The government will be flying blind with an inaccurate 2020 Census

[Yo_Mama_Been_Loggin]

The government is flying blind without accurate information about the people they support. So why is the White House trying to throw a wrench in the 2020 Census?

In March, the Trump administration proposed adding a new census question about whether household members are U.S. citizens or not. On the face of it, this might provide more information about the U.S. population. But that’s only if people actually answer the question. Demographers and population scientists are very concerned that the presence of this question will decrease participation in the census. So we’ll end up knowing less than when we started.

Twenty years ago the anthropologist Nancy Scheper-Hughes wrote about something she called “demography without numbers.” Her point was that many data sources that we rely on to understand the world are actually fairly inaccurate.

In northeast Brazil, her research found that high infant mortality rates and poverty combined to compel mothers not to record their children’s births and subsequent deaths. Fees were charged to record vital statistics, and they seemed beside the point if the child had already died. The effect of this absence in the official record was an artificially low infant mortality rate, that didn’t reflect the reality of northeast Brazil.

Professor Scheper-Hughes turned to other methods to attempt to figure out the infant mortality rate — her “demography without numbers” was composed of speaking with priests, pharmacists, hospital attendants, coffin makers, and gravediggers. Is this where we are headed?

When social and behavioral phenomena aren’t counted on purpose, it leaves those who want to tackle them at a loss. We saw a poignant example of this last month, when Canada’s Standing Senate Committee on Social Affairs, Science and Technology released a report about forced adoptions in Canada. That report offers a disconcerting example of just how much we don’t know when we don’t have accurate data. It stated “there is no official data of how many unmarried women were coerced into relinquishing their babies for adoption.” Canadian government statistics indicate that between 1945 and 1971, “almost 600,000 infants were born to unmarried women and were recorded as ‘illegitimate births.’” But there’s no way to know how many of those women were compelled to place their children in adoption and, according to the report, made to “just forget about their babies, to never speak of them again.”

And other examples abound — just to point to one, a May 2018 study from Harvard on the uncounted deaths in Puerto Rico after Hurricane Maria concluded that while the official death toll from the hurricane was 64, the actual number was “more than 70 times the official estimate.” Only last week did the government quietly increase the death toll.

If our 2020 Census is written in such a way that many people decline to answer — because they are already vulnerable and the question puts them on their guard — we will have gained nothing, but lost accuracy. The purpose of the Census isn’t to build a data wall between citizens and non-citizens. It’s constitutionally mandated in order to give us an accurate picture of who lives here — whether or not they are authorized to be here.

http://thehill.com/opinion/white-house/401407-the-government-will-be-flying-blind-with-an-inaccurate-2020-census

[Igel]

1. I've known lots of people who declined to answer questions.

In whole or in part.

They found them offensive or were suspicious of the motivation. 1980. 1990. 2000. 2010. Every year, some group objected.

When they were fundie Xians objecting and it did make the news, the tone was, "You're violating the law if you don't answer! Eat your peas!" When they were left-of-center objections, the tone was, "Bad Census Bureau! You're wrong in even asking these questions."

I didn't object or carp when people didn't return their forms. I did point out that government services and electoral apportionment rely on their returning the form, however, and if they don't and they're undercounted then they're shortchanging their local and state governments and their own voting strength.

Many questions on the census are there for the sole purpose of gathering information for the purpose of implementing laws. Every one tweaks the response rate in some way. I like the stripped down limited provision. Write down how many people are there and be done with it.


(As for the PR numbers, the problem's messier than the write wants to admit. "Death due to" implies immediacy. Not "my grandfather died because he had no electricity went out and he had to walk 5 miles every other day to the store for ice, and 3 months after Maria hit he was killed by a drunk driver on the way back home from the store." There needs to be two accurate numbers: Those killed by the storm fairly directly, and the increase in mortality in a specified amount of time, just for comparison purposes. The "actual number" is as arbitrary as the official number, masked by the word "actual" as though it had greater legitimacy. One study gave the most likely value of excess deaths given statistical treatment of the results, with rather large error bars which were completely ignored by the press who probably thought an "error bar" was the kind of drinking establishment one entered by accident; the other picked an arbitrary time frame and compared the deaths with those from the previous year for the same period ... Which, of course, misses that some things are random and random doesn't mean "it averages out".)