Printer Friendly

Implementing a computerized job-matching program with a hierarchical occupational structure.

Implementing a Computerized Job-Matching Program with a Hierarchical Occupational Structure

In vocational rehabilitation, the technology of job-matching for persons with disabilities based on "transfer of skills" is fairly well developed (Fine, 1957; Saxon & Deutsch, 1976; Saxon & Roberts, 1983; Watters, 1985). Recently, with the advent of microelectronic and computer technology, there is a proliferation of computerized job-matching programs developed specifically for the purposes of helping rehabilitation professional with this tedious, difficult, and time consuming job-matching task (Botterbusch, 1983). The majority of these Dictionary of Occupational Titles (DOT) based computer programs tend to have a strong emphasis on matching a person with a job on the basis of person-job requirements (P-JR) fit (i.e., physical capacities, skills, abilities, and aptitudes).

However, as noted by Holland (1985), person-work environment (P-WE) fit is also an important factor to consider in the job-matching process. This is because the dominant features of a work environment are believed to reflect the typical characteristics of its group members. Work environment can be analyzed in terms of: (a) consistency (providing similar rewards and demands); (b) differentiation (encourage a narrow vs. broad range of behavior); and (c) expected influence (vocational behavior, personal effectiveness, educational behavior, social behavior, and sensitization) (Holland, 1985). Research in vocational psychology has supported the premise that P-WE fit is critical to a person's long term vocational adjustment (Dawis, 1973; Holland, 1985). It can be inferred that in the computerized job-matching process, it is not sufficient to consider just how well a person meets the specific job requirements. It is equally important to consider systematically whether the person fits in with his or her work environment (Lam, Lustig, Chan, & Leahy, 1987). For instance, a disabled truck driver with a high school diploma may have the "transferrable job skills" to work as a clerk; however, his or her personality and work style are such that he or she may not fit in the white collar work culture. It is therefore essential to consider more than just transferability of skills.

A major reason the P-WE fit concept has been excluded in computerized job-matching software may be due to the difficulties of representing P-WE similarities among occupations systematically and efficiently in a software program. However, a hierarchical (tree) model proposed recently by Gati and his associates (Benyamini & Gati, 1987; Gati, 1979) for representing the structure of occupational interest fields appears to be quite appropriate for this kind of computer application. Reportedly, the similarity of relations among occupations can be represented by a hierarchical vocational structure better than Holland and Roe's circular models (Benyamini & Gati, 1987; Gati, 1979). Because many fast and efficient computer algorithms for building and searching tree-like data structure have already been developed (Horowitz & Sahni, 1976; Knuth, 1973), Gati's proposed hierarchical vocational model may be appropriate for implementing a computerized job-matching program that will consider both the P-WE fit and P-JR fit needs of the rehabilitation clients. The purpose of this paper is twofold: (1) to illustrate the concept of representing occupations with a tree-like data structure; and (2) to demonstrate how this hierarchical data structure can be used to conceptualize the development of a computer program that will systematically take into account both the concepts of P-JR and P-WE fit in job-matching. Hierarchical Occupational Structure

Data Source

Because the majority of vocational rehabilitation clients tend to find employment in the semi-skilled to skilled occupational fields (Dunn, 1974), as a demonstration, the set of 91 skilled occupations described in the Career Assessment Inventory (CAI) (Johannson, 1982) were used to represent the world of work for persons with disabilities in this study. The CAI is an interest inventory similar in design to the Strong Campbell Interest Inventory (SCII) but with an orientation toward the "nonprofessional" end of the world of work (Johannson, 1982). Like the SCII, the CAI was developed from the concept of P-WE fit as advanced in Holland's (1985) vocational theory. The CAI surveys a person's preferences for various job titles, work activities, leisure activities, school subjects, type of people with whom to associate, and sources of job satisfaction (Johannson, 1982). Information obtained from the CAI is used to help answer questions about likely P-WE fit in different occupations, as opposed to P-JR fit. The criterion (norm) groups representing 91 skilled occupations are used to help determine if an individual's expressed personal orientation and work environment needs are consistent with satisfied workers in these occupations.

The P-WE characteristics of the 91 skilled occupations in this study are defined by their mean CAI interest profiles based on six theme scales: Realistic (R), Investigate (I), Artistic (A), Social (S), Enterprising (E), and Conventional (C). These theme scales are patterned after Holland's theoretical vocational model. To represent the similarity relationships among these 91 skilled occupations in a hierarchical structure, their mean R-I-A-S-E-C profiles reported in the CAI manual (Johannson, 1982) were used as input and submitted to a hierarchical cluster analysis.

Cluster Analysis

The 91 CAI skilled occupations were clustered on the basis of their R-I-A-S-E-C profile similarity according to Ward's (1963) minimum variance method using the CLUSTAN package (Wishart, 1978). Squared euclidean distance was used as the measure of pairwise profile similarity in this research. This measure was recommended by Nunnally (1978) because it can account for accurately all essential profile information, i.e., elevation, scatter, and shape. A total of 91 x (91-1)/2 possible pairs of profile similarity indices were computed and the resulting similarity matrix was used as input for cluster analysis.

Ward's method combines clusters that will yield the least increase in the error sum of squares (the sum of the distance from each individual profile to the centroid of its parent cluster) and minimizes within cluster variance in effecting cluster mergers (Wishart, 1978). Initially, cluster analysis considers every occupation as a distinct cluster. In each subsequent partition level, two related clusters existing at the previous level are fused together until in the end all occupational profiles are merged into one all-inclusive cluster. As a result, a complete sequence of partitions, i.e., a hierarchy of occupation groupings, was produced (Berven & Hubert, 1977). The task of determining the optimal partition (the level which is the best compromise between parsimony and homogeity) from the sequence of partition levels was accomplished by the scree method (Cattell, 1968). By inspecting the plot of the changes in the error sum of squares from one partition to the next, the partition formed before a significant change in homogeneity was chosen as the appropriate clustering solution.

In the present study, a six-cluster solution was considered optimal for the hierarchy of occupation groupings. The tree representation of these six clusters of occupations is depicted in Figure 1.

As an example, the tree-representation of the skilled occupations in Cluster-1 is also depicted in Figure 2. The grouping of occupations in other clusters will be described in the text.

Cluster-1 consisted of 22 skilled occupations and was represented by occupations such as Aircraft Mechanic, Surveyor, Drafter, Electronic Technician, Machinist, and Tool/Die Maker. Based on predominant CAI designated summary Holland Code, this cluster was labelled accordingly as the Realistic-Technical cluster. Cluster-2 consisted of 15 occupations. Examples of occupations that were grouped into this cluster include Security Guard, Hardware Manager, Mail Carrier, and Purchasing Agent. This cluster was labelled as Realistic-Conventional. Cluster-3 was labelled Investigative-Service and included 12 occupations such as Computer Programmer, Registered Nurse, Radiologic Technician, and Respiratory Therapy Technician. Cluster-4 was labelled Enterprising-Service and included 25 occupations such as Restaurant Manager, Hotel Manager, Executive Housekeeper, Buyer/Merchandiser, and Reservation Agent. Cluster-5 consisted of 7 predominately arts related occupations including Author, Photographer, Interior Designer, Musician, etc. Finally, Cluster-6 was represented by 9 conventional occupations such as Nurse Aide, Cafeteria Worker, Bookkeeper, Secretary, and Teacher's Aide. As can be observed, the similarity relationships among these skilled occupations can be meaningfully represented in a tree-like data structure. Development of a Rudimentary Computerized Job-Matching Program

The cluster analysis results in this study demonstrated how the P-WE similarity relationships among various skilled occupations can be represented in a tree-like data structure. The next question is how this model and data structure can be applied in a computerized job-matching program. For illustrative purposes, a rudimentary computerized job-matching program is presented.

As a demonstration, the computer program will depict only Cluster 1 (Realistic-Technical) as its data structure (see Figure 2). The precise hierarchical relationships among the 22 specific occupations, their CAI R-I-A-S-E-C profiles, DOT physical demand ratings, DOT defined skill requirements ("tool knowledge": mathematical [M] and language [L] development), and the schematic diagram of the computer program is presented in Figure 3.

The design goal of this computer program is to help the vocational rehabilitation professional identify potential job areas that are compatible with persons with disabilities in terms of residual physical capacity, P-WE fit, and P-JR fit. Using the structure presented in Figure 3, an efficient job matching scheme can be developed.

To demonstrate how this job-matching program may work, a disabled firefighter (DOT #373.364-000) whose physical capacity has been reduced to a rating of "L" (Light: Maximum Lift = 20 lbs., Frequent Lift/Carry = Up to 10 lbs.) would be used as a hypothetical client. The R-I-A-S-E-C profile scores for firefighters are reported in the CAI manual as follows: R = 58, I = 57, A = 47, S = 45, E = 46, and C = 48 (M = 50, SD = 10). According to the DOT (U.S. Department of Labor, 1981), the physical demands factor for a working firefighter is "V", very heavy; a person on this job must be able to lift over 100 lbs and up to 50 lbs on a frequent basis. The "tool knowledge" requirements for a firefighter has a rating of 2 for mathematical development and 3 for language development. According to the DOT definition for mathematical and language development, a person on this job would need to know how to add, subtract, multiply, and divide all units of measures, possess a passive vocabulary of 5,000-6,000 words, and read at rate of 190-215 words per minute (U.S. Department of Labor, 1981).

We assumed our hypothetical client was basically satisfied with, and had performed to criteria on, his prior job. To help identify potential job change areas, the DOT code for Firefighter (373.364-000) would be entered into the computer. The code would be compared to a sorted (ascending order) array (see Figure 3) of 22 DOT codes using a binary search method (Harowitz & Sahni, 1976). Once the matching DOT number was found, the computer would follow the appropriate computer pointer (arrow) embedded in the array to the tree node containing Firefighter (see Figure 3). In this example, job attributes such as the physical demands and "tool knowledge" factors were used to represent the specific skill requirements on a job (in reality many more job attributes can be used). To find a job alternative that has a P-WE fit and P-JR fit requirements similar to a Firefighter, the computer would exit from the Firefighter node and traverse to the tree node on its left which contains the occupation Drafter. As can be observed, Drafter has a physical demands strength factor of "S" (Sedentary: Maximum Lift = 10 lbs.) and a R-I-A-S-E-C profile that is identical to Firefighter (see Figure 3). Based on the CAI information, Firefighter and Drafter may have similar P-WE fit factors in the work environment. This job is also compatible to the current physical capacity rating ("L") of our hypothetical client. However, the drafter has a higher "tool knowledge" requirements (M = 5, L = 5) (ability to do higher order mathematics, e.g., algebra, calculus, and statistics and to read and write at the advanced language level, e.g., novels, plays, editorial, technical journals, manuals, critiques, poetry, and songs) than the firefighter (M = 2, L = 3). The drafter position offers a very compatible alternative work environment for the client in terms of personal orientation and physical demands but it is less compatible in terms of job skill development.

If the counselor or client desired to know more about this job, the computer can retrieve all pertinent vocational information (training, time, employment outlook, wages and salary, etc.) from the appropriate occupational information file and display the information on the screen or produce a hard copy on the printer. If this is not an acceptable match, the computer would exit the tree node for Drafter and traverse up one level to the right and enter the tree node containing information for Electronic Technician. The sequential steps for traversing the tree is depicted in Figure 3. Of course, the more levels the computer travels up the tree, the less compatible (in terms of P-WE fit) the identified occupation would be with the Firefighter occupation. For example, the correlation between the Firefighter's R-I-A-S-E-C profile and other occupations in Cluster 1 ranged from .47 to 1.0 with a median of .79. The client must make some compromises based on P-WE fit and P-JR fit in the vocational decision making process.

After traversing all the tree nodes in Cluster 1 (Realistic-Technical 8 occupations (Forest Ranger, Park Ranger, Camera Repairman, Surveyor, Drafter, Electronic Technician, Radio/TV Repairman, Dental Lab. Technician, and Musical Instrument Repairman) were found to be compatible with the physical capacity rating ("L") and P-WE fit (R-I-A-S-E-C profile similarity) of our disabled firefighter with a varying degree of P-JR ("tool knowledge") fit. Presumably, many of these occupations identified can be potentially satisfying job alternatives for our hypothetical client in this example. The major advantage of this computer approach, therefore, is its ability to consider simultaneously job-matching both in terms of transferability of skill requirements as well as work environments. Implications

This paper demonstrated a hierarchical data model can be effective for representing the P-WE similarity relationships among a group of skilled occupations in a computer. Although 91 skilled occupations were used in this study, theoretically this model can be extended to represent any number of occupations. The hierarchical occupational structure can also be used to conceptualize the implementation of a computerized job-matching program that is consistent with traditional vocational adjustment theories.

Because the empirical relationships among occupations are reflected in the data structure and not in the computer program, this arrangement has major advantages from the computer system development perspective. That is, the expansion and re-alignment of the occupational structure (based on further vocational research, refined vocational theory, more precise measurements, and a better understanding of the relationships among different occupations) can be accomplished easily at the data level without having to rewrite the search/match routines of the computer program. This would extend the lifespan of a computer software and reduce the cost of software maintenance and enhancement.

Finally, this hierarchical structure and computer program provide an alternative means for job-matching based on already known P-WE relationships. Job alternatives so selected may be more appropriate for people with disabilities in terms of consistency with their personal orientation, preferred work environment, and skill development.

Table : Example of the Search Pattern of a Rudimentary Computer Job-Matching Program
COPYRIGHT 1989 National Rehabilitation Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1989, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

 
Article Details
Printer friendly Cite/link Email Feedback
Author:Chan, Louis Yau-Kwong
Publication:The Journal of Rehabilitation
Date:Apr 1, 1989
Words:2487
Previous Article:Levels of compliance with federal requirements in independent living centers.
Next Article:Benefit-cost analysis of transitional employment programs.
Topics:


Related Articles
Determining vocational levels of people with disabilities in Japan: A statistical prediction approach.
Toward the development of a second generation computerized job-matching system for persons with disabilities: a conceptual framework.
JVS: a career development model for community rehabilitation programs.
Analysis of the Types of Natural Supports Utilized During Job Placement and Development.
Healthy Talk.
From vocational decision making to career building: blueprint, real games, and school counseling. (Special issues: career development and the...
General job satisfaction from a developmental perspective: exploring choice--job matches at two career stages.
Factors affecting the job satisfaction of employed adults with multiple sclerosis.
The role and function of certified vocational evaluation specialists.
INCOME: a culturally inclusive and disability-sensitive framework for organizing career development concepts and interventions.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters