INTRODUCTION
“dissemble” – to give a false or misleading appearance.
The concept of “dissembling statistics” is conveniently illustrated in Chapter Three of Charles Wheelan’s 2012, New York Times best-selling book, “Naked Statistics: Stripping the Dread from the Data”, where the author cites the classic example of the George W. Bush administration’s claim that 92,000,000 Americans would receive an “average” tax cut of “over $1,000”.
The Bush administration had two choices in reporting on its tax-cuts projections. It could report the “average” amount of the projected tax-cuts, which it did, or it could have reported the “median” amount. The “median” amount was the size of the tax cut at the midpoint of the tax-cut data – the point at which there were an equal number of taxpayers who would receive cuts above and below the median amount.
The median amount was $100.
Thus the data actually showed that half of the 92,000,000 Americans – 46,000,000 – would receive tax-cuts of less than a $100 with the other half receiving tax cuts of more than a $100; relatively speaking, hardly anyone would receive a $1,000 cut.
The “average” tax cut was $1,000 because, Wheelan tells us, the calculation of the average included the very large tax-cuts that a large number of extremely wealthy Americans would enjoy under the tax-cutting program.
Thus, relative to the question of what tax-cuts middle America could expect, the average figure was effectively a dissembling statistic; not untrue, just conveniently misleading – appearing to tell Americans something important about a fact they wanted or needed to know, but instead disguising the relevant reality of that fact.
DISSEMBLING STATISTICS IN WORKERS’ COMPENSATION SYSTEMS
Workers’ compensation systems are especially fertile ground for dissembling statistics. This is because the number of claims dealt with each year is very large and the percentage of those claims in which the claims administration presents any issue of systemic importance is very low – historically about 5%.
Currently, the number of claims registered in Ontario’s workers’ compensation system each year is in the order of 200,000, and, of those, about 190,000 will be straightforward – claims that will usually present relatively little cost exposure for the Board and in which the Board has no reason to question or concern itself with the facts (medical otherwise). One might usefully call these claims the “thin” claims, reflective of the size of the file that will accumulate during their administration.
The remaining 10,000 are the “thick” claims – the claims with serious implications for the injured worker making the claim and in which there are issues of fact or medicine that are innately complicated and potentially contentious and where the Board’s cost exposure is high and the file thus correspondingly thick. The thick claims are a tiny percentage of the whole but a large number nonetheless – as I said, approximately 10,000 per year in Ontario. It is only in these thick claims that a Board’s performance is in fact put to the test.
A conveniently clear illustration of the potential in these circumstances for the Ontario Board’s statistics to be dissembling may be found in its published claim (2016 Second Quarter Shareholders’ Report) that its surveys of the level of satisfaction among injured-worker claimants show that 70% of injured workers who “have dealings with the Board” are “satisfied” with their overall experience at the Board.
Considering that the cohort of injured workers surveyed as to their level of satisfaction includes the 95% of workers with thin claims – claims which the Board will for the most part have had no reason to challenge or resist in any way – the 70% satisfaction rate is not a very helpful indicator (and neither, taking that context into account, is it a very complimentary one).
What one wants know about the injured workers’ level of satisfaction – what the WSIB should want to know – is how the workers with thick claims regard their experience with the Board; or more pertinently, how a random sample of the professional advocates active in the thick claims regard the Board’s treatment of their clients’ and their claims.
During his appearance on TVO’s Agenda on December 14, 2016, the Ontario WSIB’s President and CEO, Tom Teahan, deflected the criticism from the injured worker side of the table by citing the Board’s 92% success rate. The Ontario Board typically attributes its cost-cutting miracle over the past five years in large measure to its success at getting a large percentage of workers with lost-time injuries back to work within 12 months with no wage loss; the most recent published success rate is 92%.
As a shield against claims of the Board being unfair or acting illegally in its treatment of injured worker claimants, that 92% figure is presumptively impressive. If an organization of this size dealing with issues of this complexity can get it right 92% of the time, how, the average citizen is likely to ask, can there be any validity in the complaints.
But that 92% figure, while, no doubt, as Mr. Teahan insisted a “true” figure, is nevertheless not a figure that tells us anything important about the Board’s performance; it is innately a dissembling statistic.
The cohort on which that 92% figure is based is “all lost-time injuries”. And it is in the nature of the workers’ compensation business that a very large percentage of that cohort will be “thin” claims – claims in which the worker will have pulled a muscle, broken a finger or wrist or sprained an ankle or suffered a concussion and will be off work for a few days or a few weeks and will then return to work at full wages in the ordinary course – with no extra-curricular intervention from the Board required.
What one really wants to know is what is the return-to-work-with-no-loss-of-pay experience in the thick claims – perhaps in the cohort of claims in which there is a permanent impairment or a concern that there might be one, or, perhaps, in the smaller cohort of claims arising solely from back injuries.
Also missing from the return-to-work figures generally is any follow-up data, data on what proportion of workers with thick claims who returned to work within 12 months with no loss of pay had remained at work, still at no loss of pay for, say, the 12 months following their return to work.
For a Board with a large statistics department it is surprisingly difficult to find published WSIB statistics that are not dissembling; these would be statistics that factored out the influence of the overwhelmingly large number of thin claims where the Board’s administration is not seriously challenged; statistics, in short, that would tell us something useful about the Ontario Board’s actual performance when it is dealing with the thick claims – the claims that count.
RE