Attachment 17 - Validation in the ORS: Analysis of Approaches

Validation in the Occupational Requirements Survey:
Analysis of Approaches

Kristin N. Smyth1

U.S. Bureau of Labor Statistics, 2 Massachusetts Ave., NE, Room 4160,
Washington, DC 20212

1

Abstract

The Bureau of Labor Statistics (BLS) is working with the Social Security Administration
(SSA) to carry out a series of tests to determine the feasibility of using the National
Compensation Survey (NCS) platform to accurately and reliably capture data that are
relevant to the SSA's disability program. The proposed new Occupational Requirements
Survey (ORS) is envisioned to be an establishment survey that provides information on the
vocational preparation and the cognitive and physical requirements of occupations in the
U.S. economy as well as the environmental conditions in which the work is performed. This
paper presents various methods for reviewing and validating the reasonableness of survey
estimates. Interactive graphics in visualization software packages have been developed as
new tools to complement traditional validation processes. These methods are being
designed to properly gauge the reasonableness of the estimates as there is little historical
precedent to which these data can be compared.

Key Words: data editing, visualization, data review, data graphics
1. Introduction
The Social Security Administration (SSA) approached the Bureau of Labor Statistics
(BLS), specifically the National Compensation Survey (NCS), because NCS collects data
on work characteristics of occupations in the U.S. economy. SSA is interested in
occupational information for use in their disability programs. The SSA and the BLS have
entered into annual agreements for collecting new data on occupational information for use
in the disability programs. The goal of the Occupational Requirements Survey (ORS) is to
collect and eventually publish information that will replace the outdated occupational data
currently used by SSA. All ORS products will be made public for use by non-profits,
employment agencies, state or federal agencies, the disability community, and other
stakeholders.
An ORS interviewer attempts to collect close to 70 data elements related to the
occupational requirements of a job. The following four groups of information will be
collected:




Physical demand characteristics/factors of occupations (e.g. crawling, hearing, or
stooping)
Educational Requirements
Cognitive elements required to perform work



Environmental conditions in which the work is completed

This paper explores the validation of the estimates for ORS data. Section 2 provides
background information on the Occupational Requirements Survey. Section 3 explains the
typical validation process and the challenges posed by the ORS data elements. Section 4
explores possible validation strategies. The paper ends with a conclusion and description
of further research to be completed.
2. Background Information on ORS
In addition to providing Social Security benefits to retirees and survivors, the Social
Security Administration administers two large disability programs that provide benefit
payments to millions of beneficiaries each year. Determinations for adult disability
applicants are based on a five step process that evaluates the capabilities of the worker, the
requirements of their past work, and their ability to perform other work in the U.S.
economy. In some cases, if an applicant is denied disability benefits, SSA policy requires
adjudicators to document the decision by citing examples of jobs the claimant can still
perform despite their limitations (such as limited ability to balance, stand, or carry objects)
[1].
For over 50 years, the Social Security Administration has turned to the Department of
Labor's Dictionary of Occupational Titles (DOT) [2] as its primary source of occupational
information to process the disability claims [3]. SSA has incorporated many DOT
conventions into their disability regulations. However, the DOT was last updated in its
entirety in the 1970’s, although a partial update was completed in 1991. Consequently, the
SSA adjudicators who make the disability decisions must continue to refer to an
increasingly outdated resource because it remains the most compatible with their statutory
mandate and is the best source of available data at this time.
When an applicant is denied SSA benefits, SSA must sometimes document the decision by
citing examples of jobs that the claimant can still perform, despite their functional
limitations. However, since the DOT has not been updated for so long, there are some jobs
in the American economy that are not even represented in the DOT, and other jobs, in fact
many often-cited jobs, no longer exist in large numbers in the American economy. For
example, a job that is often cited is “envelope addressor,” because it is an example of a
low-skilled job from the DOT with very low physical demands. There are serious doubts
about whether or not this job still exists in the economy.
SSA has investigated the numerous alternative data sources for the DOT such as adapting
the Employment and Training Administration’s Occupational Information Network
(O*NET) [4], using the BLS Occupational Employment Statistics program (OES) [5], and
developing their own survey. But SSA was not successful with any of these potential data
sources and turned to the National Compensation Survey (NCS) at the Bureau of Labor
Statistics (BLS).
NCS is a national survey of business establishments conducted by the BLS [6]. Initial data
from each sampled establishment are collected during a one year sample initiation period.
Many collected data elements are then updated each quarter while other data elements are
updated annually for at least three years. The data from the NCS are used to produce the

Employer Cost Index (ECI), Employer Costs for Employee Compensation (ECEC), and
various estimates about employer provided benefits. Additionally, data from the NCS are
combined with data from the OES to produce statistics that are used to help the President’s
Pay Agent and the Federal Salary Council recommend changes in how certain Federal
employees are paid.
In order to produce these measures, the NCS collects information about the sampled
business or governmental operation and about the occupations that are selected for detailed
study. Each sample unit is classified using the North American Industry Classification
System (NAICS) [7]. Each job selected for study is classified using the Standard
Occupational Classification system (SOC) [8]. In addition, each job is classified by work
level – from entry level to expert, nonsupervisory employee to manager, etc. [9]. These
distinctions are made by collecting information on the knowledge required to do the job,
the job controls provided, the complexity of the tasks, the contacts made by the workers,
and the physical environment where the work is performed. Many of these data elements
are very similar to the types of data needed by SSA for the disability determination process.
All NCS data collection is performed by professional economists or statisticians,
generically called field economists. Each field economist must have a college diploma and
is required to complete a rigorous training and certification program before collecting data
independently. As part of this training program, each field economist must complete
several calibration exercises to ensure that collected data are coded the same way no matter
which field economist collects the data. NCS uses processes like the field economist
training to help ensure that the data collected in all sectors of the economy in all parts of
the country are coded uniformly.
SSA asked the NCS to partner with them under an annual interagency reimbursable
agreement to test the NCS ability to use the NCS platform to collect data on four groups of
information related to employer requirements for an occupation:
 Physical demand (PD) characteristics/factors of occupations (e.g. crawling,
hearing, or stooping)
 Educational Requirements
 Cognitive elements required to perform work
 Environmental conditions in which the work is completed
If BLS is able to collect these data about work demands, SSA would have new and better
data to use in its disability programs. SSA cited three key advantages of using NCS to
provide this updated data:




Reputation - SSA was impressed with the BLS reputation for producing high quality,
statistically accurate data that are trusted by our data users and follow statistically
accepted methods and principles.
Trained Workforce – SSA was also impressed that NCS Field Economists have
experience collecting information about occupations in America’s work force and
collecting data similar to that needed by SSA.
Survey Infrastructure - After attempting to develop their own survey, SSA was also
impressed with the fact that NCS has infrastructure in place across the country to
manage and implement a new survey to meet their data needs as well as systems and
processes to support all the steps of the survey.

Since 2012, NCS has been testing our ability to collect these new data elements using the
NCS survey platform. Field testing to date has focused on developing procedures,
protocols, and collection aids using the NCS platform. These testing phases were analyzed
primarily using qualitative techniques but have shown that this survey is operationally
feasible. Now it is time to turn our attention to ensuring the high quality of the estimates
produced. This paper presents the validation process that is being developed for the
estimates in the ORS program.
3. Validation Processes
ORS captures occupational information on educational requirements, cognitive demands,
physical demands, and exposures to environmental conditions. Each of the data elements
falls into two data types: categorical and continuous. Data elements that are categorical
have a set of predetermined values, one of which will be selected as a response. Continuous
data may be limited by a minimum, such as zero hours, or a maximum, such as 100 percent.
The full list of data elements is shown below:
Physical Demands
Hearing
One on one hearing
Group hearing
Telephone
Other Sounds
Noise Intensity
Keyboarding (10-key)
Keyboarding (Other)
Keyboarding (Touch Screen)
Keyboarding (Traditional)
Far Visual Acuity
Near Visual Acuity
Peripheral Vision
Sitting
Sitting vs Standing at Will
Standing/Walking
Stooping
Kneeling
Crawling
Crouching
Pushing/Pulling with Feet Only
Pushing/Pulling with Foot/Leg
Pushing/Pulling with Hand/Arm
Reaching at/below Shoulder
Reaching Overhead
Strength

Educational Requirements
Certifications, license, and training
Degree
Literacy
Post-employment training
Previous Experience
SVP
Time to Average Performance
Skill Level
Job Zone
Environmental Condition
Extreme Cold
Extreme Heat
Fumes, Noxious Odors, Dusts, Gases
Heavy Vibration
High, Exposed Places
Humidity
Outdoors
Proximity to Moving Parts
Toxic, Caustic Chemicals
Wetness
Cognitive Elements
Frequency of verbal interaction with regular contacts
Frequency of verbal interaction with other contacts
Measure of Job Control

Most Weight Lifted
Amount of Weight Lifted Constantly
Amount of Weight Lifted Frequently
Amount of Weight Lifted Occasionally
Amount of Weight Lifted Seldomly
Climbing Ladders/Ropes/Scaffolds
Climbing Ramps/Stairs, Job-related
Climbing Ramps/Stairs, structural
Fine Manipulation
Gross Manipulation
Using One Hand or Both
Driving
Driving Vehicle Type

Frequency of Job Deviations in Tasks
Frequency of Job Deviations in Work Schedule
Frequency of Job Deviations in Work Location
Job Complexity
Type of Contact with Regular Contacts
Type of Contact with Other Contacts

Once the field economists collect and code the data, they undergo a series of quality
reviews. An ORS review program is in development to create the processes, procedures,
tools, and systems that will be used to check the micro-data as they come in from the field.
The ORS Review Program is a separate process that focuses on the micro-data and includes
the goal of micro-data review. This encompasses ensuring data integrity, furthering staff
development, and ensuring high quality data for use in producing survey tabulations or
estimates for validation. The review process has been designed to increase the efficiency
of review tools, build knowledge of patterns and relationships in the data, develop
expectations for reviewing the micro-data, help refine procedures, aid in analysis of the
data, and set expectations for validation of tabulations or future estimates.
The process for validating the estimates constitutes a separate but related set of activities.
Estimate validation focuses on aggregated tabulations of weighted data as opposed to
individual data. The goal of the validation process is to review the estimates and declare
them Fit-For-Use (FFU), or ready for use in publication and dissemination, as well as
confirming that our methodological processes (estimation, imputation, publication and
confidentiality criteria, and weighting) are working as intended. Validation processes
include investigating any anomalous estimates, handling them via suppressions or
correction, explaining them, documenting the outcomes, and communicating the changes
to inform any up/down-stream processes. All results of validation are documented.
Since validation and data review have interrelated goals, lessons learned from one may
influence the other in ORS development, and some tools may be of use in accomplishing
both goals. This paper aims to explain the validation side of these activities, but as it
necessarily relates to micro-data review, certain aspects of review are referred to as well.
The validation tools in development are intended to bring attention to the estimates that are
unusual or unreliable and assist in determining what went into the construction of those
estimates. Estimates or tabulations that do not conform to expectations are considered
anomalous. Expectations cover internal expectations and those found from outside sources
(if any). Part of developing the validation process is searching for and identifying patterns

in the micro-data and recognizing relationships for use in building future systems both in
validation and also in data review, like systematic review edits, reviewing parameters,
training reviewers, etc. These patterns in the micro-data help to not only form our
expectations for the estimates but also to provide information that may help in the
development of procedures, data collection, and review parameters.
Validation in the NCS platform is done by comparing the current estimate to a set of
expectations, primarily its historical counterpart. If the current estimate has changed by a
factor that is greater than the expected threshold, the underlying data for that estimate are
checked. NCS also assesses the effects of estimation, sampling, weighting, and
methodological or procedural changes to see if these factors have had an influence on the
tabulations or estimates.
In ORS, however, there exists no perfect historical match for the information being
collected. Previous occupational research studies are out of date or in a form that make
comparisons unwieldy. With few third party or historical sources to draw from for
validation, there exists a new challenge within ORS to identify unexpected tabulations.
NCS also makes use of unstructured review of estimates by senior economists based on
salient changes (such as heavily weighted observations) in the estimates during review.
This may be of use in the ORS program as well.
The data collected during testing for ORS are of limited use for developing estimate
validation. First, the test data are not numerous enough to produce estimates that can be
used as a testing ground for how the estimates should relate to one another. They also are
not weighted. Additionally, through the various phases of testing, the ORS questions and
collection procedures have evolved, making it difficult to compare data collected in the last
phase of testing to data collected in the first.
Thus, in order to validate the estimates from testing, some different methods have been
considered. The estimates within an occupation, or SOC code, are expected to remain
consistent between phases. When the estimate differs, the procedural changes that may
have been enacted are reviewed to determine if they played a role in the unexpected
estimate. Unexpected differences in the estimates between characteristics like full/parttime and union/non-union are also examined to make sure they are not having effects where
we wouldn’t expect. Over time we should also be able to uncover any strong correlations
between certain variables, such as time spent outdoors and lifting, which can then be used
as a predictor for those estimates.
There are three different types of estimates in the ORS dataset. Some questions have
categorical data, for instance an answer of either “Yes” or “No.” Some estimates are
averages based on continuous data, like a number of hours that a task is performed. Some
estimates are percentiles, showing how much activity is performed at a certain threshold.
In addition, some tabulations are created by compiling these different kinds of data, like
the “SVP” or Specific Vocational Preparation, which combines categorical data like the
degree required with continuous information like the number of months of training required
to arrive at a new measurement. Each of these types of estimates requires validation.
Determining whether an estimate is fit for use isn’t about liking the data, it is purely
ensuring that quality thresholds for things like response rate and imputation are met. It is
not the purpose to invalidate correctly collected data, only to ensure the process is working
as it should. The end result of the estimation process should be to validate that the

construction of the estimate is good, which means that the estimation processes are working
as intended and able to support the publication of accurate estimates.
4. Validation Techniques
The objective for a validation plan in ORS is to be able to ensure that estimates are
reasonable compared to expectations. The validation options available will depend on the
data element or elements being estimated and validated. The assumption in validating the
estimates is that the underlying data have been reviewed in the ORS data review process.
However, it may still be advantageous in the beginning phases of this new survey to
leverage validation tools to inform data review or vice versa. Validation may reveal
specific areas where the initial review of the micro-data may need to be confirmed, and in
some cases, where some additional review may be necessary. Several tools for validating
certain types of elements are being developed, including validation reports, outlier search
software, and relationship dashboards and graphs. Each of these tools is described below.
Validation Reports - The most straightforward tool in conducting validation will be
automatically generated reports. These will run from a program that looks at every single
individual estimate and flags estimates that should be reviewed by trained staff, either
because an estimate differs from a set expectation or because it differs from a historical
record. To review how consistent the data within a SOC (Standard Occupational
Classification) code are, the data that underlies the estimate for elements such as time to
average performance can be evaluated using standard deviation. These reports might cast
light on the specific areas to investigate further to identify the drivers of the estimates. Such
investigations may reveal that the micro-data underpinning that estimate had outliers that
were reviewed and verified during the ORS review process. It might also reveal that while
data were verified, the weight of that observation is notably large, causing the outlier to
have a larger than expected impact. It might trigger that an underlying piece of micro-data
is errant. All of the above would be investigated and explained where and when possible
and documented so that the estimates can be declared fit-for-use.
Outlier Search - Software options can be used to expeditiously drill down from any
anomalous estimates to the micro-data identified as contributing to “outliers” to verify the
underlying information. The anomalous estimates will be noted in other reports, but the
visual nature of the dashboards allows users to click down from the estimate to the
underlying data. For example, for the physical elements, the estimates will be available as
percentile estimates, and these can be lined up in a graphical interface to quickly see if
anything pops out as an outlier. By eliminating some of the manual work of reviewing
individual responses, more data can be reviewed. Related to the outlier search is the
examination of the relationship between elements, discussed below.
Relationship Dashboards and Graphs – JMP and Tableau Software easily integrate with
SAS to produce color coded, interactive graphical interfaces from data. With these tools,
estimates can be looked at within a broader context in order to identify which estimates are
not conforming to expectations. The dashboard may, for example, compare estimates on
lifting/carrying to estimates on weather. It seems that an established relationship might be
that jobs where the incumbent lifts/carries more than 33 percent of the day should take
place outdoors and be exposed to weather. With this visualization, any unusual deviation
from that expectation is easily seen and can be explored further.

To plan for such an examination, a hierarchy of expected relationships has been established
that indicates where a strong two or one way correlation should exist. These relationships
are documented in a matrix, shown below, which makes it easy to visually comprehend the
ways in which we expect the data to interact, and resultantly, show interrelationships
between estimates. Across both axes of the matrix are the data elements and where the two
elements meet in the middle, the box is shaded with a color to either indicate that there is
a strong, weak, one-sided, or no relationship. As more data become available these
relationships can become codified with evidence that they do have a statistical relationship
to one another. The chart below shows an example of what such a matrix might look like.
In this matrix, strong relationships are shown with various hues of green shading, weaker
relationships are shown with various hues of yellow shading, and no relationship is shown
with white boxes.

We expect the estimates to have logical relationships between the elements, for example
jobs with a higher measure of cognitive demand should have higher educational
requirements. Other high level examples of relationships include expecting jobs with
higher physical demands to often be found outdoors, similar physical activities to be
grouped similarly, jobs that require contact with the public to include physical and
cognitive elements that go along with such work. While these relationships exist at the
micro-data level, we are learning and developing expectations about what they will look
like at the aggregate, or estimate, level. No estimate would be expected to contradict any
review edits. Factors such as region, industry, full/part-time, and union/non-union will be
examined to ensure that they do not have effects on the data that would be unexpected. Our
expectation is that they will not, but as we gather more data we hope to learn more about
these issues. Over time, a library of expected relationships and variations will be built up
from what is learned.

6. Conclusion and Next Steps
ORS poses unique validation challenges because of the unique nature of the data being
collected. Little previous research has been done on collecting occupational cognitive
requirements. Additionally, the physical and environmental demands are being
documented in a new way for ORS. Validation plays a uniquely important role in the
production of these estimates because of their newness. By pioneering new ways to identify
unexpected patterns we can not only validate the data but gather information that can be
used to improve the data review process, the future validation process, and occupational
requirement research overall.

References
[1]

Social Security Administration, Occupational Information System Project,
http://www.ssa.gov/disabilityresearch/occupational_info_systems.html.

[2]

U.S. Department of Labor, Employment and Training Administration (1991),
“Dictionary of Occupational Titles, Fourth Edition, Revised 1991”

[3]

U.S. Department of Labor, O*Net Online, http://www.onetonline.org/

[4]

U.S. Bureau of Labor Statistics (2008) BLS Handbook of Methods, Occupational
Employment Statistics, Chapter 3. http://www.bls.gov/opub/hom/pdf/homch3.pdf

[5]

U.S. Bureau of Labor Statistics (2013) BLS Handbook of Methods, National
Compensation Measures, Chapter 8.
http://www.bls.gov/opub/hom/pdf/homch8.pdf

[6]

See U.S. Bureau of Labor Statistics, Handbook of Methods, Chapter 8

[7]

See North American Industry Classification System website,
http://www.census.gov/eos/www/naics/

[8]

See Standard Occupational Classification website, http://www.bls.gov/soc/

[9]

U.S. Bureau of Labor Statistics, National Compensation Survey: Guide for
Evaluating Your Firms’ Jobs and Pay, May 2013 (Revised),
http://www.bls.gov/ncs/ocs/sp/ncbr0004.pdf

Any opinions expressed in this paper are those of the author and do not constitute policy
of the Bureau of Labor Statistics or the Social Security Administration.