William M. Briggs

Statistician to the Stars!

Count ’em: The White House should not control the Census

Judd Gregg rann for the door yesterday to escape from being installed as Commerce secretary in Obama’s administration. Part of why he got the willies was that he did not like Obama’s idea of forcing the Census to operate under the watchful eyes of the White House.

Here’s what our constitution says about the census:

“[An] Enumeration shall be made within three Years after the first Meeting of the Congress of the United States, and within every subsequent Term of ten Years, in such Manner as they shall by Law direct.”

Generally, and to this point in time, the word Enumeration and the phrase in such Manner as they shall by Law direct has been interpreted to mean “Everybody will be physically counted.”

So, once every decade, Census workers pop outdoors and start tallying bodies.

Everybody agrees that it is practically impossible to count every single citizen because some of us are pretty good at hiding when we see government agents heading in our direction. Other people, starved for attention, offer themselves up more than once. The gist: the official count has some error.

Some statisticians heard about this predicament and offered up their services. “We have experience making up numbers where none exist,” they said, “so we can estimate the people that are missed. To do this, we will use comforting-inducing complicated mathematical formulae. It’s science.”

This reasoning is, as readers of this blog know, irresistible to certain people. Labeling something science is enough to offer it immunity from probing or dissection. How, after all, can we mere mortals argue with science?

Other politicians liked the idea that statisticians didn’t need to actually physically count people. They could just grab a few here and there, call that a survey, plug the survey into some equations, and out would pop the desired numbers. Much cheaper and vastly quicker.

An added bonus is that the survey-method would let Census statisticians create a scientific guess on the size of any disaffected group of choice a politician might care to ask about. This was an enormous advantage because most politicians argued strenuously that their disaffected group was under-counted in the Census. The affected groups, they said, were counted just fine.

The Census is used to count how many citizens there are and where they live. These numbers are then used to gerrymander—no, sorry, draw up Congressional districts and allocate numbers of representatives. Areas which have fewer people have fewer politicians assigned to watch over them.

This idea is abhorrent to the politicians who might lose their jobs if the Census finds that fewer people now live under their jurisdiction. Thus, the loudest cries of “under-count!” are from those leaders in areas which have lost population. They are therefore eager to find a way to boost their numbers, and the best way to do that is to drag out the slogan “disaffected group!” or one of its variants.

Why is this so? Because the survey-method can, and does, count non-citizens. Now, non-citizens are fine people, but they are just what they sound like: non citizens. By trivial definition, they should not be counted as part of the official count of citizens. But if they are counted, and they are input into the arcane algorithms, they will increase the estimate of the count of people living in the area in which they were surveyed.

The mathematical apparatus that the statisticians have constructed has myriad knobs, levers, and switches that can be tweaked to produce numbers either higher or lower as directed. Twist this knob and Ames, Iowa goes up 2%, flip that switch and San Francisco drops 4%.

It would be very tempting for somebody to play with those controls to massage the numbers so that, say, the tenets of social justice are obeyed, and Congressional districts are apportioned to favor those politicians that are more properly Enlightened. A higher good is being served this way.

Which is why exactly why Obama would not want to bring a department that had been operating just fine before he came to us under his wing. Instead, presumably, he just likes statisticians around him (who doesn’t?). Obviously, he would not meddle with the independence of the Census. That is why he is drawing it closer. To not meddle.

We statisticians are lovely folk—we know some great jokes, and can integrate multidimensional integrals faster than you can crack open a peanut—but we cannot be trusted to not play with our own creations, especially when our bosses, eager for a certain result, are watching over our shoulders.

Let’s just stick with enumerate.

15 Comments

  1. If the Founders had meant an honest and truthful numeration, they’d have said “an honest and truthful enumeration.”

  2. I seem to recall this issue being addressed by the Supreme Court some years ago (probably around 2000) and I think they substantially ruled against using statistical approaches.

    The proponents argued that the statistical methods and computer models would be more accurate than taking a measurement.

    Science is definitely under attack. And the attack is not coming from the source that the media typically identifies.

  3. Even with a substantially accurate census sleazy politicians – excuse the redundancy – are wont to have their way with the gerrymander process, ala what’s happened the past decade in California. So moving responsibility for the US census from the Commerce Department to the White Temple will only change the degree of crime committed from Petite to Grande. More’s the pity. What was it Alinsky said, “Never let a good crisis go to waste”? And I thought one could trust statisticians!

  4. Re Charlie:

    Honest and truthful are implicit in the census mandate. The founders and all other rational people do not want dishonest or false enumeration. It seems that you consider the old adage, “there are lies, damn lies, and statistics”, to be beneficial. You must be a democrat from Chicago.

  5. Why should we be surprised at the Democrats plan to have the White House be in charge of the census? They have schemed to give money to ACORN. ACORN is under investigation for voter fraud. They will take fraud anyway they can get it.

  6. You failed to mention the next step which was:

    The Congress, by law directed that:

    “The Secretary [of Commerce] shall perform the functions and duties imposed upon him by this title, may issue such rules and regulations as he deems necessary to carry out such functions and duties, and may delegate the performance of such functions and duties and the authority to issue such rules and regulations to such officers and employees of the Department of Commerce as he may designate.”

  7. I remember the late David Freedman, who worked on the census for many years speaking very eloquently about his opposition to statisical adjustment of the census, and the limitations of statisical models in general. I recommend his “Statisical Models and Shoe Leather” for those who are interested in thinking clearly about these kinds of issue.

    http://www.rochester.edu/College/PSC/clarke/204/Freedman91.pdf

  8. Sorry, Paddy, irony as a mode of humor unfortunately depends on the intelligence and common sense of the audience. Clearly I didn’t gauge yours correctly.

  9. Har har. I got your joke, Charlie. Both of them.

  10. JFK,

    I am always interested in thinking clearly. Does Freedman do ladies shoes as well?

  11. Briggs

    February 15, 2009 at 7:13 am

    JFK,

    Freedman was a good guy, and sort of the informal leader for a group of statisticians who did not want to use statistical models to estimate population. It might surprise people to learn that the percentage of statisticians who feel the way we do is nearly half (according to numbers a decade old or so).

    The article you attach is excellent. Another one by prominent mathematician and statistician Persi Diaconis is this one (I don’t have an electronic copy):
    @Article{ Dia1998,
    title = “A place for philosophy? {T}he rise of modeling in statistical science”,
    author = “Persi Diaconis”,
    journal = “Quarterly Journal of Applied Mathematics”,
    pages = “797–805”,
    volume = “4”,
    year = “1998”
    }

  12. Specific use of the word “eunumeration” means sampling or adjustment with sampling is not allowed. And frankly the proponents of modeling are going to have a hard time convincing anyone that models should be used in the present climate.

  13. SteveBrooklineMA

    February 15, 2009 at 5:57 pm

    I think the census counts all people living in the US, whether or not they are citizens. People don’t have to be here legally to be counted either. Wikipedia agrees:

    http://en.wikipedia.org/wiki/United_States_Census

    I agree completely with you though. Use of modelling will only result in manipulation of the figures for political reasons.

  14. The pitch is something like this: “We’re in an economic crisis, the likes of which hasn’t been seen since [everybody now] the Great Depression! But it’s my oath, duty, and responsibility to uphold the Constitution of the United States, which requires a census every ten years. The census is expensive – VERY expensive.

    “We’ve just exited a benighted era, in which science was repeatedly rebuffed in favor of – I won’t go so far as to say religious beliefs, but one group’s beliefs, in any event. I have to tell you, that’s not the America I love. In the America I love, and the one I know you love too, weuse science – we don’t fear it. Here we have an opportunity to use science to fulfill this solemn Constitutional obligation and to save a boatload of money too. Wouldn’t it be irresponsible to let this opportunity pass us by?

    “Since the upcoming census will be the first one to use hard science rather than only the dogged determination of census-takers in years past, I think it’s important that it be overseen at the very highest levels, and by the very best people this country has to offer. That means my Cabinet. I stand before you, humbled as I am daily by the trust you have placed in me, and swear – again – to uphold the Constitution of the United States, in this instance by ensuring that not only is the census conducted using the best science, that not only is it monitored and managed by the highest levels of government, but that we are continuing our commitment to fiscal responsibility by dramatically reducing what you, and I, and every taxpayer must pay to accomplish this vital task.”

    Along those lines.

  15. Actually Jamie probably said it better and briefer. Probably best to stand clear, out of the way of the nodding nabobs of nascent opinion-persons even now dashing off approval epistles acknowledging the gloriously logical screeds on this topic soon to be emanating from our betters in the royal White Temple.

Leave a Reply

Your email address will not be published.

*

© 2015 William M. Briggs

Theme by Anders NorenUp ↑