The Charleston Gazetteâs Ken Ward Jr. reports that the State of West Virginia has added another worker’s name to the list of 2006 workplace fatality victims.  In the Stateâs coal mining industry alone, 25 workers lost their lives last year.Â
The new addition to West Virginiaâs count is Mr. Jerry Ray May, a delivery driver, who was killed on October 13 when his vehicle was involved in a collision on a haulage road at the Frasure Creek Mine No. 4 in Boone County, WV.  State officials say the fatality occurred “clearly on a bonded, permitted area,” in other words, on mine property. They also told the Gazette reporter that they were aware of Mr. Mayâs death, but forgot to post the information on the agencyâs website. (Perhaps, but in Governor Manchinâs January 10, 2007 State of the State Address he mentioned 24, not 25, mining-related deaths in West Virginia.)
Itâs sad to think we have such sloppy bookkeeping about workersâ lives and deaths. What’s worse than sloppy bookkeeping, however, is the responsible federal agency’s decision NOT to count Mr. Mayâs death in the national total of mining-related fatalities.  MSHA says “the victim was an employee of a nearby restaurant and was driving on the mine access road delivering food.”Â
As Ward points out however, MSHAâs 20-year old policy states:
âif a worker is killed on mine property, the death of that workerâ is counted as mining-related.
The policy defines a worker as:
âemployees of the mine, salesman, delivery people, all construction workers employed in any construction capacity at the mine and others with business at the mine.â
If Mr. May was delivering gravel, or fuel, or new self-contained self-rescuers would MSHA have counted his death? The policy says “delivery people” are workers, and Mr. May was a delivery driver killed on mine property. MSHA’s explanation for not counting this mining-related fatality makes no sense.
In December 2006, after Steve Twedt of the Pittsburgh Post-Gazette exposed inconsistencies in MSHA’s fatality tally, Assistant Secretary Richard Stickler promised a review of agency practices. Now would be a good time for Mr. Stickler to share the results of his review.Â
The issue that you’re addressing is a complex and tangled one where the science of statistics and the science of safety meet. What’s that they say about lies, damn lies and statistics? Being a statistician at heart, I think those people just don’t understand the intricacies in the details… but we are quick to use that to our advantage when the message is one we like, aren’t we…
(Before I begin, I need to state that I am not intimately familiar with MSHAâs business practices or itâs statistical methodology. I am merely applying the general principles that I am aware of in the areas of OSHA and Workersâ Compensation. I am assuming that they apply to MSHA as well.)
MSHAâs decision over what they do and donât consider as a chargeable fatality falls in line with their business practices. Policy decisions and politic overtones can change that definition over the years (and it probably has). The point is, MSHAâs count of what is considered a work-related fatality in the mining industry is not an epidemiological count. Because of itâs ties to policy and procedures, by definition it canât be. A chargeable fatality is one that falls within MSHAâs jurisdiction. If the jurisdiction lines change, so do the counts. So saying that the counts reflect increases (or decreases) in safety is a bit of a misnomer. It could be because of increased safety. Or it could be because policy changed and what was once listed as a chargeable fatality, no longer is.
As a hypothetical example if the administration suddenly decided that the coal mining industry was no longer subject to regulation, MSHA could technically rejoice that the number of mining deaths had suddenly been reduced in half. It may sound absurd, and I hope it never happens, but similar changes have happened through the years. A state reforms itâs Workersâ Compensation system, perhaps changing the definition of what is and isnât compensable. Lo and behold, the number of accepted workersâ compensation claims decreases. Is it because of increased safety? Perhaps. But a much more likely reason for the decline is because the door through which an injury becomes an accepted claim got a lot smaller.
If you are looking to a regulatory (or compensatory) agency for an accurate count of work-related injuries or fatalities, youâre going to get an incomplete count. A manufacturer is covered by OSHA and would be counted in OSHAâs fatality count if he was caught in a machine, but not if he was killed in a freeway accident while picking up supplies. A farm worker is covered by Workersâ Compensation and would be counted in a statesâ Workersâ Comp counts if his tractor turned over, but not if the claim was denied because his alcohol level was .08. A self-employed independent contractor who falls off a roof at a construction site isnât going to be counted by either.
If you want an epidemiological count of fatalities across the nation, the Bureau of Labor Statisticsâ CFOI program is a much better bet. Although BLS is under scrutiny for its supposed undercounting of injuries in its annual survey, the Census of Fatal Occupational Injuries is just that â a census â so it isnât subject to the same problems of underreporting and faulty estimation.
Now, for CFOI, a mining death is just that â a death that occurs in the mining industry. So the delivery driver working for the restaurant wouldnât be considered a mining death even if he died 1,000 feet into a coal mine when the whole show collapsed around him â it would be listed as a death in the services industry, because that is the industry in which he was employed. It also doesnât count deaths that happen to non-workers on job sites. A visiting friend isnât considered in the census unless she is working herself at the time of her injury. But it does count all work-related fatalities in the United States â regardless of workersâ compensation coverage or regulatory jurisdiction. Itâs just about as complete a count as you can possibly get.
In terms of safety though, no count, rate, percentage share, or any other statistic is really going to give you a complete picture. If anybody gives you a number and says âHere. Things are working!â Iâd look twice.