TOWARD A FRAMEWORK FOR STRUCTURED JOB-COLLABORATION DESIGNVitaly Dubrovsky Clarkson University Potsdam, NY 13699 Anatoly Piscoppel Moscow University Moscow, USSR In: Proceedings of 35th Human Factors Society Annual Meeting, (969-973). San Francisco, CA: Human Factors Society, 1991.
This paper introduces the idea that job design and design of collaboration are inseparable. On this basis, job design is reformulated as job-collaboration design. Combining activity theory and systems approach, the paper introduces several functional-structural models and interprets conventional job design decisions in terms of the modeling operations. It describes a sequence of stages for the first phase of structural job-collaboration design, which uses the introduced models and results in a design of the ideal job-collaboration network, which is supposed to function strictly according to the standards. Structured methodology for designing social support mechanisms, which compensate deviations from the standards experienced by real organizations is on our future agenda.
According to the conventional definition, job design determines (1) what tasks are to be performed, (2) what tasks to be assigned to what jobs, and (3) how the jobs are to be linked together (Davis and Wacker, 1982). We believe that what tasks to be assigned to what jobs depends on the collaboration context, or linkage between tasks, and, at the same time, collaboration relationships depend on how the tasks distributed among the jobs.
In other words, job design and design of collaboration are inseparable. Job design should be reformulated as job-collaboration design.
This paper introduces several functional-structural models which we believe will: (1) contribute to a theoretical framework for job-collaboration design, (2) address important problems in this area, and (3) provide bases for structured job-collaboration design decisions and for evaluation of the design by means of computer simulation. Being functional, these models represent job-collaboration design in the broadest terms possible, thus permitting integrated approach to heterogeneous socio-technical reality of jobs and collaboration (Singleton, 1974). Functional definitions also represent the problems in the most abstract form, thus providing the best chance for innovative solutions (Bailey, 1982). Being structural, these models can serve as simple construction units, which can be assembled into complex models, in order to represent the reality (Shchedrovitsky, 1964).
The Activity Theory Approach
Developing the framework, we combined the "activity theory" and systems approaches to job-collaboration design. The activity theory assumes that human actions are not determined by natural laws, but primarily determined by standards, e.g. societal, cultural, and social norms, industrial codes, organizational rules, professional paradigms, technical specifications, instrumental scripts, and alike (Shchedrovitsky, 1975). Even deviations from the standards are standardized, i.e., described, categorized, and the appropriate sanctions or compensatory actions are determined (Dubrovsky, 1988).
Following the systems approach (Emery, 1969; Shchedrovitsky, 1975; Davis and Taylor, 1979; Davis, 1982), we assumed that a socio-technical system constitutes the context for job-collaboration design, and all design decisions should be considered in at least three dimensions: productive, organizational, and individual. Activity theory adds another systems dimension: self-reproduction of society and its units (Shchedrovitsky, 1975). In this dimension, a socio-technical system and its units should be designed as a self-maintained networks of collaborating jobs.
A TASK AND A PRODUCTION ACT
Tasks and standards. The definition of a task as a normative representation of worker's activity and its outcome can be traced back up to Frederick Taylor (1911). Each task must be viewed as a component of at least three structures, which impose their requirements and constraints on both, the task performance and its outcomes, or products. First, task corresponds to a certain function in the mission of the organization (Bailey, 1982) and thus is subdued to the requirements dictated by the environment, e.g., cultural, industrial, and organizational standards, customer's technical specifications etc. (Verman, 1973). Second, each task is performed by an individual or a group and the products are usually used by other individuals or groups in their task performance. In this way, the task is included in the collaborative network which sets its own requirements and standards, usually internal to the organization, e.g., report formats, procedures, and coordinating schedules. Finally, groups of tasks assigned to a certain organizational position constitute a job and are subjects to such constraints as time, workload, and alike, which are usually specified in the job definition. All these different standards are key determinants of tasks. Our first?cut definition of a task is the smallest unit of worker's activity, which starts with a standard material and results in a standard outcome, or product.
Task and its subtasks. It is a common view that tasks can be decomposed into smaller tasks or can be combined into larger tasks. The above definition prohibits such a view. The definition requires both, standard material and standard product. If a task is decomposed into subtasks, a subtask either does not result in the standard product, but in an "interim product", or it does not start with the standard initial material, but with the "interim material", or both (Figure 1). For example, preparing this paper included such subtasks as writing the text with a word processor, formatting the text according to the written specifications, saving it as a file under a certain name, printing a hard copy, cutting and pasting on the plates (initial material) according to the layout, and sending the finished plates to the publisher along with the copy right agreement and hardcopies of transparencies (the product). Suppose our secretary agreed to help us, so we need only to write the text, sign the agreement, and type the content of the transparencies. Division of the initial task into two parts is not enough to create two separate tasks. To transform the parts of the work into tasks we: (1) must standardize the interim product/material, e.g. a file in the ASCII format on a floppy disk (initial material requested by the secretary) and (2) to comply with the standard, we have to perform additional "interface" subtask: save our Word Perfect file as a DOS file on a floppy.
Task and variance control. Such "absolute" distinction between task and its subtasks has another important meaning. In real work, task performance experiences deviations or disturbances in the throughput states which are called "variance" (Herbst, 1974). The idea of a task as an additive unit of human performance implies that all variance is handled by an actor during performance of subtasks, so the resulting outcome will correspond to the standards, or specifications. In our example, we decided that we will format our paper as well. Since we wanted our paper to be approximately five page long (which could be found only after formatting), it was more efficient to make necessary changes before giving the text to the secretary to avoid having it repeatedly returned from her. In other words, it proved to be more efficient to handle the variance by ourselves.
Figure 1. Task-subtask relationship.
Practical implications. According to the task/subtask distinction, any re-distribution of work requires the appropriate changes in the standards along with addition or deletion of the "interface" subtasks.
In the case of job simplification (Taylor, 1911), when a complex task is divided between several workers, additional specifications for each interim product/material must be introduced and the additional "interface" subtasks must be defined to meet the new specifications. The above mentioned variance puts limitations on task simplification and sometimes even makes job enlargement necessary.
By the same token, certain specifications must be removed from the tasks to be assembled in a larger task in the course of job enlargement (e.g., Conant and Kilbridge, 1965) and/or enrichment (Herzberg, 1966). "Minimal critical specification" is the ideal of this approach (Herbst, 1974). Also the appropriate "interface" subtasks must be eliminated.
In modern computer-supported jobs, both task simplification and task enlargement may be required in the ongoing process of job redesign (Fitter, 1982; Ostberg & Chapman, 1988). In this case, it is especially important to take care of specifications and "interface" subtasks.
Representation of a Task as a Production Act
F. Taylor (1911) stated that task specifies what should be done (object) and how it should be done (action). This "action-object" representation of a task is commonly used by analysts and designers (e.g., Shneiderman, 1987). Some authors add other components to a task: user mental models, tools (Norman, 1986), and skills (Bailey, 1982). Texts in Industrial Engineering and Ergonomics include production process, facility layout, environment, and incentives in the job design considerations.
We will represent a task as a production act. The activity theory defines a production act as a unit of societal self-reproduction (Shchedrovitsky, 1975). As such, a production act reproduces a standardized action and results in a standard product. The structure of a production act is constituted by components, that are necessary for the successful action and potentially include all components mentioned in task analysis and job design literature (Figure 2).
Support functions. Each time when the act is reproduced or repeated, all its components must be in place in the "ready" form. This readiness is achieved by performance of support functions (supply, maintenance, and disposal) in relation to each component of the act. These functions can be performed as simple subtasks (e.g. adding paper to the tray of a copier by the operator), as special production acts (repairing a copier by a repairman), a specialized organizational unit or even a service organization (installation of a new copier).
Coordination functions. Above, we placed control of variance inside a task. We assumed that the discrepancies, which occurred during performance of earlier subtask (if any), are handled during performance of subsequent subtasks. This requires coordination of the subtasks. Below we will discuss other coordination functions.
It must be noticed that the set of act's components should not be viewed as complete, but rather as "opened construction set". Other kinds of components can be added as needed. We suggest that, for the sake of simplicity, only those components should be depicted that are objects of support or coordination tasks or subtasks. For example, if one should take into consideration the incentives activities, a "motives" component must be added to an act.
Figure 2. A production act.
At these level of abstraction, a production act (Figure 2) can represent a simple task as well as activities of entire organization. To make the representation more specific, we will define several modeling operations upon a production act and will provide interpretation for the operations in the job design terms.
Structural Decomposition of an Act
"Structural decomposition" is a modeling operation, which decomposes a whole entity into parts, and at the same time, links the parts together, so the wholeness is retained (Shchedrovitsky, 1964). In the process of job-collaboration design, structural decomposition divides an act, which represents the activity (mission) of the entire organization, into a number of acts (tasks) linked into a collaboration network. The integrity of the structurally decomposed organization is retained: it remains to be a production act.
We would like to stress that structural decomposition of an act differs from "linking" a number of acts into collaboration network. A main advantage of the structural decomposition is that the completeness and integrity of job-collaboration design are always guaranteed, since the support and coordination are functional constituents of the model which are respectively responsible for self-maintenance of the organization and stability of its functioning (Figure 3).
Figure 3 depicts a simple collaboration network which is obtained by structural decomposition of an act into two acts connected by a collaboration link. A complex collaboration network can be obtained by a sequence of structural decompositions corresponding to hierarchical divisions of the organization, down to the tasks of the smallest self-maintained organizational units (Davis, 1982).
Figure 3. A unit of collaboration network.
Types of Collaboration
A collaboration link means that a product of one act is used as a component of another act. The first act functions as "production", the second act functions as "usage". There are three types of collaboration: production, when the product of one act is used as a material of another act; support (e.g., when a product of one act is used as a tool in another act), and coordination (e.g., a production schedule as a result of managerial decision, becomes a goal requirement for the manufacturing act). Collaboration links can be specified according to the components of the usage act. For example, support links (supply, maintenance, and disposal) in relation of an actor's skills can be specified as training, development training, and retraining.
Organization as a Self-maintained Collaboration Network
The activity theory approach requires that a cooperative network must be self-maintained. To meet this requirement, a collaboration network has to provide support (supply, maintenance, and disposal) for components of all acts included in the network. The unfolding of the collaboration network should not stop until each of the support activities is defined. Short term operational maintenance should be defined as a subtask (e.g., change the ribbon in the typewriter). Middle term maintenance (repair and modification) should be defined as special production acts. Long term maintenance functions -- supply and disposal should be defined as collaborating acts (e.g. purchase, delivery, and installation)
Types of Coordination
"Variance control" and "prerequisite" management. We borrowed the concept of coordination and most of the terminology from Malone and Crowston (1990). They define coordination as "the act of managing interdependencies between activities to achieve a goal" (Malone and Crowston, 1990, p. 361). The first type of coordination discussed above, we called "variance control", where the way the discrepancy is handled
depends on the occurrence of the discrepancy and its nature. The second type of coordination deals with the "prerequisite" interdependency, which is typical for production and support collaboration links (e.g. equipment must be repaired on time to be used). Two more types of coordination correspond to the interdependencies derived from two modeling operations defined upon components of production acts.
Management of "sharing". One or more elements of two or more production acts can be identified. This modeling operation means that two (or more) acts "share" the same components, e.g., the equipment, the budget, etc. There are two kinds of sharing interdependencies. The first kind is "sharing" indivisible object, such as a tool, or a machine. Time coordination, (e.g., scheduling and shifting) permits this kind of sharing. The second kind is "sharing" parts of an object, such as budget money or office space. Corresponding coordination decisions are distribution and assignment.
Coordination of concurrent activities. Acts of production usually include more than one element of the same type: actors, tools, pieces of machinery, etc. This fact can be represented in the production act model by "element multiplication". Special kind of interdependency derived from the element multiplication is called "simultaneity". For example, group decisions made face-to-face and flight control at the airport require coordination of concurrent activities. Identification of multiple components creates a mixed kind of "sharing", such as usage of a set of tools or multiple machinery in more than one act. Temporary allocation is the corresponding coordination activity.
According to a commonly accepted definition, a job is a cluster of tasks assigned to an organizational position. This assignment can be represented as a modeling operation of identification of the actors of several production acts (with assignment of a position title).
The Structure of a Job
Work module. While a task is the basic element of a job, a "work module", or combination of several tasks is the basic unit of a job (Bailey, 1982). Typical faculty job includes "teaching", "academic advising", "research", and "services" work modules. Teaching work module may include such tasks as preparation and teaching of scheduled classes; preparation, proctoring, and grading of tests and examinations; preparation of course outlines; and others. Different cohesiveness principles can be used to combine tasks in the work module: organizational function, skill level, common data, sequence, common tools, etc. For example, the faculty work modules correspond to the organizational functions, or parts of the university mission. Usually jobs include from one to four work modules and from four to nine tasks per module (Bailey, 1982).
Work submodule. Sometimes it is useful to divide a work module into several "work submodules". For example, a typical college faculty job includes research as a work module. This module, in turn, may include management of a research budget account, which requires performance of budget preparation, rebudgeting, purchase, reallocation of funds and charges, and account closure tasks. We suggest that these tasks should be combined into a separate work submodule, since they are apparently more cohesive with each other than with other research tasks. In some large university departments, management of research accounts is assigned to administrative assistant instead of faculty. In case of the appropriate reorganization, the entire work submodule should be reassigned to another position. A structural model of a job is depicted in Figure 4.
Figure 4. The structure of a job.
A Job in The Collaboration Network
A job and production acts. Assignments of tasks to jobs should be done with consideration of the collaboration network context, so all interdependencies among production acts will be taken into consideration. This can substantially simplify the coordination tasks. A job can be represented in the collaboration network by assembling of the appropriate production acts along with their collaboration links. All components of the acts should be combined and grouped appropriately, e.g., goals, knowledge, and skills. The middle part of an act (between "actor" and "support") should be repeated as many times as many tasks are assigned to the job and in groups and subgroups, which correspond to respective work modules and submodules. All actions should have names of the respective tasks. The resulting structure should have the name of the appropriate organizational position.
A network of collaborating jobs. Combination of production acts into jobs along with their collaboration links dramatically changes the topology of the collaboration network. The resulting collaboration network has jobs as its nodes (instead of production acts) with sometimes multiple collaboration links between two jobs. In the research budget management example, all budget tasks are resulting in sending different documents (budget, rebudget memo, completed acquisition forms, etc.) to the Finance Office. We suggest that in such cases, for the sake of simplicity, only one link should be shown on the model and the names of all products should be associated with the link. This job collaboration network represents a first-cut job-collaboration design, or a design, which made under unrealistic assumption, that all products exactly correspond to the standards.
Based on the idea that job design and design of collaboration are inseparable, we reformulated job design task as job-collaboration design. We introduced several functional-structural models and interpreted conventional job design decisions in terms of the modeling operations. Such interpretation has a potential for development of structured methodology for job-collaboration design. We suggested that job-collaboration design should start with an abstract representation of the entire organizational activity as a simple production act. At the first stage of the design, a sequence of structural divisions of the production act should be performed according to the hierarchy of divisions in the organizational design. The result of this stage is a network of collaborating production acts with the requirements for support and coordination activities. At the second stage, the supportive acts are defined, and the resulting network becomes a self-maintained one. At the same time, all interdependencies between tasks are determined and the appropriate coordination acts are defined. The result of this stage is a network of collaborating production, support, and coordination acts. At the third stage, the task are allocated between organizational positions with consideration of the interdependencies. On the basis of the task/subtask distinction, we formulated principles for task composition-decomposition, which usually are performed at this stage of job-collaboration design. Resulting is a design of an ideal job-collaboration network, which is supposed to function strictly according to the standards.
Real organizations experience all kinds of deviations from the standards. Such deviations are compensated by social mechanisms supporting collaboration (e.g., Dubrovsky, 1987). Structured methodology for designing social support mechanisms for job-collaboration networks is on our future research agenda.
Bailey, R.W. (1982). Human Performance Engineering: A Guide For System Designers. Englewood Cliffs, NJ: Prentice-Hall.
Conant, E.H. and Kilbridge, M.D. (1965). An interdisciplinary analysis of job enlargement: technology, cost, and behavioral implications. Industrial and Labor Relations Review, Vol. 18, pp. 377-390.
Davis, L.E. (1982). Organizational Design. In G. Salvendy (Ed.), Handbook of Industrial Engineering. New York: Wiley & Sons, pp. 2.1.1-2.1.29.
Davis, L.E. and Taylor, J.C. (1979). Technology and job design. In L.E. Davis & J.C. Taylor (Eds.) Design of Jobs (2nd edition), Santa Monica, CA: Goodyear.
Davis, L.E. and Wacker, G.J. Job design. (1982). In G. Salvendy (Ed.), Handbook of Industrial Engineering. New York: Wiley & Sons, pp. 2.5.1 - 2.3.31.
Dubrovsky, V.J. (1987). Social exchange in group consensus development: Face-to-face versus electronic mail. In Proceedings of 31st Human Factors Society Annual Meeting, New York, NY, pp. 701?705.
Emery, F.E. (Ed.). (1969). System Thinking. London: Penguin.
Fitter, M.J. (1982). Information systems and organizational implications of job redesign. In J.E. Kelley & C.W. Clegg (Eds.), Autonomy and Control a the Workplace: Context for Job Redesign. London: Croom Helm.
Herbst, P.G. (1974). Socio-Technical Design. London: Tavistock.
Herzberg, F. (1966). Work and the Nature of Man. Cleaveland: World.
Malone, T.W. and Crowston, K. (1990). What is coordination theory and how can it help design cooperative work systems? In Proceedings of the Conference on Computer-Supported Cooperative Work, Los Angeles, CA, pp. 357-370.
Norman, D.A. (1986). Cognitive engineering. In D.A. Norman and S.W. Draper (Eds.) User Centered System Design. Hillsdale, NJ: Lawrence Erlbaum.
Ostberg, O. and Chapman, L.J. (1988). Social Aspects of Computer Use. In M. Helander (Ed.), Handbook of Human Computer Interaction. Amsterdam: North-Holland, pp. 1033-1049.
Shchedrovitsky, G.P. (1964). The Philosophy of Systems Research. Moscow, USSR: Znaniye.
Shchedrovitsky, G.P. (1975). Fundamental concepts and categories of the activity theory. In B.V. Sazonov (Ed.), Development of Automated System Design: Theory and Methodology. Moscow: Stroyizdat, pp. 72?161.
Shneiderman, B. (1987). Designing The User Interface: Strategies for Effective Human-Computer Interaction. Reading, MA: Addison-Wesley.
Singleton, W.T. (1974). Man-Machine Systems. Middlesex, England: John Wiley & Sons.
Taylor F.W. (1911). The Principles of Scientific Management. New York: Harper & Row.
Verman, L.C. (1973). Standardization, Connecticut: Archon Book.