I wrote about the Reliability of the Occupational Requirements Survey in May 2022. This week, a vocational witness left this reply:
Here is where I am, as a Vocational Expert. Since the ORS groups things by SOC code and not a specific DOT code, it is impossible to break down all the variables to a point that it is reliable. You can have a SOC with, let's say, 20 DOT codes. We can not assume that each job would match the ORS information. And, as an expert, I have to use all the ORS data, even down to "ramps and stairs" and "ladders ropes and scaffolds". Pretty much made up my mind to walk away until they fix this mess which we all know might never happen.
This is a really good response. How does anyone break down the SOC groups to apply to individual DOT codes? Fair comment. If a VW cannot break down the data in the ORS to the individual DOT codes, how does a VW break down the one variable from the Employment Projections (the foundation of the OOH) or the Occupational Employment and Wage Statistics (OEWS, used by SkillTRAN) -- job numbers -- to any individual DOT code?
Whatever methodology a statistician uses to break down job numbers, that statistician would use the same techniques for breaking down the physical, cognitive, environmental, or experiential requirements of work described in the ORS. SkillTRAN uses an occupational density model (an opaque phrase). That method uses the industry designation or the job duty descriptions to pick industries appropriate for that DOT code. SkillTRAN engages that process for all DOT codes. Using OEWS and County Business Patterns data, SkillTRAN accumulates the job numbers for that SOC/OEWS code at those industry (NAICS) intersections, and then divides by the number of DOT codes. Adding up the job numbers in each SOC-NAICS intersection results in a job number estimate. SkillTRAN uses the suspect equal distribution method for calculating job numbers at the SOC-NAICS intersection.
The ORS is a blunt knife. With the data that is now available, the question asks whether it is possible for the job number suggested by SkillTRAN or the VW to be reliable. Vague discussions don't help. Examples crystalize the problem.
There are 1,590 DOT codes aggregated in production workers, all other including 52 of the 137 sedentary unskilled DOT codes. The 2022 EP estimates that 275,300 jobs exist for this broad occupational group. The OEWS estimates 243,500 jobs. The ORS reports a null estimate for sedentary jobs regardless of skill level. SkillTRAN reports less than 0.5%. Any estimate over 1,376 jobs for all sedentary occupations (skilled, semi-skilled, and unskilled) does not conform to the ORS.
Production workers, all other contains 405 light unskilled DOT codes. The ORS reports 11.1% of the jobs represent light work. Let's round off and call it 30,000 light jobs. Any testimony that there are 30,000 light unskilled jobs in a single light unskilled DOT code conflicts with the ORS. Accounting for the 26.4% of unskilled work and engaging the assumption that skill levels distribute across the exertional levels permits less than 3,000 light unskilled production workers in 405 DOT codes.
This application of the ORS does not try to tease out the job numbers for a DOT code. This application of the ORS extracts the job numbers for the set of jobs that have the overarching characteristic. It is more difficult when the question layers the claimant with multiple limitations -- e.g., social, manipulative, pace. But if we know that light production workers, all other represent 30,000 jobs and unskilled workers at all exertional levels in that group represent fewer than 75,000 jobs, we know that no single light unskilled DOT code could ever represent more jobs. We know that the 52 sedentary DOT codes represent less than 1,400 jobs.
The ORS, like the EP and the OEWS, does not lend itself to DOT job numbers. That troika of sources does establish the number of jobs within an occupation-industry intersection and the aggregate number of jobs with the critical characteristics at issue. Inside of those boxes, the VW can exercise experience. But they can never reliably estimate job numbers outside of the boxes erected by the data sources.
Using the ORS is not easy. I agree with Gilkison. The difficulty is why the agency calls the VW an expert. You are called upon to engage in statistical analysis that is not part of the general requirements to be a VW. So when the ALJ asks for a sedentary unskilled reasoning level 1 occupation, tell the ALJ that lens inserter does not exist in any significant number. You can back up that testimony with the EP, the OEWS, the ORS, and County Business Patterns.
Use the data.
___________________________
Suggested Citation:
Lawrence Rohlfing, A Vocational Expert Responds -- and I Reply, California Social Security Attorney (May 17, 2024)
https://californiasocialsecurityattorney.blogspot.com
The author has been AV-rated since 2000 and listed in Super Lawyers since 2008.
