Methodology for Calculating Fatality Rates - Pacific Standard

Methodology for Calculating Fatality Rates

Author:
Publish date:

This methodology accompanies Antonia Juhasz's feature, "Death on the Dakota Access," which originally appeared in the September/October 2018 issue of Pacific Standard.

section-break

Basic Methodology

I used fatality statistics published by the Bureau of Labor Statistics and employment data from the BLS Current Employment Statistics survey. BLS classifies workers by industry using the North American Industry Classification System codes. I use NAICS 237120/Oil and Gas Pipeline and Related Structures Construction.

I generated employment-based oil and gas pipeline construction worker fatality rates and employment-based national average fatality rates. My data is internally consistent: I can compare my pipeline worker fatality rates to my national fatality rate average to determine how much greater the pipeline worker fatality rate is for the national rate.

However, my data cannot be directly compared to BLS Census of Fatal Occupational Injuries published fatality rates. This is because the BLS uses data from the Bureau of the Census' Current Population Survey (CPS) to generate hours-based employment data. However, the CPS does not generate hours-based data for all industries, and does not do so for oil and gas pipeline and related construction workers. Readers will note that I do not directly compare my fatality rates to the CFOI fatality rate data other than to confirm that my findings for pipeline workers are consistent with those reported for oil and gas extraction workers generally: significantly higher than the national average.

For those seeking to reproduce my findings, it is critical to note that the BLS data is not static. Rather, the BLS regularly revises all of its employment, fatality, and other related data going back in time when new data is made available. Also, the policies dictating how fatality information is reported, gathered, and shared by employers and the federal government can and have changed based on who is in the White House. Thus, data generated is subject to change based on the BLS revising the underlying data.

Finally, using a much more complicated methodology described below, I generated hours-based employment data for oil and gas pipeline construction workers, an hours-based fatality rate for oil and gas pipeline construction workers, and compared it to the CFOI national fatality rate. I did this as a check on my employment-based findings. My hours-based findings were nearly identical to my employment-based findings, reaffirming the validity of my calculations and findings.

A. The Data and How to Calculate Employment-Based Fatality Rates

1a: Number of Fatalities "Oil and Gas Pipeline and Related Structures Construction" Workers.

Bureau of Labor Statistics -> Injuries, Illness, and Fatalities -> National Data (in Left Hand Bar) -> Fatal Injuries Data -> Miscellaneous CFOI Data Tables; All Worker Profile, 2003-2016

Use: "Oil and gas pipeline and related structures construction."

1b: Employment Numbers – Total Oil and Gas Pipeline and Related Structures Employment

Use: NAICS Code 237120.

Bureau of Labor Statistics -> Current Employment Statistics -> CES Databases -> Multi-Screen -> Create Customized Tables

Steps:

  • Select "Seasonally adjusted" and click "next form."
  • Within "Supersector" choose "20 Construction" and click "next form."
  • Within "Datatype" choose "01 All Employees Thousands" and click "next form."
  • Within "Industry" choose "20237120 Oil and Gas Pipeline Construction" and click "next form."
  • Click "Retrieve data": set parameters at 2003 to 2018 using the dropdown at the top of the screen; select "include annual averages"; click "go."
  • Use the month of May.

According to an interview with the BLS, "seasonally adjusted" is the most commonly used data set.

The BLS does not calculate an annual average for each year. To remain consistent with the employment data providing by BLS for all workers, we use the month of "May" here as well.

1c: Employment Numbers – Total National Employment

Bureau of Labor Statistics -> Occupational Employment Statistics

For each year, choose "National" and "May."

Use: "All Occupations Total."

2: Number of Fatalities All Workers

Bureau of Labor Statistics -> Injuries, Illness, and Fatalities -> National Data (in Left Hand Bar) -> Fatal Injuries Data -> Miscellaneous CFOI Data Tables -> All Worker Profile, 2003-2016

Use: "Total" fatalities for all workers for each year.

3. Calculate Fatality Rate

Divide the number of fatalities by the total number of employees in each NAICS code, and multiply by 100,000 per the BLS.

4. Compare to National Average Fatality Rate

Divide the industry fatality rate by the national average fatality rate.

B. The Data and How to Calculate Hours-Based Fatality Rates

1: Number of Fatalities "Oil and Gas Pipeline and Related Structures Construction" Workers.

Bureau of Labor Statistics -> Injuries, Illness, and Fatalities -> National Data (in Left Hand Bar) -> Fatal Injuries Data -> Miscellaneous CFOI Data Tables -> All Worker Profile, 2003-2016

Use: "Oil and gas pipeline and related structures construction."

2: Hours of Employment – Total Oil and Gas Pipeline and Related Structures Hours of Work

Use: NAICS Code 237120.

Bureau of Labor Statistics -> Current Employment Statistics -> CES Databases -> Multi-Screen -> Create Customized Tables

Steps:

  • Select "Seasonally adjusted" and click "next form."
  • Within "Supersector" choose "20 Construction" and click "next form."
  • Within "Datatype" choose "56 Aggregate Weekly Hours of All Employees, Thousands" and click "next form."
  • Within "Industry" choose "20237120 Oil and Gas Pipeline Construction" and click "next form."
  • Click "Retrieve data": set parameters at 2006 to 2018 using the dropdown at the top of the screen; select "include annual averages"; click "go."
  • The data is complete only as of 2007.

This is where it gets tricky.

  1. Calculate the aggregate monthly hours worked by multiplying each month by four.
  2. Calculate the aggregate annual hours worked by summing the aggregate monthly totals for each year. Multiply by 1,000.
  3. Divide the annual total by 56 to get an average weekly amount for each year.
  4. Multiply the average weekly amount for each year by 50 to get a total number of hours worked per year for a fifty-week work year (per the BLS).

3: Calculate Fatality Rate

Calculate hours-based fatality rate: Number of fatalities divided by aggregate hours worked for 50 weeks. Times 200,000,000 (per BLS).

4: Compare to National Average Hours-Based Fatality Rates

Bureau of Labor Statistics -> Census of Fatal Occupational Injuries (CFOI) – Current and Revised Data -> Fatal Injury Rates

Hours-based fatal work injury rates listed for each year going back to 2006.

Divide the industry fatality rate by the national average fatality rate.

Detailed Methodology Discussion

The Bureau of Labor Statistics annually publishes the Census of Fatal Occupational Injuries, which includes fatality rates for select groups of workers. It has occasionally published fatality rates for oil and gas extraction workers, most recently in 2014. It has never done so for oil and gas pipeline workers. I therefore calculated fatality rates myself using data provided by the BLS, and with the guidance of the BLS, the Centers for Disease Control and Prevention, and the Bureau of the Census' Current Population Survey (CPS) experts.

No other agency or group that I could identify has produced fatality rates for oil and gas pipeline workers other than industry support company ISN, but its data is neither accessible nor replicable, nor are all of its findings publicly available. Those which are available, however, are consistent with those I generated, showing that oil and gas pipeline construction workers have fatality and incident rates on par with oil and gas extraction workers and, like extraction workers, often higher than the rest of the broader oil and gas industry.

I used fatality statistics published by the BLS and employment data from the the BLS's Current Employment Statistics survey. My data is necessarily employment-based data. I generated oil and gas pipeline worker fatality rates and an employment-based national average fatality rates. The data is internally consistent: I can compare my pipeline worker fatality rates to my national fatality rate average to determine how much greater the fatality rate is for the former. But this data cannot be directly compared to the BLS CFOI fatality rates for other industries because the BLS uses employment data from the Bureau of the Census' Current Population Survey to generate hours-based employment data. However, the CPS does not generate hours-based data for all industries, and does not do so for oil and gas pipeline construction workers (nor water and sewage pipeline workers). Thus, we cannot directly compare our employment-based fatality rates to the CFOI hours-based fatality rates.

The CES does provide hours-based employment data for specific sectors, but only as of 2007. Using this data is more complex, however, then the employment-based data, and thus prone to error and less replicable. As described above, as an exercise to determine if my employment-based findings were consistent with hours-based results, I did run the hours-based numbers, generating hours-based fatality rates for oil and gas pipeline and related construction workers for 2007–17. The results were virtually identical.

To additionally verify the validity and robustness of my pipeline worker calculations, I also calculated oil and gas extraction worker fatality rates using employment-based data for my own purposes. Readers will see that I do not directly compare my fatality rates to the CFOI extraction worker data, other than to confirm that my findings for pipeline workers are consistent with those reported for extraction workers generally: significantly higher than the national average. In the article, I reference the CFOI oil and gas extraction worker fatality rates published by the AFL-CIO in its Death on the Job annual report, as these are the numbers readers are most familiar with as they are oft-cited in press and academic reports.

The BLS classifies workers by industry using the North American Industry Classification System codes. I use: NAICS 237120/Oil and Gas Pipeline and Related Structures Construction. For definition, see: 2017 NAICS Manual, p. 128.

More details from federal government contracting website GovTribe:

NAICS 237120: Vendors that win government contracts in this category are primarily engaged in the construction of oil and gas lines, mains, refineries, and storage tanks. The work performed may include new work, reconstruction, rehabilitation, and repairs. Specialty trade contractors are included in this group if they are engaged in activities primarily related to oil and gas pipeline and related structures construction. All structures (including buildings) that are integral parts of oil and gas networks (e.g., storage tanks, pumping stations, and refineries) are included in this industry.

Examples:

  • Distribution Line, Gas And Oil, Construction
  • Gas Main Construction
  • Gathering Line, Gas And Oil Field, Construction
  • Natural Gas Pipeline Construction
  • Natural Gas Processing Plant Construction
  • Oil Refinery Construction

Like all NAICS categories (including that for oil and gas extraction), it is not a perfect fit, but it's the best and a sound data option for our purposes. Most importantly, it most likely accounts for only a portion of the actual fatalities suffered by workers engaged in oil and gas pipeline and related structures construction and related activities. It excludes, for example, some workers who one would want to consider, such as NAICS 486210/Pipeline Transportation of Natural Gas Workers and NAICS 486110/Pipeline Transportation of Crude Oil Workers. But, not all of these workers make sense to include. Thus, I reference in the article worker fatalities of those engaged in oil and gas pipeline maintenance, for example, but did not include these NAICS categories in the overall calculation.

There is also a problem with how pipeline workers are classified, as it is well-known that workers suffering fatalities while constructing oil and gas pipelines are often misclassified. For example, DAPL worker Nicholas Janesich's fatality was miscategorized as an oil and gas extraction worker death. It is also well-known that the data is inadequate as it relies heavily (not exclusively) on company self-reporting, raising concerns that many deaths attributed to "natural causes" are instead workplace incidents; it varies by state; and also does not account for fatalities caused by work-related illness. Often, journalists will supplement BLS fatality data with state-level reporting and media articles. While I contacted some states, I did not have capacity to contact all. And while I reference media reports on worker deaths and reviewed these regularly, I chose not to include them in the data set here because I am building a first-of-its kind oil and gas pipeline fatality report and I wanted it to be as straightforward and replicable as possible. Subsequent efforts should add to my work by compiling such additional sources.

Comparisons to Other Industries & NAICS Codes Used

NAICS 2121/Coal Mining

Same steps as above.

NAICS 211/213112 (incl. 213111)/Oil and Gas Extraction Workers

Same steps as above, except: The Centers for Disease Control and Prevention (CDC) wrote in 2015 that it used the following methodology when calculating fatalities for oil and gas extraction workers.

Note: While these were previously three distinct and separate codes, the BLS now reports (as of Dec 2017) to us that NAICS 213111 is now subsumed under NAICS 213112.

As the CDC wrote previously, explaining its choice of NAIC categories:

There are three types of companies in the oil and gas extraction industry: oil and gas operators that control and manage leased areas (NAICS 211), drilling contractors that drill the wells (NAICS 213111), and well-servicing companies that provide all other types of support operations that prepare a well for production and completion (NAICS 213112).

Employment-based fatality rates were calculated. Employment data for NAICS 213111 was not available using the Employment, Hours, and Earnings from the Current Employment Statistics survey tool from the BLS. The data is available via the BLS Quarterly Census of Employment and Wages.

QCEW Data

All steps are identical to above, except for the calculation of employment, or "Total Workers." Instead, employment data can be found using this tool:

  • Select the industry.
  • Select "US000 U.S. TOTAL."
  • Select "Private" (all that's available for 213111).
  • Select "1 All Employees."
  • Select "0 All Establishment Sizes."

NAICS 23711/Water and Sewer System Construction, for Water and Sewer System Pipeline Workers

There are no NAICS codes for alternative energy, but fatality numbers are posted by the BLS for "Solar photovoltaic installers" and "Windmills, wind turbines."

Background & Additional Information

CFOI fatality rates for years 1992 through 2007 are employment based—that is, they are measured in terms of the number of fatalities per a given number of workers. These rates measure the risk of fatal injury for those employed during a given year, but they do not take into account the amount of time workers are exposed to the risk. Fatality rates calculated in terms of injuries per number of employed workers alone do not fully capture the relative risk of incurring an occupational fatality. This discrepancy is most noticeable among groups that do not typically work a 40-hour work week, such as older and younger workers and workers in retail trade and mining. A more ideal measurement would account for the time a worker is exposed to dangers associated with fatal injuries. As a result, BLS introduced hours-based rates in June of 2009.

2017 NAICS Manual: See for NAICS definitions

  1. Oil and Gas Pipeline and Related Structures Construction (NAICS 23712): Manual p. 128
  2. Oil and Gas Extraction (NAICS 211): Manual p. 105
  3. Support Activities for Mining (NAICS 213): Manual p. 115
  4. Drilling Oil and Gas Wells (NAICS 213111): Manual p. 115
  5. Support Activities for Oil and Gas Operations (NAICS 213112): Manual p. 116
  6. NAICS 2121/Coal Mining: Manual p. 107
  7. NAICS 23711/Water and sewer system construction: Manual p. 127

Related