about religion, because religion on the Democratic side is somewhat conflated with race. I, however, haven’t seen any poll of the Democratic caucus in Iowa that asked about religiosity.)
The Census Bureau doesn’t ask about religion. However, the Association of Religion Data Archives collects congregation enrollment numbers every decade, creating a sort of religion census that complements the rest of the Census’ data. It breaks down at the county level, so we can get a general idea of not just how religious people in each county are, but also what denominations they belong to.
IA-01
Linn County: Evangelical percentage: 13.8 (Non-denominational 3.4, Lutheran Church — Missouri Synod 2.7); Mainline percentage: 16.8 (United Methodist Church 6.9, Evangelical Lutheran Church of America 5.5); Catholic percentage: 19.4; Unclaimed: 47.2
Black Hawk County: Evangelical: 11.1 (Vineyard 2.2, LCMS 2.1); Mainline: 19.0 (ELCA 8.4; UMC 4.0); Catholic: 15.3; Unclaimed: 45.7
Dubuque County: Evangelical: 6.0 (Non-denom. 1.7, LCMS 1.7); Mainline: 8.0 (ELCA 3.3, UMC 2.3); Catholic: 53.3; Unclaimed: 31.6
IA-02
Scott County: Evangelical: 11.6 (Non-denom. 3.5, LCMS 3.3); Mainline: 15.1 (ELCA 6.5, UMC 4.6); Catholic: 18.5; Unclaimed: 51.9
Johnson County: Evangelical: 7.8 (Non-denom. 1.7, Evangelical Free Church 1.5); Mainline: 12.4 (UMC 6.6, ELCA 2.7); Catholic: 12.0; Unclaimed: 64.6
Jefferson County: Evangelical: 10.2 (Non-denom. 4,4, Foursquare 1.2); Mainline: 17.4 (UMC 5.9, Disciples of Christ 3.9); Catholic: 6.1; Unclaimed: 62.7
IA-03
Polk County: Evangelical: 12.5 (Non-denom. 2.9, Assemblies of God 2.2); Mainline: 14.4 (UMC 4.5, ELCA 3.8); Catholic: 14.6; Unclaimed: 55.0
Dallas County: Evangelical: 14.4 (Non-denom. 7.3, LCMS 2.0); Mainline: 28.8 (ELCA 18.3, UMC 6.4); Catholic: 11.7; Unclaimed: 44.1
Pottawattamie County: Evangelical 12.2 (Non-denom. 3.3, Christian Churches 3.0); Mainline: 14.4 (UMC 5.8, ELCA 4.4); Catholic: 9.6; Unclaimed: 61.1
IA-04
Woodbury County: Evangelical: 16.8 (LCMS 5.0, Lutheran Congregations in Mission for Christ: 4.8); Mainline 17.3 (UMC 6.9, ELCA 5.5); Catholic: 22.3; Unclaimed: 40.9
Story County: Evangelical: 15.9 (Non-denom. 6.4, LCMC 2.7); Mainline 21.0 (ELCA 10.3, UMC 7.7); Catholic: 9.9; Unclaimed 50.9
Sioux County: Evangelical: 38.5 (Christian Reformed 26.5, LCMS 3.8); Mainline 39.5 (Reformed Church 33.0, ELCA 3.0); Catholic 7.3; Unclaimed 14.6
So, if those Pew crosstabs are accurate, the counties with the most “unclaimed” persons (their term for people who aren’t members of any church) would be the likeliest in the Sanders column; once again, that would be Johnson County (home of the University of Iowa), and Jefferson County (where all those transcendental meditators don’t really fit into ARDA’s schema). The most religiously devoted counties would most likely fit into Clinton’s column: The heavily Catholic (53.3 percent) Dubuque County, Dallas County (in the exurbs, and heavily Lutheran), and most of all, Sioux County in the state’s northwest corner.
Sioux County deserves a few extra words. It’s remarkably more religious than the rest of the state (which, as a whole, is also somewhat more religious than the nation)—only 14 percent of its residents are “unclaimed.” (And "unclaimed" doesn't necessarily mean non-religious; it simply means not an enrolled member of a church.) That’s reflected in how it’s also perhaps the most conservative county in the state: It went 83-16 for Mitt Romney in the 2012 general election, and it’s also a stronghold for the social conservative candidate in the caucus (it gave 53 percent of the vote to Mike Huckabee in the 2008 caucus), meaning it should be the core of Ted Cruz’s support this year on the GOP side.
You may also notice that the denominations are somewhat different than the rest of the state: Instead of Lutherans and Catholics, it’s heavily Reformed and Christian Reformed (the Reformed Church is probably the most conservative of the mainline churches, and the Christian Reformed Church is its more extreme spinoff). These churches are very much associated with Dutch-Americans (the other place, besides northwest Iowa, where they’re prevalent is western Michigan).
This corresponds with the Census Bureau’s results on the “ancestry” question, which is one demographic category I didn’t delve into much because the rest of the state is uniformly heavily German-American: 35.7 percent of the state’s residents, as a whole, claim German ancestry, much higher than any other ancestry. That goes as high as 47.9 percent of Dubuque County residents with German ancestry. Sioux County, on the other hand, is as uniformly Dutch as any place in America: 47.3 percent of its residents claim Dutch ancestry.
Sioux County, of course, simply doesn’t have many Democrats in it. As seen with its presidential numbers, it’s heavily Republican. But … again, bear in mind the delegate allocation rules in the Democratic caucus. Sioux County’s few Democrats punch above their weight, unlike the many college students in secular Johnson County, who punch below their weight.