|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 196
o The Classification of
o
Occupations: A Review of
Selected Systems
THE CLASSIFICATION STRUCTURE OF THE DOT
The purpose of the DOT occupational classification system is to organize
occupations into groups that are similar in the sense that they tend to
make similar demands on workers or in which workers with specific
qualifications or characteristics are likely to find satisfactory employment.
The first edition DOT noted in its foreword (p. xi):
As a product of the [Employment Service] Research Program, [the DOT] is part of
a directed effort designed to furnish public employment offices in this country with
information and techniques that will facilitate proper classification and placement
of work seekers.
Getting qualified workers into appropriate jobs is a task that can be done most
adequately when the transaction is based on a thorough knowledge of both worker
and job.... Thus, it becomes part of the duties of public employment offices to
learn as much as possible about jobs and workers in order to be able to act as an
effective placement agency. If a foundry superintendent wants the public
employment office to send him a cupola tender, the office must know enough about
the work and worker to be able to refer a registrant who has been previously
classified as qualified and capable of doing the work required.
The DOT was developed to provide Employment Service interviewers and
cQunselers with the information necessary to classify workers and jobs
appropriately in order to match them.
The fourth edition DOT reflects the continued primacy of job-worker
matching as the reason for its existence. The first sentence of its
196
OCR for page 197
The Classification of Occupations
197
introduction lists job matching as the primary justification for producing
the DOT (p. xiii):
The Dictionary of Occupational Titles is an outgrowth of the needs of the public
employment service system for a comprehensive body of standardized occupational
information for purposes of job placement, employment counseling and occupa-
tional and career guidance, and for labor market information services. In order to
implement effectively its primary assignment of matching jobs and workers, the
public employment service system requires a uniform occupational language for
use in all its offices. This is needed to compare and match the specifications of
employer job openings and the qualifications of applicants who are seeking jobs
through its facilities.
CREATING OCCUPATIONAL TITLES
The process by which the millions of jobs in the economy are grouped into
the occupational titles in the DOT iS crucial in determining the usefulness
of the DOT as a matching tool. The fourth edition DOT describes this
process in general terms (p. xv):
Work is organized in a variety of ways. As a result of technological, economic and
sociological influences, nearly every job in the economy is performed slightly
differently from any other job. Every job is also similar to a number of other jobs.
In order to look at the millions of jobs in the U.S. economy in an organized way,
the DOT groups jobs into "occupations" based on their similarities and defines the
structure and content of all listed occupations. Occupational definitions are the
result of comprehensive studies of how similar jobs are performed in establishments
all over the nation and are composites of data collected from diverse sources. The
term "occupation," as used in the DOT, refers to this collective description of a
number of individual jobs performed, with minor vanations, in many establish-
ments.
The process of arriving at the 12,099 occupations defined in the fourth
edition involves two steps, which are described in detail in chapters 6 and
7. First, on the basis of actual observation of workers in a number of
positions, a job description is written by completion of a job analysis
schedule. Then the job descriptions are grouped into occupations, and
composite descriptions are prepared for inclusion in the DOT. Conceptual-
ly, these two steps are similar; both "job" and "occupation" are theoretical
entities. The central question in creating these entities is how to delineate
the boundaries, by deciding how much heterogeneity should be tolerated
within them.
The same kind of question arises in the next step in the process:
arranging the 12,099 occupational definitions into a classification struc-
ture. The remainder of this chapter is devoted to this topic.
OCR for page 198
198
GROUPING OCCUPATIONS
WORK, JOBS, AND OCCUPATIONS
Traditionally, the Employment Service has used the occupational titles in
the DOT in a relatively straightforward manner to match available jobs and
workers. For example, those who indicate that they are plumbers (or have
worked as plumbers) are matched with any openings for plumbers on file
at the local Employment Service office. The matching procedures are
relatively uncomplicated if there are job openings for plumbers (and if the
plumber is satisfied with one or more of the available positions).
A more difficult question arises when there are no openings in the
occupation in which a worker is classified. In such instances the
Employment Service either must send the worker away without offering
him or her any opportunity for employment or must make fundamental
decisions about the similarity of occupations and accurate estimates about
the degree of transferability of the worker's skills and experience in past
occupations to one or more alternative occupations. Interviewers appar-
ently do this quite frequently. As we have already noted, assessing the
transferability of skills goes beyond the paradigm for job-worker matching
that originally motivated the DOT and was expressed in its first edition; the
underlying principle is, however, extremely important.
Workers typically can perform in many occupations besides the ones in
which they have previously been employed; moreover, many skills are
learned on the job. It may be that for a large number of jobs, previous
work experience is more or less irrelevant. Among the 12,099 occupations
described in the DOT we find a great many that appear to involve similar
skills and aptitudes. For example, workers who have experience as a
Landscape Laborer (408.687-014) may be reasonable referrals for occupa-
tions such as Laborer, Brush Clearing (459.687-010) and Laborer, Golf
Course (406.683-010~. We note, however, that these occupations are not
grouped together in the DOT'S classification scheme.
Since the DOT classification is used to organize files of job openings and
applicants in local Employment Service office job banks, the location of
occupations in the classification structure will effectively determine to
which job openings a job seeker is exposed. This is particularly the case if
the lists of job openings are extensive, as they are in large labor markets.
The Landscape Laborer (408.687-014) mentioned above might have to
search through listings for many jobs before coming upon an opening for a
Laborer, Brush Clearing (459.687-010~. To the extent that any ordering
scheme makes it easier to locate an appropriate job, it obviously increases
the employment opportunities of workers, especially since (in the offices
we visted) more than 70 percent of all Employment Service job referrals
OCR for page 199
The Classification of Occupations
199
are made by workers themselves finding appropriate job openings in the
lists maintained by the employment office.
It is simple to observe that a system that directs landscape laborers only
to openings in that field and does not refer them to employment
opportunities in related areas is overly restrictive. It is a far less simple
matter to develop a general solution to this problem. The difficult question
is how to decide, in general, what constitute reasonable occupational
referrals. How does one decide what occupations are similar? How can one
expand the employment opportunities of workers who seek work at the
local Employment Service offices?
Two plausible approaches to these questions are embodied in the
classification structure of the fourth edition DOT. First, occupations are
organized in groups essentially according to technologies; 559 of these
occupational groups are represented by the first three digits of the DOT
occupational code. Second, each occupation is also characterized by the
requirements it places on workers in terms of their interaction with data,
people, and things; these requirements are represented by the second three
digits of the code. Either set (or both sets together) of digits could be
viewed as a reasonable index of the similarity of occupations, although it is
important to note that neither has ever been validated against an external
standard. In the fourth edition the occupational titles and their definitions
appear in order according to the numerical DOT codes. This ordering
means that the technologically defined groups have precedence in the
classification system over the worker function groups.
The DOT Code: The First Three Digits
The occupational groups represented by the first three digits of the code
appear to have been developed in an ad hoc manner by considering a
composite of industry; work field; machines, tools, equipment, and work
aids (MTEWA);i and materials, products, subject matter, and services
(MPSMS). (See chapter 6 and U.S. Department of Labor (1972:5-7) for a
detailed discussion of these concepts.) In assigning the first three digits to
an occupation, the job analyst or the definition writer is instructed to
consult the DOT'S existing classification, particularly the narrative descrip-
tions of the major categories and divisions, in order to identify in which of
the 559 occupational groups the occupation belongs. This process involves
iIn the work field "logging," for example, the following descriptions of tasks are suggested:
"Climbs tree, using climbing spurs and safety rope, and cuts limbs, knots, and top from tree
with ax and handsaw" (U.S. Department of Labor, 1972:89).
OCR for page 200
200
WORK, JOBS, AND OCCUPATIONS
a good deal of judgment, since some of the categories are very similar. In
the absence of more specific guidelines, the DOT code may be assigned in a
somewhat arbitrary manner. Moreover, reliance on the existing clas-
sification of the third edition DOT as the bench mark for the assignment of
category codes in the fourth edition undoubtedly discouraged rearrange-
ments of the classification and changes in codes. While definition writers
for the fourth edition could in principle recommend that categories be
combined or eliminated or that new ones be created, there is no
documentation of such recommendations.
The DOT Code: The Second Three Digits
The third edition DOT and various trial matching programs used the
worker function scales (the second three digits of the DOT code) to identify
occupational groups,2 but the fourth edition makes no attempt to do so.
Attempts to classify occupations solely on the basis of their complexity in
relation to data, people, and things have been generally unsuccessful. An
automated matching system based on worker function codes did not work
out, nor did a manual matching attempt in Pittsburgh during the
experimental period prior to publication of the third edition.3
Despite the failure of the worker function scales to serve as an adequate
basis for matching, the concept is probably useful in developing a
classification system for matching. The worker functions are intended to
summarize characteristics of workers required by the job (such as their
interests and aptitudes).4 They were developed from a realization that
every job is actually a job-worker situation and that to describe such a
situation adequately, knowledge of the characteristics of both the job and
the worker is required. The worker traits required by a job are not
2The 22 major categories of the worker function scales, called areas of work, appear to have
been developed in an ad hoc manner. Within each category, occupational groups of related
worker function codes are created, but the correspondence of groups to codes is not unique.
The same worker function code (e.g., .288) often appears in many different groups. This is
perhaps not surprising, since worker function codes attempt to measure the complexity of the
job and omit reference to specific skills, which are often important in placement.
3Interview, Adaline Padgett, occupational analyst, Division of Occupational Analysis, U.S.
Employment Service, August 1979.
4In their description of the Functional Occupational Classification Project, Fine and Heinz
(1958) recount the process by which all the occupational measures created for a group of
4,000 experimental occupations were used as bases for sorting the data into groups of similar
occupations; they concluded that the worker functions form the best groups because the
profile of the occupations on all the other variables was fairly consistent within worker
function groups, at least more so than for groups formed on other bases. It should be pointed
out that the techniques for discerning common patterns in data have advanced significantly
since the mid-1950s when this research was done. Fine and Heinz sorted the data repeatedly
in a search for consistent patterns.
OCR for page 201
The Classification of Occupations
201
captured by the considerations that currently enter into determining the
first three digits of the DOT codes, which appear to be based primarily on
technological processes.
Summary
Because the transferability of skills is generally regarded as the most
appropriate criterion for assessing the similarity of occupations with a view
to matching jobs and workers, the two aspects of the classification
structure inherent in the DOT codes (the first and second sets of three
digits) were ostensibly designed to capture two important elements of the
transferability of skills. The first element is job-specific knowledge or
skills the technological aspects of the occupation, the particular subject
matter, and the materials and equipment used. These are described by the
first three digits of the DOT code. The second element involves the qualities
of workers that are required by jobs. These are thought to be captured by
the worker function codes (the second three digits) because the worker
function configurations "profile" a variety of worker traits consistently.
These two bases of classification of the occupational titles in the DOT are
conceptually quite appropriate in the judgment of the committee. The
implementation of these concepts in practice, however, appears to be
somewhat inadequate, and it remains an open question whether these two
elements of the transferability of skills could not be better tapped by
indicators based on other methodologies.
THE KEYWORD SYSTEM OF THE EMPLOYMENT SERVICE
More an alternative mechanism for matching job applicants with job
openings than an alternative classification system, an automated keyword
system has been implemented in a number of Employment Service offices
throughout the country. This system was the subject of a limited staff
review, reported in Appendix G. The main conclusion that should be
drawn from this review is that although automation of the matching
process is highly desirable, the keyword system as it is currently
implemented suffers severe difficulties and needs to be thoroughly reviewed
by a committee of experts, a task that goes beyond the charge to our
committee.
EXISTING ALTERNATIVE CLASSIFICATIONS FOR JOB
WORKER MATCHING
Several alternative methodologies for constructing occupational clas-
sifications have been developed in recent years. The task inventories and
OCR for page 202
202
WORK, JOBS, AND OCCUPATIONS
the Position Analysis Questionnaire (PAQ) techniques described below may
be especially useful in identifying the types of skill components involved in
particular tasks and jobs. The identification, measurement, and clas-
sification of skills is at the heart of the problem of personjob matching.
With a thorough understanding of the skills required by a broad variety of
jobs as well as sound measures of those skills, it might be possible to
develop taxonomies of persons and occupations that would facilitate
differential placement, counseling, guidance, and education (see Altman,
1976; Canada Employment and Immigration Commission, Occupational
and Career Analysis and Development Branch, 1978; McKinlay, 1976~.
The feasibility of developing classifications based on knowledge of skill
content, at least in certain fields of work, is illustrated by descriptions of
certain military occupations (Morsh, 1966~. The PAQ data for 746 jobs
have been used to create job families based on worker-oriented dimensions
of skill, omitting the technological aspects (Straw et al. (1977~; also see
Colbert and Taylor (1978), Taylor (1978), and Taylor and Colbert (1978~.
Another inventory, the Occupational Analysis Inventory, has been used to
group 1,414 jobs into 21 clusters (Pass and Cunningham, 1976~.
Classifications based on job dimensions derived from structured job
analysis or task inventories, when combined with taxonomies of human
performance (Fleishman, 1975), appear to provide another way to address
directly the issue of the suitability of workers' skills, abilities, or other
characteristics for specific categories of jobs. As Dunnette (1976:516)
notes, there now exist several methods for describing or predicting how
efficiently different persons may be expected to perform various work
functions:
Further research . . . should focus on developing . . . short, easily adminis-
tered, and easily understood behavior description inventor~ies] which may be used
as a common basis for classifying jobs, tasks, job dimensions, human attributes,
aptitudes, skills, and tests and inventories into the same taxonomic system.
Because of the level of detail involved in inventory approaches, research
has been limited so far to a few work areas. Many more areas would have
to be studied to make these techniques generally useful in developing
classifications for personjob matching purposes. Despite this limitation
the inventory approach is one that appears to be worth pursuing.
Counseling psychologists have attempted to create classifications of
workers and jobs of the sort suggested by Dunnette. The resulting systems
for personjob matching do not involve as much detail as do the task
inventory approaches and are also more indirect, usually relying on the
characteristics of persons rather than the characteristics of the work itself
for the development of the matching scheme. Two attempts, the
OCR for page 203
The Classification of Occupations
203
Minnesota theory of work adjustment (Borgen et al., 1972; Dawis and
Lofquist, 1974, 1975, 1976; Lofquist and Dawis, 1969; Rosen et al., 1972)
and Holland's theory of careers (Holland, 1566, 1968, 1973a, 1976; J.
Holland and G. Gottfredson, 1976; Holland et al., 1972) use classifications
of occupations to explain vocational adjustment and vocational choice,
respectively. These two schemes are of interest for three reasons: First,
both theories have developed occupational classifications for the specific
purpose of matching workers with jobs. Second, both theories incorporate
independent but parallel classifications (Holland) or characteristics
(Minnesota) of persons and occupations and explicit procedures for
specifying the degree of match between a person and a number of
occupations. Third, both perspectives have generated substantial re-
search.5
MINNESOTA THEORY OF WORK ADJUSTMENT
According to the Minnesota theory, the greater the correspondence
between a person's abilities and the patterns of aptitudes required by a job,
the better his or her performance (satisfactoriness) and the greater his or
her persistence in the job. Similarly, correspondence between a person's
"needs" (values, interests) and patterns of occupational reinforcers leads,
according to the theory, to job satisfaction and persistence. Recently,
Dawis and Lofquist (1974, 1975) have also used occupational aptitude
pattern clusters and occupational reinforcer pattern clusters to form a
classification of occupations and have shown how this classification is
related to the DOT and Holland classifications.
In general, the evidence about the usefulness of the Minnesota theory
implies moderate support for the theory and its associated tools (Betz et
al., 1966; Elizur and Teiner, 1977; Weiss et al., 1965, 1966~. The theory
predicts satisfaction more efficiently than performance, and researchers
have found relatively stable differences among occupations in their
patterns of reinforcers and aptitude requirements. It is also clear that
predictions of performance (satisfactoriness, satisfaction, and persistence)
are relatively inefficient, even with the aid of this elaborate and carefully
constructed set of tools for personjob matching. These relatively weak
predictions of important job-related criteria are not, however, limited to
this particular theory. In this area of research, strong statistical associa-
tions between predictors and criteria are rare (Dunnette, 1976; Ghiselli,
1973; Schletzer, 1966~.
5For reviews, summaries, critiques, and important tests, see Osipow (1973), Walsh (1973),
McCormick (1979), McKinlay (1976), L. Gottfredson (1978), and Rounds et al. (1978).
OCR for page 204
204
WORK, JOBS, AND OCCUPATIONS
The major limitations of the Minnesota theory are twofold. First, it
relies on a number of indirect approaches to the determination of job
characteristics, inferring them, for example, from employee or supervisor
ratings or from the characteristics (especially abilities) of workers who are
employed in an occupation. Second, the range of occupations for which
occupational reinforcers and aptitude patterns are available is currently
limited; data for only 148 occupations are available. Nevertheless, the
Minnesota work demonstrates that, in principle, it is possible to engineer
the independent assessment of persons and jobs in parallel ways so that the
degree of workerjob match can be estimated. Such an approach could
prove effective in capturing the two elements of skill transferability noted
above, particularly the qualities of workers that are required by jobs.
HOLLAND CLASSIFICATION OF CAREERS
The second counseling approach to personjob matching is illustrated by
Holland's (1973a) theory of careers. Holland has developed a typology of
persons and occupations that includes six types: realistic, investigative,
artistic, social, enterprising, and conventional. Inventories such as the
Vocational Preference Inventory (Holland, 1978), the Self-Directed Search
(Holland, 1973b), the Strong-Campbell Interest Inventory (Campbell,
1977), or 1'Inventoire Personnel (DuPont, 1979) are used to locate
individuals in this topology, and, in turn, the modal characteristic of
incumbents in a particular occupation is used to characterize that
occupation.6 For counseling purposes, matches are made between the
personalities of individuals and this characterization of occupations.
Underlying the Holland classification is the notion that vocational choices
are expressions of personality; thus there should be greater similarity
between the personalities of incumbents of the same occupation than
between incumbents of different occupations. Personjob congruence is
said to exist when the personality type that a person most resembles
accords with the category into which a given occupation falls. Congruence
leads, according to the theory, to satisfaction, success, and stability or
tenure in an occupation.
6Often, when data on profiles of job incumbents are unavailable, Holland and his colleagues
resort to indirect approximations involving a substantial degree of judgment in classifying
occupations. Use has been made of Strong Vocational Interest Blank data, Ruder Preference
Inventory data, PAQ data, and observations of regularities between the Holland occupational
classification and the DOT classification (Holland, 1973a; Holland et al., 1972). Approxima-
tion techniques exist for assigning a Holland category to all 1960 and 1970 census
occupations (L. Gottfredson and V. Brown, 1978) and to all third edition DoT.titles
(Viernstein, 1972).
OCR for page 205
The Classification of Occupations
20S
Tests of Holland's theoretical formulations have had mixed results. On
one hand, despite the amount of subjective judgment involved in the
classification of occupations, the occupational classification shows strong
relationships of expected kinds with the Minnesota occupational reinforcer
scales; the DOT worker functions, svP, and GED ratings; self-direction; and
prestige (L. Gottfredson, 1978; Rounds et al., 1978~. The Holland
occupational classification has also been shown to be efficient in organizing
occupational mobility data in that the category of a worker's later job is
substantially predictable from knowledge of the category of a worker's
earlier job for those who change jobs (G. Gottfredson, 1977; Holland et al.,
1973; Nafziger et al., 1974~. Finally, evidence verifying the dimensions of
vocational interests that underlie Holland's classification of persons
implies that his groupings are reasonably sound (Guilford et al., 1954;
Hanson and Cole, 1973; Nafziger and Helms, 1974~.
On the other hand, the classification appears to be most useful when it is
supplemented by a general measure of occupational level such as the GED.
It focuses primarily on occupational preferences, and the measurement of
occupations is indirect. The occupational classification resembles in some
ways the approach to development of the Occupational Ability Patterns of
Dvorak (1935) and Patterson and Darley (1936) during the depression,
paying little direct attention to the details of the work performed or the
skills required to perform them. Also, it organizes occupational mobility
and congruence data better for older than for younger people (G.
Gottfredson, 1977; L. Gottfredson, 1979~.7 Variation in the methods used
to classify occupations or persons results in slightly different classifications
("identifications" in Sokal's (1974) terms).
Moreover, Holland's theory is incomplete with respect to the roles
played by social class, intelligence, and special aptitudes in the allocation
of persons to jobs. The theory incorporates a number of secondary
propositions about the degree of congruence among the personality and
occupational types that have not been discussed here (see Holland, 1973a).
In general, the research tests of these secondary propositions have yielded
weak support (G. Gottfredson, 1977; Nafziger et al., 1974; Rounds et al.,
1978~. The proposition that congruence leads to success in an occupation
is largely untested, and the evidence that congruence leads to job
satisfaction is very weak (see the studies cited by Rounds et al. (1978~.
Perhaps this reflects the fact that individual traits change over time as a result of
occupational experience, in such a way as to create greater conformity between individual
and occupational characteristics (Kohn and Schooler, 1973).
OCR for page 206
206
SUMMARY
WORK, JOBS, AND OCCUPATIONS
In short, the Holland and Minnesota approaches to personjob matching
illustrate the value of independent but parallel assessments of persons and
jobs and of the resulting occupational classifications, but both approaches
employ limited mechanisms for the assessment of actual job content and
skill requirements. An approach to occupational classification that seeks a
middle ground between the extreme specificity of task analysis and the
detailed examination of human abilities exemplified by Fleishman's (1975)
work, on one hand, and the more global but indirect approaches to the
parallel classification of persons and jobs illustrated by the Holland and
Minnesota schemes, on the other, may be a fruitful approach to the
improvement of the classification of occupations for the purpose of
matching workers and jobs.
A MOBILITY-BASED APPROACH TO JOB-WORKER
MATCHING
The transferability of skills between occupations should be the primary
basis for classifications whose purpose is job-worker matching. The
mobility that occurs in the labor market, specifically the changes between
occupations that workers sometimes make when they change jobs,
provides one indicator of the transferability of skills between occupations.
If workers move frequently back and forth between a pair of occupations,
we can infer that the occupations require similar aptitudes and skills, or at
least that those who perform one occupation are generally capable of
performing the other; otherwise, transfers would not occur.8 Clas-
sifications that have been developed for the purpose of job-worker
matching should group together those occupations among which workers
commonly transfer. As we have seen, in the DOT classification many jobs
that appear to require similar skills are placed in widely different
occupational categories. For example, Dispatcher, Radio (379.362-010) is
classified as a protective service occupation, while Dispatcher, Traffic or
System (919.162-010), which involves essentially the same skills, is
classified as a miscellaneous transportation occupation. Similarly, Engrav-
er, Hand, Hard Metals (704.381-026) is classified as a benchwork
~Obviously, one-way transfers must be treated more cautiously, since they may represent
promotion ladders. It would not be desirable, for example, to send an assembly line worker to
an opening for foreman even though foremen are almost entirely drawn from the ranks of line
workers. In practice, however, this is not much of a problem, since supervisory personnel are
almost always promoted from within. See Appendix H for a discussion of ways to use
unidirectional transfers to infer career ladders.
OCR for page 207
The Classification of Occupations
207
occupation, while Die Maker (979.281-010), which involves similar tasks,
is classified as a miscellaneous occupation.
Naturally occuring mobility between occupations' categories is a
sufficient but not necessary indicator of the transferability of skills. There
are many jobs between which mobility does not occur despite similarity in
content, because of custom, discrimination, or other reasons (McKinley,
1976~. For example, women who are secretaries move into managerial jobs
only rarely, primarily because of tradition and prejudice, even though
secretarial skills such as planning and coordinating may be highly relevant
to many managerial jobs. Hence it would be unwise to rely on mobility
patterns as the only or even the primary basis for assessing occupational
similarity.
However, a mobility approach may provide a useful supplement to
traditional methods of assessing the similarity of occupations and the
transferability of workers, by providing an empirical criterion for judging
the similarity of occupations and the substitutability of labor.9 Whereas
the classifications we have reviewed above rely mainly on analysts'
judgments regarding the similarity of jobs, in the mobility approach,
occupations are grouped solely because of high degrees of movement
between them. The nature of occupations need not be analyzed in order to
identify similarities to be used as a basis for classification; it is necessary
only to locate movement among occupations, whatever their nature. (The
mobility approach must, however, rely on other approaches to define the
basic occupations; 100 million positions in the economy must first be
classified into a reasonable number of occupational titles before movement
between occupations can be assessed.) In this section we describe the
potential of mobility data as a basis for constructing classifications and
enumerate the advantages and disadvantages of this approach.
We have undertaken some exploratory analyses to assess the feasibility
of developing alternative classifications based on the available job mobility
data. Basically, we attempted to group in clusters those jobs between
which the rates of transfer were high. Technical details of these analyses
are provided in Appendix H; for similar work, see Dauffenbach (1973~.
Our analyses have led us to several general conclusions:
1. Mobility data can be useful for constructing an occupational
classification that is useful for placement, but the basic occupational titles
for which mobility data are collected must be defined by other procedures.
Occupational mobility data can contribute little to the definition of
9See Roe et al. (1966) and Holland et al. (1973) for earlier studies of classifications using
mobility data.
OCR for page 208
208
WORK, JOBS, AND OCCUPATIONS
occupations as clusters of similar jobs. For this work other methodologies
such as job analysis or task analysis are required.
2. Some plausible statistical models for transfers are available and can
be used as a guide in evaluating and generating classifications and career
ladder orderings.
3. It is technically feasible to construct occupational categories so that
most transfers take place within relatively small groups and according to
career ladders. Computations for developing such a classification might
cost several hundred thousand dollars if the full set of 12,099 DOT titles
were used. New algorithms would have to be developed.
4. It is technically feasible to apply this kind of analysis to the job
history data currently gathered from Employment Service clients, since
these job histories are routinely assigned DOT codes.~°
5. Classifications based on observed transfers among occupations
suggest hypotheses about how the observed mobility has come about.
Independently generated data on task and skill similarities, and also on the
social characteristics of incumbents of occupations (e.g., age, sex, and
race), could be used in conjunction with data on mobility rates to further
our understanding of how people move among jobs.
6. Because some transfers may be excluded (or included) for reasons
other than those having to do with the transferability of skills, such
classifications should not be used uncritically. It is necessary to examine
the job content of the occupational categories suggested by the mobility-
based clusters in order to include any additional potential transitions and
in order to exclude absurd clusters created as artifacts of the statistical
algorithm.
ADVANTAGES AND DISADVANTAGES
The mobility approach to developing placement-oriented classifications
has several advantages. First, it can be developed from data already
collected by the Employment Service, which in its day-to-day operations
routinely collects work histories from applicants. For each local labor
market and for the specific clientele they deal with, the Employment
Service collects all the data needed to find out what occupational linkages
commonly occur. Second, the mobility approach allows for great flexibility
and continuous improvement. Since the underlying mobility matrices can
be continuously updated by using data from the ordinary operations of the
Employment Service, classifications for matching can be altered as labor
i°To be useful, these data would have to be preserved as a nine-digit occupational code. At
present, the third through ninth digits are discarded when the interview data are keypunched.
OCR for page 209
The Classification of Occupations
209
market conditions change. For example, by using regularly updated
transition matrices for a local labor market, the procedure could reflect the
fact that the opening of a new automobile assembly plant had created new
employment opportunities for workers formerly employed as coal miners.
Third, this method avoids the ad hoc judgments of program designers,
occupational analysts, or vocational counselors in deciding what are
similar and dissimilar occupations for the purpose of job referral; it relies
instead on the actual experiences of workers as they test various
alternatives in the labor market. Fourth, the approach would overcome
problems inherent in the overly narrow occupational classifications of the
DOT, since all occupations between which workers routinely transfer
would be grouped together.
The mobility approach, however, is not without its disadvantages. First,
as noted earlier, the resulting categories will reflect in part the current
practices of employers rather than the potential possibilities for transfer
inherent in the nature of transferable skills among occupations. Employers
may perpetuate, even unwittingly, discriminatory or stereotyped hiring
practices, or they may fail to perceive the potential of workers to move
into new occupations. To the extent that this occurs, the use of a
classification based on actual transitions will continue to perpetuate these
undesirable limitations on workers' employment opportunities. Second,
and analogous to the first disadvantage, the resulting classification will
reflect in part the current preferences and possibly limited horizons of job
seekers themselves. Such a classification might not expose workers to what
has not been tried before. Third, if the resultant groupings are based on
data generated by Employment Service activity, they will rehect in part the
practices of the Employment Service itself. When workers with particular
occupational histories are referred most often to job openings in certain
other occupations on the basis of currently used classificatory practices,
these patterns in referral practices will naturally tend to appear also in
data on placements. Fourth, the reliance on job histories to provide
mobility data may result in classifications that meet the needs of new
entrants and labor market reentrants inadequately. Fifth, regularities in
occupational transfers per se may tell us little about the desirability of the
transfers from the point of view of either the employer or the employee.
Placements may differ in terms of stability or tenure of employment,
productivity or performance, and employee satisfaction or employer
perceptions of satisfactoriness. Classifications based on mobility data may
group together, then, placements of differing usefulness; such clas-
sifications do not provide information on the likely quality of the matches,
though labor market information (such as job tenure) could be used to
supplement the classification.
OCR for page 210
210
WORK, JOBS, AND OCCUPATIONS
STRATEGIES FOR IMPROVING OCCUPATIONAL
CLASSIFICATION FOR JOB-WORKER MATCHING
These disadvantages suggest that the mobility approach to developing
placement classifications must be supplemented by other information.
First, and most crucial, a reliable set of occupational definitions is
necessary to provide the data base for mobility studies. Job analyses, then,
must continue to be the basic building materials of classification systems.
Second, because of current limitations on labor market mobility, addition-
al indicators of the transferability of skills must be developed in order to
encourage employers, workers, and the Employment Service to try new
types of matches. Again, job analysis approaches are appropriate, as are
vocational counseling approaches. Moreover, mobility patterns might be
studied to identify groups of occupations for which specialized approaches
such as task analysis would be particularly useful. Third, because without
supplemental information the mobility approach treats all matches as
being equally good, additional information about the quality of the
matches must be developed. There are two plausible approaches to
developing this information. The quality of the match could be inferred
from labor market data on, for example, the average job tenure of
particular types of matches (e.g., coal miners in steel mills), or the quality
of matches could be assessed by directly querying workers and employers.
Either approach could contribute to improving the quality as well as the
quantity of matches. Fourth, mobility data must be supplemented by
information about new entrants and returning workers. Direct skill and
ability assessment will continue to be useful in developing placement
possibilities, not only for those with limited labor market experience but
also for those workers who want to change careers.
OTHER METHODOLOGIES
Among the other alternative methodologies that may provide independent
assessments of occupational similarity, a prime candidate is task analysis
(including task inventories and position analysis questionnaires). Similar in
many respects to traditional job analysis, task analysis aims to describe
occupations in terms of the types of job tasks that are performed. It differs
from job analysis in both the explicitness of its attempt to assess the
similarity of occupations and its method of measurement. Task analysis
has been extensively used by the military services and to a lesser extent by
other government agencies such as the Public Health Service. By using
data rating the extent to which various jobs involve a common set of tasks,
it is possible to apply clustering and scaling procedures to construct a
OCR for page 211
The Classic cation of Occupations
211
simplified description of the similarity of these jobs. The resultant
description may be used to construct a classification of jobs in which
similarity is taken to mean similarity in task content. This procedure could
provide an alternative perspective to the mobility approach to the
similarity of occupations.
Worker characteristics can also be used to assess the similarity of
occupations. As we note above, classification systems developed by
Holland in his theory of careers or in the Minnesota theory of work
adjustment tap important dimensions of occupational similarity. More-
over, the techniques developed by vocational counselors to assess the
quality of matches from the point of view of both workers and employers
provide useful tools to assess the success of various classification schemes
in generating appropriate placements. These techniques are also useful in
providing knowledge of the skills, abilities, and aptitudes of workers that
supplement knowledge gained from job histories; they will thus be
particularly important for new entrants, reentrants, and those wishing to
explore different areas of work.
Classifications that are truly ideal for placement must make use of a
variety of approaches. Further research on developing classifications for
job-worker matching is particularly necessary along two lines. First, the
use of mobility data to indicate the transferability of skills and to locate
plausible job-worker matches should be investigated further. Second,
methods for assessing worker characteristics, such as skills, aptitudes, and
interests, and indicators of the adequacy of matches, such as satisfaction,
performance, and persistence, should be investigated.
A RESEARCH PROGRAM FOR DEVELOPING CLASSIFICATIONS
A research program intended to develop or improve classifications for
placement purposes might evaluate several aspects of the resulting
classifications:
1. What heuristic value do the classifications have for contributing to an
understanding of the transferability of skills, barriers to labor market
mobility, or the segmentation or Balkanization of labor markets in both
desirable or undesirable ways?
2. How successful are the classifications in generating satisfactory
placements? What proportion of job referrals made using a classification or
matching scheme results in placements (i.e., employer decisions to hire and
applicant decisions to accept employment)? How long do the placements
last? Do persons referred to jobs continue working at those jobs for an
acceptably long period of time? Put another way, do alternative matching
OCR for page 212
212
WORK, JOBS, AND OCCUPATIONS
procedures make any difference for the employment stability of users of
the Employment Service?
3. What are the long-range outcomes of placements made in terms of
the income, job satisfaction, and performance of the persons placed?
4. How easy is it for employment interviewers, applicants, and
employers to understand and use each system?
5. Are different classifications useful for different aspects of job-worker
matching?
CONCLUSION
In this chapter we have reviewed the classification structure of the DOT as
well as alternative bases for systems of occupational classification and have
raised some of the conceptual issues involved in developing classifications
for job-worker matching, in particular the notion of occupational
similarity and the transferability of skills. We have suggested the use of
data on naturally occurring patterns of labor mobility to evaluate, refine,
and develop new occupational classifications.
Our analyses lead us to conclude that mobility-based methods may
provide a flexible methodology for evaluating and developing classification
systems for use in placement. They have the unique advantage of using the
actual histories of workers in the labor force as guides for defining what
are appropriate (and inappropriate) matches to make for individuals with a
given occupational background. This method avoids ad hoc judgments and
permits greater flexibility than previous centralized, once-a-decade exer-
cises in occupational grouping.
Nonetheless, our work also indicates a clear need for more traditional
occupational analysis procedures. At a minimum, such procedures are
needed to define the basic occupational titles. There are, however, other
important reasons for shunning excessive reliance on mobility data in
making placement decisions. Any history of occupational mobility reflects
not only the potential range of the transferability of workers' skills
between various occupations but also the patterns of discrimination in
hiring and promotion that now exist (or previously existed) in the labor
market. So, for example, the fact that administrative secretaries do not
commonly advance into management occupations may reflect patterns of
sex discrimination in hiring and promotion rather than any inherent lack
of transferability of their skills. Any placement system guided exclusively
by the history of labor mobility between occupations would build the past
biases of the market into its future operations.
These considerations dictate that any mobility-based approach to
describing the similarity of occupations should be supplemented by other
OCR for page 213
The Classification of Occupations
213
methods that do not depend on the past functioning of the labor market.
An independent perspective on the similarity of occupations is required.
Traditional occupational analysis procedures might play this role, al-
though job analysis as currently practiced in the occupational analysis
program has not been especially successful in defining the similarities of
disparate occupations. In an approach that emphasizes required worker
characteristics, the most ambitious attack on this problem has been the
ratings measuring the complexity of a job in relation to data, people, and
things of the occupational analysis program's functional job analysis
approach. The validity of these ratings has not been studied systematically,
however, and their relationship to the potential transferability of workers
from one occupation to another remains to be shown. Moreover, as
chapter 7 indicates, the reliability of these measurements is questionable.
Any attempt to apply these particular measures as independent indicators
of occupational similarity should be grounded in future studies of their
criterion-related validity and ongoing quality control of their measure-
ment.
Other alternative methodologies that should be explored are task
analysis or other forms of structured job analysis and personjob matches
based on vocational preference theories such as that of Holland. The
integration of (1) task analysis data obtained from representative samples
of workers, (2) direct observation of jobs using more traditional job
analysis procedures and the judgments of trained analysts, and (3) the
assessment of workers' traits and personjob matches using techniques
developed by vocational counselors, with (4) study of the naturally
occurring patterns of labor mobility would provide a more adequate basis
for developing classification systems and operational procedures for use by
the Employment Service in matching jobs and workers.
Criterion validity could be demonstrated by showing the relationship, if any, between the
ratings of occupations on DATA, PEOPLE, and THINGS variables and the ease with which
workers transfer between jobs in these occupations
Representative terms from entire chapter:
labor market
