Politicians I almost always support, and a bunch of people on my side of the political spectrum are making hay out of the Star Tribune’s report on recent census data saying black household income plunged 14% from 2013 to 2014, and the poverty rate rose form 33% to 38%. This doesn’t seem odd for one year? You know how we rag on conservatives for repeating claims that aren’t facts, but are too politically convenient to not use? Why can’t they just exercise some skepticism? In this case, hate to say it, now we’re doing it, repeating what’s convenient without giving it a close look. Black poverty and the racial income gap is our issue, and a big jump like is so handy for making our case. Too bad it makes no sense. To cut to the chase, the only explanation for the big increase that does make sense is statistical noise from a small sample size.
Just to be clear, I’m not accusing the Star Tribune reporters of lying. I think they tried to get this right. I’m likewise not accusing the Census Bureau of lying. I don’t doubt they’re just reporting what they found. The paragraph that should have clued us in that something was funky, even before thinking things through, was this:
The American Community Survey (ACS) is an ongoing survey conducted by the U.S. Census Bureau, with questions on a variety of topics such as income, race, employment, educational attainment, commuting to work and housing. The survey goes out each year to about 3.5 million households nationwide (about 100,000 in Minnesota), with a response rate of about 97 percent. Data estimates generated from these surveys have a degree of uncertainty associated with them, called sampling error. In general, the larger the sample, the smaller the level of sampling error.
So how big was the sample? The Star Tribune didn’t say. How big is big enough? I don’t claim to know. What I did notice was that in seeking an explanation for the plunge in income, there wasn’t another one that held up to even cursory scrutiny. And other statistics don’t fit. Let’s think this through.
What could cause such a plunge in income? An economic crisis would do it. If the time period being measured was 2008 or 2009, that would make sense, but this time there wasn’t an economic crisis. Growth may be unacceptably slow, but that’s not a recession let alone a repeat of late 2008. And that still begs the question of why only black income fell. It might fall more than other racial groups’ income in a bad recession, but to plunge when most stay the same and Asian income supposedly rose drastically? That’s according to a chart that accompanied the print version of the linked Star Tribune article but which isn’t in the web version.
The article says the state demographer speculated that there might have been a rise in blacks going back to school and cutting back working hours. That many people? And only blacks? Seems implausible. We do have a large population of African refugees in Minnesota, and they count as black, but it’s not like they all arrived last year.
Was there some disaster that hit mostly black people? That might account for it, and Hurricane Katrina flooding New Orleans shows such racially disparate disasters are possible, but nothing like that happened here in 2013-2014.
What else would we expect to happen if black income plunged 14%? Unemployment would shoot up — except it stayed the same. Blacks might be affected the worst, but other groups would see decreases too — but they didn’t. About that plunge in Asian poverty as drastic as the increase in black poverty, what would cause that? Did a load of large employers move into the state but hire only Asians?
Whatever explanation you want to offer, it has to account for the increase in black poverty, and the decrease in Asian poverty, and the unemployment rate staying steady, and the economy in Minnesota being overall pretty good. If you have something other than sampling size, chime in.
Is it possible the error was the other way, that 38% is right and the earlier 33% was too low? Sure. But that still means there wasn’t a real rise, and my argument is with using these figures to say there was a rise. Isn’t it possible there is still a racial income gap? Yes, and that seems indisputable, so why do we need a suspicious decrease in income to make that point. We don’t need for the movement in income or poverty to be true to have a solid case that the income gap is real and black poverty is high.
So why then use a questionable claim? Using it means that if it turns out to be false, our whole case is undermined. I say “if” to leave open the possibility it’s real and explained by something no one has thought of yet, but you can’t claim something is a fact when it’s merely marginally possible. So I’m asking those of you on this side of the issue to stop repeating this, and use some skepticism not just about the claims of conservatives, but about what appear to be facts to support our side. I know and like State Sen. Jeff Hayden, I agree with him just about always, but sorry Senator, I’m looking at you. Hayden is right that the issue hasn’t received enough urgency, but hanging the case on this particular report will backfire if it doesn’t hold up and, frankly, it likely won’t hold up.
Maybe the way to make the case is to admit doubt about the rise in poverty but, even if the exact measure isn’t sure, there’s no reasonable doubt that there are big racial disparities in income. The black unemployment rate might actually be steady, but it’s still higher than for whites (I didn’t see anything about the unemployment rate for other non-whites). The problem has been persistent and deserves more attention. Call for specific measures to address and demand to know why there has been no movement on them so far.
But don’t use what might not hold up. As a general rule, a way I discipline my own thinking is to ask what I would think of a possible fact if it came from conservative sources. Then show that same skepticism regardless of the source. To put it another way, what questions might you ask if you preferred the new information to be false? Ask them. I’ll go further and say that the more shocking and the more convenient for our case, the more we should question it. In this one case, I didn’t really have to stop myself because such a big change seemed utterly implausible. Convenient, but implausible. Unfortunately, given how DFL politicians are using it, and so many people on the left are passing it along as fact, maybe we need to do better.
In short about the matter at hand, the story is probably wrong, and we don’t need for it to be right to make our case about the racial income gap.
From Greg Laden: As a thumbsuck rule of thumb, if you sample this sort of thing at about n=200, you can be roughly confident that you are probably within 5% either way. The margin of error on some of the ACS data from 2013 that I looked at calculates to about 4 percent.
There may well be a statistical or sampling reason for this shift, but 14% is too big for that, probably. Perhaps some other kind of error, like a change in reporting. On the other hand, of the 50 states, only about 7 are way out of whack like Minnesota is, so if there is a 1 in 20 chance of an accidental random statistical outlier, then we’d expect to see three or four for the US.
In any event, without some sort of further explanation about process, a number like this is meaningless because … well you have to give it meaning!
From Dan Burns: My own thought when I saw this is that there probably was a spike in blacks moving to MN, from elsewhere in the U.S. and from abroad, due to the better opportunities here, who are not yet employed. The claim that African-Americans in Minnesota are worse off than in Mississippi, which the Strib put right near the top as I recall, is what spiked my skepticism meter. That wouldn’t happen all of a sudden, and it would have become plenty evident before now.
I agree now that it likely is at least primarily a sampling anomaly.
From Hank Fischer: Eric had it right. On Sunday, Sept 20 A letter to the editor of the Trib showed how the info got messed up.
INCOME AND OPPORTUNITY
Consider all of the factors that can influence the data- Strib letter on racial income disparity
While racial disparities abound, both in the state and nationally, good policy requires good data, not yellow journalism or grandstanding politicians (“Income plunges for blacks in state,” Sept. 17, and “Black leaders assail Dayton on racial gap,” Sept. 18). What changed? That’s what you need to ask when you see such a stark, one-year shift in data for a particular group, such as the reported one-year plunge in black or African-American household income. And the answer is there in the U.S. Census Bureau’s American FactFinder website, and it’s about shifting demographics, not economics.
From 2013 to 2014, there was a reported net increase of 12,000 recent immigrants from Africa to Minnesota. That year saw the addition of 7,000 black or African-American households to the state, about 5,000 of which are nonfamily (and thus single-income) households. As a result of this demographic shift toward single-income households, median black or African-American household income showed a drastic fall, even while individual income remained statistically flat at about $16,000 per capita.
If you want to increase median household income quickly, then building a good recent- immigrant dating scene would be the way to go. However, if you want to address the real problem, then policies must address the low per capita income in all communities of color.
GAKU SATO, St. Paul