By Craig McDermott, cross-posted from Random Musings
As the US Census Bureau releases local data to various states for redistricting efforts (Louisiana, Mississippi, New Jersey, Virginia this week, Arkansas, Indiana, Iowa, Maryland next week), they're holding a number of press conferences.
Wednesday, Dr. Robert M. Groves, Director of the Census Bureau discussed "the upcoming release of state redistricting data products."
The transcript of the presentation is here; a .pdf of the Powerpoint slides is here.
As the data for Arizona hasn't been released yet, I didn't expect much from the presentation other than to provide a little background for when AZ data *is* released. However, my ears perked up when the following slide was presented –
This slide (page 6 of the presentation .pdf) shows how much actual state census counts varied from pre-census estimates based on demographic analysis. The estimates were actually pretty close –
– In 34 of 50 states, the actual counts were within +/- 1% of the estimates, and 46 of 50 states were within +/- 2%.
As you can see from the graph, the margin of error spread was pretty balanced, but the interesting point (for AZ readers, anyway) is that the one state where the actual count was more than 2% less than the estimates was, of course, Arizona (page 9 of the transcript .pdf). In AZ, the actual count was 4% less than the estimates.
When asked by a reporter from the Arizona Republic (Ron Hansen) about the variance, Dr. Groves didn't have an explanation or even speculation, saying only that they're looking into it and that they'll have more information when they have more information, specifically on local level variances and explanations.
However, I'm not a trained statistician (as if you hadn't noticed 🙂 ), so I will be happy to engage in a little speculation.
I think there are three main reasons for the variance – fear, hatred, and economics -
1. SB1070 and the related anti-immigrant hysteria. Many immigrants either have left the state or simply avoid contact with public officials (such as census workers) as much as possible. Even legal immigrants fear the harassment that comes from contact with emergency and public service personnel.
2. The hatred of the federal government that has taken hold of the Arizona GOP and its adherents. Many people simply refused to respond to either the mailed surveys or when actual workers were sent out to "fill in the gaps." They don't hate state or local governments, because in most of AZ, those are run by people who are "good ol' boys," just like them.
While the reasons may have differed, a significant part of AZ's population self-selected themselves for undercounting.
3. The cratering of Arizona's economy seems to have led to an significant outflow of residents. Anybody who canvassed neighborhoods for any candidate in the 2010 election noticed a huge number of empty homes. This may not seem to be purely political, but as more people watch the Republican majority in the legislature and the rest of the state government focus on tea party issues/corporate giveaways while ignoring the state's economic and fiscal crises, it shouldn't be surprising that many have just given up hope of making a good life for themselves and their families.
That trend seems to have been accelerating over the last 18 months or so, and may have skewed the estimates – people that were here when the estimates were formulated weren't when the physical count was conducted.
We'll see what happens when the local level data for AZ is release (personally, I expect an outcry of "we wuz robbed!" from the RW blogosphere).
All local level data will be released by the end of March (a statutory deadline). However, the exact date for the release of Arizona's data hasn't been announced yet.
Each week, the Bureau will announce which states' data will be released the following week. After that, the data will be shipped to the states' leaders (i.e. – the governor and caucus leaders in the legislature). Once the receipt of the data by the leaders is confirmed, the data will be released to the general public and media, generally 24 hours after the state leadership gets it.
Eventually, the data will be available via FTP download here and on the Census Bureau's American FactFinder page here.
A related blog post from the Population Resource Center is here.