Data has always been putty to be molded by propagandists into informational power. Through the inclusion of irrelevant data or the exclusion of relevant data, these students of Stalin use the data to fabricate a “reality.” They rely on the obfuscatory possibilities afforded by rows and rows of numbers, complex charts, vast tables and busy figures to create an impenetrable mangrove of data. They count on citizens not caring quite enough (or not being sophisticated enough) to dig into the details. It’s easier for many to accept the conclusion if they’re not able to penetrate the analysis. This is exactly what keeps the editorial page of the Wall Street Journal going.
For the record, I do not object to a data exploration exercise which seeks to illuminate the jobs situation in Wisconsin and the nation. But I do object to the use of data and analysis as agitprop.
We should all heed the sage advice of President Ronald Reagan. Доверяй, но проверяй; “doveryai, no proveryai,” Trust, but verify.
One of the challenges in any political system is to detach the data and analysis from the political process as much as possible. Dispassionate technocrats are what you need to crunch numbers and give you the information you may need to steer the ship of state whether it’s a town, a state or the nation. My father was one of these technocrats. He was the deputy director of the United States Census. He was a consumate statistician who reveled in gigantic datasets.
He told me that once the technocrats become political, it’s game over. The data no longer serve a purpose to help guide policy, but become a tool to enforce ideology. Welcome to Wisconsin.
Remember Colin Powell sitting at the UN with a vial of “yellowcake” uranium? That’s a classic example of the politicization of data. The Bush administration was so thoroughly convinced that Saddam Hussain was working on building a nuclear bomb that they created phony data in the form of a vial of “yellowcake” to prove it.
We know now that the whole episode was a hoax, a conjuring trick to goad America into a war-of-choice. And we know what that cost us. What it cost our soliders and what it cost the Iraqi people.
Now, in the politically charged atmosphere of Scott Walker’s Wisconsin we are being asked, nay implored to believe that the numbers produced by the very technocratic and non-political Bureau of Labor Statistics are somehow deficient when they demonstrate that Wisconsin trails the nation in job creation (50th out of 50). We are being asked by Walker’s Department of Revenue (not the technocratic Wisconsin Department of Workforce Development who report data to the BLS) to believe that another dataset is somehow more representative of the “true” jobs situation in Wisconsin. Color me skeptical. But let’s have a look.
First, I’m not going to talk about the numbers. The numbers here are not relevant. What is relevant are the datasets, how they’re used and what’s an appropriate analytical approach. There are others better positioned than I to dig into the fact / fiction question of the numeric details. I chose to question the entire Walker enterprise here.
First, let’s talk about the data. The two datasets at issue here are the Establishment Survey and the Household Survey. The Establishment Survey is the survey used by the BLS to calculate employment levels in the economy and it’s the one that Walker’s DoR is claiming is somehow deficient. The Household Survey, a survey conducted by the Census Bureau for the BLS, is used to determine unemployment levels and types of unemployment (the “U” unemployment series).
What are the principal differences in the datasets and why does the BLS prefer one over the other when reporting on state job creation? First, you have to understand that the BLS relies on the Establishment Survey to determine levels of employment at the state and local level. This survey polls employers about their hiring, firing, etc. The Household Survey is just that, a survey of individuals at the household level.
So why use the Establishment Survey and not the Household Survey for calculating levels of employment within a state? Simple. The Establishment Survey relies on businesses located within specific states to determine employment levels. Household Survey data does not. It relies on households and the employment of those 15 years of age or older. It makes no assumptions as to where those people are employed.
So, for instance, if there is a factory in Waukegan, IL and half the workforce lives in Racine, they would be counted as employed Wisconsinites in the Household Survey but not as employees in a Wisconsin business as part of the Establishment Survey. Should Governor Walker (or any Governor for that matter) be allowed to take credit for Illinois adding jobs but hiring Wisconsinites?
It is due to these boundary effects that the Establishment Survey and not the Household Survey is used to calculated state job numbers. The ability of a state government to positively (or in the case of Governor Walker, negatively) effect the creation (or destruction) of jobs at the state and local level is only measurable with the Establishment Survey data.
According to Laura Dresser, a labor economist at The Center On Wisconsin Strategy at The University Of Wisconsin, Walker’s new numbers are little more than an incredibly transparent effort to create a false reality just in time to mislead Wisconsin voters who will cast their ballot in a few short weeks. Pointing out just one of the flaws in the ‘new and improved’ Walker method of measuring job growth, Dresser says, “It seems that they’re attributing employment growth in other states to Wisconsin.”
The second reason why we don’t use Household Data is that no other state uses household data to calculate state job numbers. One of the important features of this massive data collection exercise is the ability to create apples-to-apples comparisons between states. I can look at comparable data between Wisconsin and Illinois and see how much better Illinois is doing at the creation of private sector jobs.
This facilitates the exchange of policy ideas between states with similar populations and employment opportunities. Policy makers can see the impact of their policies on jobs. Of course if you’re Governor Walker, you don’t want to see those comparisons. They make him look like a complete failure compared to the rest of the nation.
So if the question you want answered is: “How many people regardless of state-of-origin are employed by Wisconsin businesses?” or “How many jobs were created by Wisconsin businesses?” you want to use the Establishment Survey data. The Household data is inappropriate.
But if the question you want answered is: “How many Wisconsin residents are employed regardless of where, geographically, they are employed?” then the Household Survey is the data you want. The Establishment Data is inappropriate.
See how that works? The data you use depends on the question you ask. Not all data can answer all questions.
As a nation, we can have a dialog about the quality of the Establishment Data versus the Household Data across the states. We can create a framework that allows employment analysis based on the Household Data across all the states so that comparable numbers can be created.
Walker’s efforts to inflate his jobs numbers are just so much agitprop. Enjoy the show!