So, after much discussion last week over whether or not Oakland is experiencing an “exodus” due to crime, we now have a Trib article weighing in with Census Bureau data suggesting Oakland’s population has decreased dramatically since 2000. First, I’ll note that the Census Bureau’s official 2007 population estimate for Oakland is 401,489, not 372,000 as the article suggests. So where does the smaller figure come from?
So, there’s this thing called the American Community Survey. Wait. Let’s back up. So there’s this thing called the decennial Census. Every ten years since 1790, the government tries to count everyone living in the US. When we started, this was accomplished by sending US Marshals out on horses to look for settlers. Now, it’s this massive operation that involves tons of advisory committees, setting up regional offices, hiring hundreds of thousands of people, and costs billions of dollars. Because counting every single person in a country of over 300 million is…well, it’s really hard. And over the years, the Bureau has found that some populations are harder to count than others, particularly children, racial and ethnic minorities, and more generally, those of large cities. After an embarrassing debacle with the 1990 census, which the Bureau eventually concluded had undercounted the US population by a whopping 16 million people, we took a number of steps to ensure greater accuracy in 2000.
How to best do this was a subject of much controversy, and eventually, a Supreme Court case, which is actually a pretty fascinating story on its own, and one I might try to tell here someday during a slow news week or something if people are interested. The short version is that the Bureau wasn’t allowed to do the count the way they wanted. Instead, they relied on a whole lot of fieldwork, partnerships with local government, technology, and more user-friendly forms. The standard form, sent to 83% of housing units, asked only seven questions (the smallest number since 1820). The remaining 17% of households received a longer form (called, um, the “long form”) with an additional twenty-seven questions reflecting more detailed demographic information.
The government has traditionally used long form data to create the social and economic community profiles used in allocating federal funds. Problem is, ten years is an awful long time, and one in which significant demographic changes can occur. So the Bureau decided to address this issue by initiating a project to replace the long form with something called the American Community Survey. Surveys with the same questions as the Census long form are sent randomly to 250,000 households every month, for a total of 3 million households annually. These responses are then used to create annual demographic profiles for all states, cities, counties, and metropolitan areas with a population of 65,000 or more. The Bureau began testing the ACS in 1996 and began full implementation of the project in 2005.
So last week, the Census Bureau released a new set of ACS estimates, based on the combined survey results from 2005 to 2007. That’s what today’s Trib story, “Exodus of blacks saddens Oakland” is about. And I know I already said this, but the point is not made in the story, and it’s really important, so again, I’ll say that this is a demographic profile, not an official Census Bureau population estimate. Although the ACS three year survey results mark Oakland’s estimated population at 372,247, the Bureau’s official 2007 estimate for Oakland is 401,489.
So why would the numbers be different? For starters, the ACS numbers do not reflect group quarters populations. That’s people living in nursing homes, residential treatment centers, group homes, halfway houses, dorms, shelters, and so on. In 2000, this was estimated as 7,175 people in Oakland. The ACS is now starting to try to count group quarters, but the challenges of doing so are ongoing. Hopefully, the issue will be resolved within a few years.
The second problem, one which I think will be more difficult to resolve, is the difficulty in getting accurate responses from traditionally undercounted populations. Oakland’s significant racial and ethnic diversity is a blessing in many ways, but a curse when it comes to getting good population counts. Because the ACS does not enjoy the same level of resources when it comes to field workers and follow up as the decennial Census, it is difficult to ensure a satisfactory response rate, particularly among populations where significant distrust of the government is a common characteristic. The detailed questions on the survey are viewed by many recipients as frightening and invasive, weighting methodology is a work in progress, and ensuring adequate funding for the project has been an ongoing problem in Congress.
So what does this mean for Oakland? Well, the failure to count group quarters is a problem, and likely a significant factor in the undercounting of our African American population. How significant the problem of non-response is – well, that’s harder to tell. I think it’s fair to say that while we don’t necessarily have good reason to dismiss the estimates out of hand, we do have good reason to question them. Ultimately, we’ll have to wait until we get the results of the 2010 Census to determine how good a marker of demographic trends the ACS has turned out to be.