Friday, February 18, 2011

Grading PolitiFact: John Boehner and adding 200k federal jobs

To assess the truth for a numbers claim, the biggest factor is the underlying message. 
--Bill Adair
The issue:




The fact checkers

Louis Jacobson:  writer, researcher
Martha Hamilton:  editor


Analysis:

The graphic above states the issue sufficiently well.

Here's a video of the press conference, with the relevant statement from Boehner occurring near the end:




I'm unable to make out the reporter's question to Boehner just prior to his comment, and so far unable to locate a transcript of the exchange.  As a result, the context as provided by PolitiFact is about all we have to work with.

To calculate the number of new federal jobs, PolitiFact first used two different metrics via the Bureau of Labor Statistics:
The first is the overall rise in federal employees between January 2009 and January 2011. The net increase was 58,000.

The second is the number of federal employees without counting U.S. Postal Service workers. Over that same two-year period, the increase was 140,800.
Should postal workers count?  The U.S. Postal Service is set up as a quasi-independent entity.  The argument might be made either way, but the keepers of stats clearly see some value in keeping track of federal employment other than postal workers.  The 140,800 figure may be valuable, then.  PolitiFact points out that 140,800 is less than 200,000 and then offers a different count:
We also checked with John M. Palguta, vice president for policy with the Partnership for Public Service, a non-profit that promotes government service, and he confirmed our general conclusion using numbers from a different database.

He dug into the U.S. Office of Personnel Management’s on-line federal workforce data source, "FedScope." He found that in fiscal years 2009 and 2010, respectively, the federal government filled a net 59,995 and 47,062 new permanent, full-time, non-seasonal, non-postal jobs. Combined, that means that federal employment rose by 107,057 jobs -- well short of 200,000.
An organization that extols the value of public service may not be the ideal place to go for unvarnished analysis.  Farming out a significant portion of the fact check to such a figure seems curious.  Why, for example, does Palguta exclude non-permanent, part-time and seasonal jobs?  If the government shifted entirely to a part-time workforce would we have zero federal government employment in Palguto's eyes?  Palguta's totals are suspect unless we receive some sort of explanation.


PolitiFact:
We checked with Boehner’s office to see what his statement was based on. Aides said they had used figures from December 2008 to January 2011, which produced an increase of 153,000 federal, non-postal jobs. Then they factored in, on a discounted basis, the temporary jobs required to carry out the 2010 Census. According to the Census Bureau, such temporary employment peaked at 585,729 in early May 2010.
The base total from Boehner's office (153,000) differs from that reached by PolitiFact, apparently because the start date varies by a month.  Is it fair to count the difference between December and January during the transition from an election year?  It seems plausible, since the old administration may be less likely to make new hires than the new administration.

Also note that Boehner's 200,000 estimate included temporary Census jobs, a practice PolitiFact questioned in the past, though with a different set of circumstances.  The Census jobs do not really belong in a complaint about federal government expansion except in the context of the past practices of the Census.

Discounting the Census jobs and the estimate from Palguta, we have a range from 140,800 to 153,000 new federal jobs compared to Boehner's claimed 200,000.

PolitiFact finds Boehner false:
All told, we find that Boehner’s 200,000 number is way off. We rate it False.
That was the last line of the fact check.

Some of you from the Bill Adair school of fact checking may be wondering "Um, where's the emphasis on the underlying point?  The most important aspect of a numbers check is the underlying point."

I could detect no evidence that PolitiFact considered a point underlying Boehner's 200,000 figure.  It seems possible to hypothesize, however, that Boehner was making the underlying point that federal employment had increased considerably over the past two years under President Obama.  An increase of 140,800 is short of 200,000 but may be sufficient to make that underlying point at least somewhat true.

Further, if the increase was as high as 153,000 then Boehner could legitimately round up to 200,000 (the nearest 100,000).

It's true that we don't ordinarily round to the nearest 100,000 in normal communications.  The nearest 10,000 or nearest 1,000 are more commonly used.  For the sake of argument, let's take the figure 195,000 (Boehner could reasonably round up from that figure) and compare the Boehner rating to an early PolitiFact fact check of Ron Paul.

Paul was rated "Mostly True" for an estimate of American war casualties ("over 5000 Americans") that was inflated by about 15 percent (using 5001 as the minimum baseline).  Boehner's estimate was inflated by 38.5 percent using the minimum baseline for rounding to the nearest 10,000 (195,000).  Using the figure Boehner's office used the inflation percentage changes to just over 27 percent.

Could a 12 percentage point difference turn a rating from "Mostly True" to "False"?  Or does it need to be about 25 percentage points off?

Or is the key to Boehner's rating an underlying point that PolitiFact neglected to discuss?


The grades:

Louis Jacobson:  F
Martha Hamilton:  F

The fact check leaves too many unanswered questions, including one about the proper role of the underlying point in a fact check about numbers.  If the fact check is fact and not opinion then PolitiFact should employ consistent methods.  Justifying a rating by the underlying point in one case and ignoring any underlying point in other cases critically undermines any appearance of objectivity.


March 29, 2011:  Added some clarifying language in the paragraph explaining methods of rounding.

No comments:

Post a Comment

Please remain on topic and keep coarse language to an absolute minimum. Comments in a language other than English will be assumed off topic.