Delay may be the result of bad assumptions

Discussion in 'Wireless Estimator Site Discussions' started by Michael S Landa, Feb 24, 2012.

  1. Michael S Landa

    Michael S Landa Industry Observer

    It is more likely the programs airing was delayed due to faulty data based on bad assumptions. The Propublica Frontline report may have used a Bureau of Labor Statistics formula for its calculations, however, they failed to use the same data or same data sources that the Bureau of Labor Statistics would have used. There are two manuals which determine how statistics are developed and calculated by the Bureau; the BLS Handbook of Methods and the Occupational Injuries and Illnesses Classification Manual, both available online.

    Failure to comply with the processes in either will result in inaccurate results. Moreover it would be difficult to make a logical or reasonable argument for comparing BLS data developed using BLS methodology with data not using BLS methodology.

    Lastly, not using all fatalities, not just one (falls) of the six events or exposure tracked by the BLS, in the numerator will skew the result. This skewing misrepresents the results as accurate when they are not.

    U.S. Department of Labor, Bureau of Labor Statistics (1997, April). BLS Handbook of Methods. United States Government Printing Office, Washington, DC.
    U.S. Department of Labor, Bureau of Labor Statistics (2007, September). Occupational Injuries and Illnesses Classification Manual. United States Government Printing Office, Washington, DC.
  2. Michael S Landa

    Michael S Landa Industry Observer

    The following comments were received from the BLS about the methods used by both the Propublica Frontline report and’s determination of “Most dangerous” concerning the calculating, comparing and reporting of fatal falls (CFOI) using BLS formulas and data.

    Dr. Landa,

    Thank you for contacting the Bureau of Labor Statistics Injuries, Illnesses, and Fatalities (IIF) program. Our program tabulates annual counts and rates of nonfatal workplace injuries and illnesses in the private, state, and local government sectors, case and worker characteristics for cases that involved days away from work, and annual counts and rates of fatal workplace injuries. Much of our data can be located online at

    The Census of Fatal Occupational Injuries (CFOI) uses many different types of Source Documents to identify and verify fatal occupational injuries. CFOI uses Source Documents like death certificates, workers’ compensation reports, reports from the Occupational Safety and Health Administration (OSHA), and media reports to produce the most accurate data on fatal workplace injuries.

    Currently, the IIF program uses the Standard Occupational Classification (SOC) system to define occupation and the North American Industry Classification System (NAICS) to define industry. Neither classification system has a category dedicated solely to tower climbers. As such, the IIF program does not publish any data on tower climbers since there is not a corresponding category in SOC or NAICS. For this case and other instances where our classification systems are insufficiently detailed, researchers will sometimes use BLS data to craft their own categories using existing BLS data in their research. Any products of research like should be cited as “Based on individual research using data from the Bureau of Labor Statistics, Census of Fatal Occupational Injuries (CFOI)” and are not considered official products of BLS.

    It is the policy of the IIF program not to review or comment on outside research. Documentation of CFOI rates can be found in the BLS Handbook of Methods: Please note that CFOI fatal injury rates are only calculated for those occupations and industries which meet minimum thresholds of fatal injuries and employment. Published rates can be found on our web page:

    Due to differences in methodology, we would strongly discourage you from comparing any rates that are calculated in a way that differs from the CFOI methodology against any rates calculated by the CFOI program. In addition, I am attaching a document entitled elephant.pdf which provides information on the problems involved in calculating rates for industries or occupations with few workers.

    I hope this information is helpful. Please contact us again at (202) 691-6170 or if you have any additional questions on workplace injury, illnesses, or fatality statistics.

    Joyce Northwood
    Office of Safety and Health Statistics

    It should now be clear both publications promulgated faulty conclusions based on bad assumptions.

    Michael S. Landa

    Attached Files:

  3. David Lehrer

    David Lehrer Industry Observer

    The only thing I could find on the Frontline/Propublica report was in the Wireless Estimator story that said their promotional ad said that "Independent contractors who are building and servicing America’s cellular infrastructure are 10 times more likely than an average construction worker to die on the job."

    I guess you're convinced that your "more likely" is more qualified then their "more likely" when you state that "It is more likely the programs airing was delayed due to faulty data based on bad assumptions."

    Plus you have absolutely no idea as to what methods were used by Propublica and Frontline reporters as you state above since the investigation isn't out yet.

    You've taken a one sentence promo and turned it into a thesis.

    It seems like everybody is hung up on arguing over whether it's "the most dangerous job".

    Let's say it can be documented that it's not the most dangerous job in the nation year after year and workers are not 10 times more likely to die.

    But we're still left with the undeniable fact that it is "one of" the most dangerous jobs in America and too many people are dying.

    That's the inconvenient truth that statistics can't be manipulated to disguise...and something needs to be done about it.
  4. Michael S Landa

    Michael S Landa Industry Observer

    I spoke with Liz Day, chief researcher for ProPublica, on Thursday, February 16, 2012 for more than thirty minutes; in that time Ms. Day outlined the entire research methodology. I challenged the assumption she and both made concerning how to properly apply the fatality rate formula. Since falls are one of a handful of events it was improper to use falls as the numerator; the total number of fatalities is what BLS used therefore you can’t compare ProPublica and figures with those produced by the BLS. Secondly, the denominator used by both publications was 10,000 when, in fact, there is no population outside of 237130, which is the official NAICS code used by the BLS for tower construction. Because of these facts, statements such as "Independent contractors who are building and servicing America’s cellular infrastructure are 10 times more likely than an average construction worker to die on the job" are misleading.

    David is absolutely correct in his assessment of the undeniable fact that to many workers die on the job, my objection is that all this noise surrounding journalist reports obstructs the formulation of a coherent plan to reduce the number of falls from elevation.

    Two years ago I tried to put together a working group of six climbers and six owner/operators to analyze the problem and all I heard were crickets and a couple of smart remarks for climbers who had absolutely no concern with the issue at hand.

    Research I started two years ago indicates that of 76 fatalities only 2 could reasonably be considered accidents. Fifty-one were ‘failure to tie-off’, that is two thirds of all fatalities in the initial group could have been prevented simply by following OSHA regulations. I am now looking at over 250 tower fatalities since 1984; from that list I will determine which actually involved communications towers the request the complete OSHA investigation files. Similar statistical analysis will be done on the complete files and at some point in time I will publish the results in a peer reviewed journal. The accidents studied had to have sufficient information in the summary to determine whether or not personal fall perfection equipment was worn. Those other cases included riding the line, riding the load, and the use of a capstan to hoist personnel also had to have sufficient information in the summary to make a determination. The idea behind any scholarly research is to add to the body of knowledge and not to sensationalize.

    In spite of these preliminary findings, I am unwilling to universally condemn climbers as exhibiting risky behaving in all circumstances and under all conditions. There is another factor which likely impacts individual perception and therefore ones perception of danger both in terms of degree of danger and immediacy of danger.
  5. Wireless Estimator

    Wireless Estimator Administrator Staff Member

    Since there is no NACIS or SIC code for Pachyderm pedagogy to identify that there are "600 known elephant trainers in the US," I presume that we'll just have to take Guy Toscano's word for it in your document link since he's from the government and he's here to help us.

    The SIC code system is not accurate; the NACIS is worse. Our industry is dumped into some Dumbo-written NACIS code of 237130. It also includes nuclear power plant, solar, and wind power construction, amongst other disciplines that have nothing to do with wireless construction.

    Since there is no central government agency with the role of assigning, monitoring, or approving NAICS codes for wireless contractors, companies oftentimes use the code that will give them the lowest insurance premium for their NCCI code for workers compensation rates.

    NAICS codes contained on a form or citation received from OSHA are sometimes incorrect, but can be changed by contacting the federal agency.

    Even if our industry had over 100,000 workers and its own unique code, tower construction and maintenance wouldn't be on the BLS's radar because many different codes are used.

    Only 18 of 83 fatalities of falling from a communications structure reviewed by Wireless Estimator were businesses in the NACIS code of 237130. (See Below)

    Electrical Contractors - 238210 - is a common NACIS code found in 19% of 83 tower-fall-related fatalities.

    Two separate fatalities of workers falling were from companies using the NACIS code 238230 - Painting and Wall Covering Contractors.

    This one is truly a stretch. Two men died last April when a gin pole failed. According to OSHA, the company's SIC code is: Commercial Physical and Biological Research. The NACIS code is even less plausible: Research and Development in the Physical, Engineering, and Life Sciences.

    When he was OSHA's chief in 2008, Edwin G. Foulke, Jr. unabashedly talked about the high fatality rate of tower technicians.

    He neither used the NACIS's 237130 industry code nor waited for the Bureau of Labor Statistic's annual August unveiling of the occupations with high fatal work injury rates.

    Instead, Assistant Secretary of Labor Foulke, working with his directors and industry businesses, was able to use his own methodology and drill down into the wireless construction sector and expose the unacceptably high death rate in 2006.

    Fatality rates have seen a decline over the years, partly through Foulke's NATE partnership program and his desire to have the industry acknowledge its high fatality rate and address it.

    In this case, parading the 800-pound gorilla in the open was a good thing.

  6. Michael S Landa

    Michael S Landa Industry Observer

    I tend to favor government statistics for their uniform and consistent application of generally accepted statistical methodology; in this particular case, because the Wirelessestimator calculations are being compared to BLS statistics. With regard to the issues concerning NAICS and SIC codes, we have to work with what we have, so long as Wirelessestimator produces statistics that are compared to BLS statistics then BLS methodology should apply to both sides of the equation. It is important to remember all deaths are recorded somewhere, it is therefore logical to assume taking fatality data out of somewhere and putting into another place NOT used by the BLS will result in faulty statistics.
    NAICS and SIC codes are not assigned by anyone, they are selected by establishments; Wirelessestimator would likely select SIC code 2711 Newspapers: Publishing, or Publishing and Printing (NAICS code 516110 Internet Publishing and Broadcasting). Establishments may have multiple codes depending on what areas, industries and customers they serve.

    The assertion that the former Assistant Secretary of Labor developed his own methodology to expose the high death rate in 2006 is flawed logic. Mr. Foulke’s public remarks largely reflect Wirelessestimator data, with little or no critical thought and certainly no ground breaking analysis. If there is a record (hard copy or electronic copy) of Mr. Foulke’s methodology from which a researcher can make an independent analysis, please provide it.

    The purported decrease in fatalities by Wirelessestimator is limited to the nine years (2003-2011) tracked by Wirelessestimator, a 28 year look at fatalities shows a clear trend increase.

    Fatalities 1984 - 2011

    Attached Files:

  7. Wireless Estimator

    Wireless Estimator Administrator Staff Member

    1. Earlier this month you said you were looking at over 250 tower fatalities since 1984, and "from that list I will determine which actually involved communications towers the[n] request the complete OSHA investigation files."

    Since your tower fatality chart shows approximately 250 fatalities, are these fatalities based upon OSHA responding to your Freedom of Information Act requests in less than three weeks?

    2. If you didn't spend the approximately $17,000 it would take for the FOIA requests and/or didn't get OSHA to respond in a timely manner (many times it has taken OSHA six months or more to complete a Wireless Estimator FOIA request), how did you obtain and vet your fatality statistics to produce your trend line chart and compare it to Wireless Estimator's published fatality statistics?

    3. Wouldn't it be obvious that your data points for your 28-year trend line should result in an increase, since 1984 is generally accepted as the year when analog service was first launched, and cell towers were few and many miles between - and in most states the total number of tower erectors wouldn't exceed the safe occupancy level of a McDonald's?

    You're correct that our fatality data clearly shows that there is a trend line indicating lower fatalities from 2003 through 2011.

    You're incorrect if you believe our fatality data is not accurate.

    But you can add to our body of research by responding to this post and sending us the OSHA investigation ID numbers that you're privy to, that are not in our fatality database from 2003 through 2012.
  8. towerman

    towerman Friend of the Community

    Why do you think it is better to go back 28 years to see a fatality trend which you say is going up rather than a 9 year trend that shows it going down?

    I think the industry should look at the last ten years or so and not when cellular first got started since there were very few climbers then and it would be natural to see fewer deaths at that time.

    There weren't climbing harnesses in 1984 and nobody, but nobody was using 100% fall protection. NATE wasn't even founded until about 12 years later and it was a while after that before safety really kicked in.

    For a guy that is so into research methodology and documentation, your chart gets a failing grade.

    It would be like you showing that the trend line has always been up for the Dow since 1940, but investors should ignore a trend line that is seriously down such as in 2007 when within a couple of years it dropped 50%.

    I don't think Wirelessestimator's statistics are faulty since they document every one, but they certainly could be in error. Do you disagree with their number of fatalities? Can you document it?

    Do you have documentation for your 250 fatalities that you can share? Otherwise your chart is worthless and misleads those professionals who are truly concerned about tower climber safety.
  9. Michael S Landa

    Michael S Landa Industry Observer

    Go to Start date: January 1, 1984 End date December 31, 1993 keyword: tower and hit Submit copy results to Excel and repeat with the process by changing dates to January 1, 1994 End date December 31, 2003 and January 1, 2004 End date December 31, 2011. Click on any accident summary you want and after you have viewed all summaries let me know what you think!
  10. Wireless Estimator

    Wireless Estimator Administrator Staff Member


    I was hoping that your 28 years of data was going to add to our nine-year body of research, but it's now apparent that you've done absolutely nothing to benefit the safety community other than provide a two minute OSHA database search that provides distorted data due to your faulty keyword requests.

    Running a Ditch Witch trencher across your keyboard could have drastically improved the results of the data used for your flawed analogies which at one point, to some, might have teetered towards credibility.

    A cursory review of your chart below shows that in 1984 there were five tower-worker-related fatalities. Using your search terms, it appears that there were eight.

    But more importantly, your search failed to look for other identifying keywords to find additional fatalities.

    Our extensive research has recognized that there will be additional industry-related deaths where "tower" will not appear in the keyword or abstract search. Just one of many you missed was the young man who was featured in "Tower Dogs" and fell to his death during the production period of the NBC Dateline special.

    In addition, your data doesn't include fatalities that OSHA opts to not investigate such as if the deceased was a sole proprietorship.

    Fast forward from 1984 to 2002 and your research chart shows that there were 7 fatalities. Using your suggested search terms, it appears that there are 19 verified fatalities of tower-work-related deaths - a 271% difference.

    Perhaps that's good enough for your precision levels of uniform and generally accepted government statistical methodologies; unfortunately it provides a disservice to the industry.

    But I must give you credit for accuracy in your chart from 2003 through 2011.

    Or rather, I must give Wireless Estimator credit since it's reflective of their data and not what you would have found using your identified search method using the OSHA database.

    I have no idea why you try to find fault with everybody else's research, programs and other endeavors to explore avenues to make our industry safer.

    You've previously stated that the National Association of Tower Erectors' "Star Program" initiative requirements may provide workers with additional skills, but it will not measurably improving safety.

    In the same series of keystrokes you identify that you have "examined the data in the OSHA accident database on multiple occasions and conclude that 66 percent or more of all falls from elevation were a result of risky behavior."

    It's evident that you're not good at capturing information from OSHA's database, so should that 66% be asterisked with a note stating that there could be a margin of error potential of 271%?

    You have stated that ProPublica, Frontline and Wireless Estimator have promulgated faulty conclusions based on bad assumptions, but your recent posts have allowed Dorothy's dog Toto to open your curtain of credibility, revealing that your console of wheels and levers are nothing more than poorly researched lackluster opinions.

    Your future announcements through the Wizard of Oz's microphone should be weighed with extreme prejudice.

    Angie Shyrigh likes this.
  11. Michael S Landa

    Michael S Landa Industry Observer

    I questioned your methodology not your integrity, honesty or motives; however, you seem content on questioning mine, very professional.
  12. Jimmy Colbert

    Jimmy Colbert Industry Observer

    Looks like nothing more than a glorified google search. Whats it supposed to tell you?
  13. Andrew

    Andrew Industry Observer

    Argue all you want about stats. It doesn't mean much. I started climbing before harnesses, however, we were allowed reasonable time to complete projects. Now that the "Turf Contractors" are in control every job needs to be finished yesterday.

    Tower hands are being pushed too hard for not enough money. In the last year I had a large ice shield dropped on me, luckily it was curved and didn't drop far. The curve of the shield prevented it from destroying both my legs.

    I got away with bruised shin bones and a good slamming to the ground. More recently, I narrowly missed getting hit by a 300' tape measure that came up a tool line without a down lead on it. I don't have any stats to say how fast it came at me , trust me, it was high speed.

    A very apologetic foreman w/8 plus years of experience confessed he was over tired from working sunrise to sunset for several weeks straight. Figure that into your statistics.

    P.S. Good thing I had my hardhat on and the JSA was filled out.
    One thing for sure, the status quo is going to kill again.
  14. The Estimator Webmaster

    The Estimator Webmaster Administrator Staff Member

    For the record, now that the Frontline and ProPublica pieces have been aired/published they also provided a well documented methodology for their statistical calculations. It doesn't appear that the postponement from February to May had anything to do with "bad assumptions" as this thread suggests.

Share This Page