Monday, April 02, 2012

Grading PolitiFact: Reince Priebus and the CBO

HALF TRUE – The statement is partially accurate but leaves out important details or takes things out of context.
--Principles of PolitiFact and the Truth-O-Meter
The above is not to suggest that PolitiFact has eliminated its inconsistency with respect to defining the "Half True" rating.  But that's another story.

The issue:

(clipped from

The fact checkers:

Louis Jacobson: writer, researcher
Bill Adair: editor


I repeatedly try to give PolitiFact the benefit of the doubt as trying to do its job fairly and objectively.  Cases like this one featuring Reince Priebus make granting the benefit of the doubt very difficult.  Cases like this one look like simple partisan hackwork instead of journalism.  Comparing this fact check with one of President Obama helps highlight the mockery of journalistic objectivity, especially since writer/researcher Louis Jacobson and editor Bill Adair contributed to both stories.

PolitiFact introduces us to the statement by RNC chairman Reince Priebus (bold emphasis added):
Priebus pointed to what he saw as a range of problems with the law, but we zeroed in on one claim. "While millions will be added to the government rolls, millions more will also lose their current health-care coverage," Priebus wrote, later explaining, "According to the (Congressional Budget Office), as many as 20 million Americans could lose their employer-based insurance thanks to Obamacare."

Is the health care law really going to leave 20 million American workers without health insurance from their employers?
Note that the second paragraph asks a question that alters Priebus' claim.  The Obama story does the exact same thing, so both stories contain the same type of error on this point.  Both men said "as much as" and PolitiFact dropped the phrase in both cases while identifying the target claim of the fact check.

When we asked the RNC to back up its number, a spokeswoman pointed us to a March 2012 study by the Congressional Budget Office. We often cite CBO’s research because we consider their work to be independent, nonpartisan and credible. We wondered whether Priebus had used CBO’s figure appropriately, and with sufficient context.
It's fair to judge Priebus on whether Priebus used the CBO figure accurately and appropriately along with sufficient context.  If PolitiFact uses similar standards for figures like the president then so much the better.

Ranging for ranges

The CBO study in question was undertaken to estimate the impact of the health care law on the number of people obtaining health care coverage from their employer. It took up this task with care, noting that surveys of what companies may do in the future about health care "have no consequences for the responders, do not require careful analysis or extensive deliberations, and are necessarily based on limited information about the various ways that the (law) will affect the market for health insurance."

Still, after taking into account the various cross-cutting factors, CBO came up with a "baseline" estimate -- essentially, its best guess. CBO, along with the similarly nonpartisan Joint Committee on Taxation, settled on a range of 3 million to 5 million fewer people, on net, obtaining coverage through their employer each year from 2019 through 2022 than would have been the case before the law was passed.
At first blush, a range of 3 million to 5 million each year from 2019 through 2022 creates a total decrease of 12 million to 20 million persons covered by employer insurance by 2022 under the baseline scenario.  But that apparently isn't what the CBO numbers mean.  Rather, the 3 million to 5 million figure is a running net total of persons losing employer-provided insurance.  Accordingly, the 5 million figure is not an upper boundary on the estimate as the report later makes clear (bold emphasis added):
In the judgment of CBO and JCT, their estimates of the effects of the insurance coverage provisions of the ACA on sources of coverage and the federal budget are in the middle of the distribution of possible outcomes. However, assessing the effects of broad changes in the nation’s health insurance system requires assumptions and projections about a wide array of technical, behavioral, and economic factors. As a result, any projections of those effects are clearly quite uncertain. To illustrate that uncertainty, CBO and JCT have estimated the sources of insurance coverage and federal budgetary outcomes that would result from the ACA under certain alternative assumptions.
While 3 million to 5 million people is nothing to sneeze at, it’s also quite a bit lower than the 20 million figure Priebus cited. So where did the 20 million number come from?
PolitiFact apparently takes the "3 million to 5 million" figure as a range where 5 million represents an upper boundary on the loss of employer insurance, otherwise comparing 5 million to Priebus' 20 million represents an apples-to-oranges comparison.  Priebus wasn't offering a mid-range figure.  He clearly said "as many as 20 million."  His language denotes an upper boundary rather than a mid-range figure.

So where did the 20 million figure come from?

CBO supplemented its "baseline" estimate with four alternative scenarios, tweaking its model in a variety of ways to account for the possibility that companies or individuals will behave differently than the baseline estimate assumed.

The four alternative scenarios produced a wide range of outcomes. One actually resulted in a net gain of 3 million people with employer-sponsored insurance. The other scenarios resulted in a decline of 10 million, a decline of 12 million, and -- here it is -- a decline of 20 million.
PolitiFact's description is reasonably accurate, although finding the 20 million figure in the report is easier than PolitiFact portrays it.  The report mentions the outcome ranges before delving into the details of the four alternative scenarios (bold emphasis added):
In the four alternative scenarios discussed below, the ACA changes the number of people who will obtain health insurance coverage through their employer in 2019 by an amount that ranges from a reduction of 20 million to a gain of 3 million relative to what would have occurred otherwise.
Contrary to PolitiFact's portrayal, the 20 million figure is the first figure one encounters in the report's description of the alternative scenarios, not the last.  The report describes its baseline estimate as a mid-range figure and offers a gain of 3 million and a loss of 20 million as the range.

Contextual cherry-picking

So Priebus is right that CBO estimated that "as many as 20 million Americans could lose their employer-based insurance thanks to Obamacare." But this is the most extreme outcome of five possible outcomes presented, and it’s not the primary estimate, which is about 5 million, or one-quarter the number Priebus offered.

So Priebus has engaged in cherry-picking. He also ignores some important context.
Did Priebus engage in cherry picking?

Priebus said the CBO "estimated as many as 20 million" might lose employer-based health insurance.  His statement clearly refers to upper boundaries on the CBO estimate, so in what respect is the 20 million figure "cherry-picking"?  None of the five scenarios the CBO scored was accompanied by a low-end or high-end figure.  The figures were mid-range for each scenario, and for the third alternative scenario it follows that the CBO countenanced the possibility that the net loss of employer-based insurance would exceed  the 20 million estimate, though by how much the report does not say.

The cherry picking angle comes as a tough sell given the properly-understood context.  What about the supposed critical context that Priebus allegedly ignores, however?

It’s important to note that not everyone who "loses" their employer coverage will do so involuntarily. According to CBO’s "baseline" estimates, 3 million people will spurn their employer’s offer of insurance and turn instead to another source, such as the health insurance "exchanges" created under the Obama health care law -- virtual marketplaces for each state where applicants will be able to compare and purchase plans that offer at least a minimum bundle of benefits. In many cases, they will do this because they consider the employer’s offering to be unaffordable or lacking too many features they need. So those people aren't being forced off the employers' plans, they are choosing to switch.
PolitiFact is furiously spinning the context of the report.  Those 3 million who supposedly "spurn" their employer's offering of coverage are technically ineligible for subsidized coverage under the state exchanges.  That fact is explicit in the CBO report (bold emphasis added):
Another 3 million people who would have had employment-based insurance under prior law and will still have an offer of such coverage under the ACA will instead choose to obtain coverage from another source. Under the legislation, workers with an offer of employment-based coverage will generally be ineligible for exchange subsidies, but that “firewall” will presumably be enforced imperfectly, and an explicit exception to it will be made for workers whose offer of employment-based coverage is deemed unaffordable.
It's not clear from the report where PolitiFact bases its claim that workers will spurn employer coverage because that coverage lacks desired features.  The CBO report doesn't mention such reasons (contrary to the impression PolitiFact foments).  It simply mentions that an unaffordable insurance offer from an employer is expected to permit employees to obtain subsidized coverage from a state exchange.

The CBO report goes on to explain that employees are expected to seek out the least expensive insurance option.

It's worth mentioning that employees looking for the least expensive option do so with the threat of the individual mandate penalty hanging over their heads.  The employees may choose to switch, but the decision is hardly free from coercion by the health care law as well as from the price of employer-based insurance.  Some might say the employee is caught between a rock and a hard place.

The underlying argument?

Now the tough question:  How does all of this relate to the rating of Priebus' claim?  None of it addresses the accuracy of the 20 million figure.  That figure is a reasonable approximation given the context of the report.  The "critical context" must address some underlying argument from Priebus that PolitiFact has yet to identify.

But there's a problem with PolitiFact's approach.  Priebus' remarks suggest he cites the 20 million figure to support the argument that the reform decreases the value consumers receive for their insurance dollar.

Priebus (bold emphasis added):
While we pay more for health care, we will get less in return. According to the CBO, as many as 20 million Americans could lose their employer-based insurance thanks to Obamacare. Six million seniors will lose their prescription-drug coverage under new Medicare rules. Obamacare also slaps restrictions on families that use flexible health-savings accounts. These accounts save money, but will be cut in half, increasing costs for 20 million Americans.
PolitiFact's gripes do not address Priebus' argument.  Rather, they try to fault Priebus for ignoring the number of people who obtain insurance through the law.  But again there's a problem for PolitiFact's approach:  Priebus acknowledges gains in the number of insured persons, albeit in the form of subsidized government-run insurance exchanges. 

Priebus (bold emphasis added):
Health-care costs are up and will continue to skyrocket for the foreseeable future. While millions will be added to the government rolls, millions more will also lose their current health-care coverage.
Moreover, in the broader context Priebus' op-ed was published as part of a debate by the Orlando Sentinel, with HHS secretary Kathleen Sebelius contributing an op-ed defending the ACA.  Is it fair to expect Priebus' to use his space taking note of the things Sebelius was likely to highlight?

PolitiFact doesn't have Priebus' excuse for neglecting context. Where's the mention of the debate? Where's the fact check of Sebelius?  It's easy to forgive Sebelius for leaving out mention of ObamaCare's drawbacks when there's no fact check of her debate claims, as it happens.

At the end of the day, PolitiFact's claims that Priebus "cherry-picked" his figure and that he omitted critical context both rest on very shaky ground.

The phantom implication

There's one more complaint from PolitiFact before we move to the conclusion:
Contrary to Priebus's [sic] implication, the law is expected to result in a huge gain in the number of people getting coverage.
PolitiFact asserts the existence of the supposed implication instead of identifying it in the text of Priebus' remarks.  It's doubtful any but a partisan from the left would notice such a thing.  Important points in debate get emphasized so the audience doesn't miss them.  Merely implied points tend to get ignored by the audience.

PolitiFact offers no argument why Priebus' alleged implication is strong enough to serve as a significant factor in rating the truth of his statement.

Wrapping up the Obama comparison

President Obama claimed "as many as a quarter" of American students do not finish high school, which PolitiFact translated as a dropout rate as high as 25 percent.  PolitiFact found no evidence that the rate was actually that high and elected not to dock Obama under its "burden of proof" grading criterion, instead choosing to give Obama "credit for offering a range."  Obama used a dubious figure that PolitiFact found suspect and ended with a "Mostly True" rating.

The Priebus rating is different in that he was reporting a figure from a report rather than providing an upper bound for a real-world statistic.  PolitiFact confirmed the number from the report and gave Priebus a "Half-True" rating based on a series of poorly reasoned complaints.

In Obama's case, PolitiFact made no attempt to identify an underlying point.  In context, Obama was making the case for education beyond high school.  His implication was that a high percentage of students wouldn't qualify for higher education because they haven't even finished high school.  But that doesn't appear to follow.  The relevant issue was the value of the education received.  A person with no high school diploma might well be better equipped for higher education than one possessing a high school diploma.  PolitiFact did Obama a favor by overlooking the underlying argument.


Our ruling

Priebus’ number does appear in the CBO report he references in the op-ed, but it’s not the primary estimate -- it’s one of four alternative scenarios, and easily the one with the biggest decrease. So the number is cherry-picked.

His claim also suggests that 20 million people are being forced out of the coverage, when in fact many will choose voluntarily to switch to better coverage. And he conveniently ignores the estimates that 9 million people who didn't have employer coverage will get it because of the law, at least according to CBO’s estimates.

His claim is partially accurate but leaves out important details and takes things out of context, which is our definition of Half True.
1)  No other number in the CBO/JCT report can qualify as an upward boundary estimate of the loss of employer-based insurance.  With no other number from which to choose and the fact that the third scenario implicitly grants that more than 20 million could lose employer-based insurance, this isn't a good example of cherry-picking.

2)  PolitiFact recycles its spin on the "many" who will choose not to accept coverage from their employer.  The report says employees will take that option when it is less expensive.  Therefore, the "choice" to change is a form of price rationing employer coverage.  PolitiFact obscures the facts.

3)  Priebus achieved the type of accuracy that PolitiFact often rates "True," and PolitiFact's caveats carry the thinnest of justifications.  Indeed, PolitiFact easily outdoes Priebus when it comes to leaving out important details and taking things out of context.

The grades:

Louis Jacobson:  F
Bill Adair:  F

Jacobson and Adair receive flunking grades for a number of reasons, including the failure to explain that the CBO/JCT report figures did not represent high-end ranges but rather mid-range estimates for each scenario.  The team likewise ignored Priebus' inherent justification for using the highest number from the third alternative scenario as the upper boundary prediction.  Nor did PolitiFact provide the context of Priebus' column as a counterpart to a column from an Obama administration official.  These hardly exhaust the list of failures, but they justify the flunking grades.

Correction 4/2/2012:  Hat tip to Jeff Dyberg for pointing out a nearly redundant "of the eduction."  Removed the needless and misspelled phrase.  Oops. Also duplicated PolitiFact's error of adding an extra "s" to the possessive "Priebus'" in the concluding paragraph.  Eliminated the extra "s" accordingly.

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.