The Case of the Disappearing Black Voter, Part II
Forty percent of Sasha Issenberg's Milwaukee black voters have never existed.
I'm not being fair, but in a recent post for Slate, Sasha Issenberg begins by dramatically claiming that, “Sixty percent of Milwaukee's black voters have disappeared” and goes on to detail how that could be a concern for Democrats this November.
Matt, the blogger at Milwaukee, pointed out that Issenberg's numbers don't make sense. While Matt is sensitive to the difficulties involved with turnout drives in Milwaukee's low-income African-American neighborhoods, he checked out the math. Using data from the League of Young Voters and analysis from the New Organizing Institute, Issenberg reports that “160,000 African-American voters in Milwaukee were no longer reachable at their last documented address – representing 41 percent of the city's 2008 electorate.” But Matt points out that there aren't 160,000 black voters who were able to go missing citing a City of Milwaukee report which lists that there the voting age African-American population is 154,335.
Since Matt is a semi-anonymous blogger who responds to my suggestions to improve his graphs and answer my data analysis requests, I trust him implicitly – at least over Sasha Issenberg who I've only been following on Twitter for a week, since I read that great New York Times blog post. Still, I wanted to check it all out. I mean fact-checking is 2012's Clap Your Hands Say Yeah.
American Fact Finder using the 2010 Census lists 161,053 for the 18 and over Black or African-American alone or in combination with one or more races population group in Milwaukee.
Hmmmm. The existence of 161,053 people at least makes it possible to lose 160,000 people, but it's still pretty unlikely that Democratic allied groups can only locate 1,053 black people. Tom Barrett didn't do that poorly this summer. At first, I thought Issenberg or his sources conflated Milwaukee the city with Milwaukee County. But Milwaukee County only has 173,862 black people of voting age, so probably not.
Back to the Issenberg article:
Starting in April, they spent eight weeks knocking on 120,882 doors across 208 of Milwaukee's 317 wards to raise awareness of the gubernatorial recall election scheduled for June. The doors had one thing in common: the voter file said they were all home to a registered voter whom a commercial data vendor had flagged as likely to be African-American.
120,882 doors? American Fact Finder only lists 88,217 black households in Milwaukee (city) and 94,661 black households in Milwaukee County using the 2010 Census data.
I don't know how unnamed commercial data vendor built their list of doors to knock on, but I hope the League of Young Voters didn't pay by the address because the 2010 Census makes clear that around 27% of those addresses aren't going to contain an African-American household.
Having a good number of old addresses to check might be useful for a rigorous turnout drive, but it's probably not a good idea to use the entire list for an estimate of a group's population size when we have the Census data. Issenberg reports the League of Young Voters found “31 percent of their targets.” Perhaps we can do a better job figuring out how many “missing voters” that there might be in Milwaukee with respect to this list?
31 percent of 120,882 doors on the list is 37,473. Since there were 88,217 black households in Milwaukee (let's stick with the 2010 city data to make our job easier), the League of Young Voters probably found more like 42% of all the black households in the city. Missing address for 58% percent of African-American households is close enough to 60% for Issenberg's 3/5 claim in the title and lede to pass a fact check. At least my fact check. It's the implications of that number which fall apart a bit.
So how many black voters are “missing” from Milwaukee?
The League of Young Voters is missing addresses for 58% of African-American households, so assuming that a “missing” household and a “confirmed” household are the same size (1.8 people), they are missing addresses for 58% of the African-American voting age population which, using the 2010 Census data, would be about 93,000 people. Using the household total we can say that this number could be as small as around 50,000 and as large as 124,000, but as Mat pointed out there aren't 160,000 missing address.
Assuming that there are only 93,000 “missing” African-American voters amounts to the Democrats having a particularly hard time targeting a little over 33% of the 2008 city electorate as opposed to the claimed 41%. For 41% of the entire electorate to be missing, each missing African-American household would have to be made up of much less than 1.8 people which is certainly possible. But more to the point, CNN's 2008 exit polls found 91% of blacks voted for Obama which means Democratic turnout efforts are without addresses for at least 40% of their former voters (and presumably there are also white, hispanic, and other voters who have moved without notifying the DNC as well whom we have not tried to count). I'm sure that's a number which creates a plenty big hurdle for Democratic efforts.
But I'm curious how big a deal it is to not have this many addresses. Are the African-American residents of Milwaukee so much harder to pin down than those in Detroit, Philadelphia, Richmond, and Miami? What percentage of African-American voter addresses did the 2004 and 2008 get out the vote efforts begin with? Also, do missing African-American addresses really matter for actual door-knocking campaigns given how predictably located African-American voters are in Milwaukee as a result of the incredibly high rates of racial segregation in housing?
Anyways, I'm still really looking forward to Sasha Issenberg's new book – which is coming out this week – but I hope he has been a little more careful with the details than in this blog post at Slate.