The effects of structured development methods on the job satisfaction of programmer/analysts : a theoretical model online

perceived by programmer/analysts . It will limit the options available to
the programmer/analyst in systems development.

Page 15



H5 (B) : Skill variety will be positively related to job satisfaction .

H6 (a) : The use of structured methods will reduce the task identity
perceived by programmer/analysts . The development task will be split into
smaller tasks when structured methods are used.

H6 (B) : Task identity will be positively related to job satisfaction .

H7 (A) : The use of structured methods will limit the level of autonomy
perceived by programmer/analysts . Using structured methods will allow
project managers to better specify the tasks to be performed by
programmer/analysts.

H7 (B) : Autonomy will be positively related to job satisfaction .

H8 (a) : The introduction of structured methods will improve the
productivity of programmer/analysts and it will improve the quality of their
work . This is based on the previous research on the impact of structured
methods on productivity and quality.

H8 (B) : Changes in productivity and quality will not have a significant
effect on job satisfaction . The research with other types of jobs has shown
little relationship between job satisfaction and performance (Locke, 1976) •
There is some evidence that there is a relationship between current
performance and future satisfaction, but not between current job
satisfaction and current performance (Wanous, 197'«) • Locke claims that
individual differences, measurement problems, and contextual differences
play a major role in moderating the relationship between performance and job
satisfaction. Controlling for these factors is out of the scope of this
study.



Page 16



Environmental Variables

The hypotheses discussed above can be tested with a quas i -exper iment .
That is, we can examine differences in the dependent variable and in the
intervening variables between a treatment group - a group of
programmer/analysts using structured methods - and a control group - a group
of programmer/analysts not using structured methods. However, in a
quasi -exper iment we cannot control for differences between the treatment and
control group or for peculiarities of a particular research site. We can
attempt to measure as many of these factors, which we will call
environmental variables, as possible. This will allow us to determine the
degree of validity of the quasi -exper iment . In this section, we will
consider three types of environmental variables - characteristics of
individual programmer/analysts, characteristics of project teams, and
characteristics of the research site.

We will assume the quas i -exper iment is a pretest-posttest with control
group design. That is, we will be concerned with changes in job
satisfaction between the pretest and posttest and with differences in the
amount of the changes between the treatment and control groups.

The environmental variables can threaten both the internal and external
validity of the quasi -exper iment . For example, differences between the
programmer/analysts in the treatment and control groups could falsely lead
us to attribute a change in job satisfaction to the introduction of
structured methods - a threat to internal validity. In addition, if there
were significant differences between the programmer/analysts at the research
site and programmer/analysts at other organizations, the general i zab i 1 i ty of
the quasi -exper iment would be affected - a threat to external validity. In
this section we will rely on Cook and Campbell's (1979) classification of

Page 17



threats to val idi ty .

Characteristics of programmer/analysts - Differences in several
characteristics of programmer/analysts in the treatment and control groups
could affect both internal and external validity. This includes background
variables, such as job longevity, age, amount of systems development
experience, tenure in the project group, amount of experience with
structured methods, and level of motivation and satisfaction with job
contexts .

For example, differences in job longevity could threaten the internal
validity of the quas i -exper iment . Suppose the treatment group contained a
much larger percentage of new hires than the control group. We would expect
that the job satisfaction of the new hires would decline after six months
when the novelty of their new job wears off. This change in job
satisfaction would be independent of the introduction of structured methods.
We could, however, falsely attribute this change to the introduction of
structured methods. Cook and Campbell would classify this as a threat to
internal validity due to the interaction of selection and maturation.

Background differences between the treatment and control groups could
also threaten external validity. For example, if the treatment group was
older or more experienced in systems development or more experienced with
structured methods than the control group, we could not be sure that the
results of the quas i -exper iment could be generalized to other treatment
groups containing less experienced programmer/analysts. Further, Katz
(1977) has shown that age and job longevity differences moderate the effects
of job characteristics on job satisfaction. If most of the members of the
treatment group were working at the company for at least ten years, it would
be difficult to generalize the results to organizations containing less

Page I8



experienced programmer/analysts. Cook and Campbell would classify this as a
threat due to the interaction of selection and treatment.

Other individual differences, such as level of motivation and
satisfaction with job contexts, could also affect the general i zabi 1 i ty of
the quasi -exper iment. Hackman and Oldham (I98O) identify these as
constructs that moderate the effects of job characteristics on job
satisfaction. Suppose the treatment group was significantly more motivated
than the control group. We would then be uncertain that any differences in
the level of job satisfaction between the groups were due to the use of
structured methods or to the interaction of structured methods with
motivation.

Project team characteristics - Differences in project teams could

affect external validity. One aspect of team differences is the leadership

style of the project leader and of the programmer/analyst's peers. Bowers

and Seashore (I966) have identified four leadership characteristics:

Support - behavior that enhances someone else's feeling of personal
worth and importance.

Interaction Facilitation - behavior that encourages members of the
group to develop close, mutually satisfying relationships.

Goal Emphasis - behavior that stimulates an enthusiasm for meeting the
groups goal or achieving excellent performance.

Work Faci 1 i tation - behavior that helps achieve goal attainment by such
activities as scheduling, coordinating, planning, and by providing
resources such as tools, materials, and technical knowledge (Bowers and
Seashore, page 2*47) .

These characteristics can be applied to both the project leaders and other

members of the project team. Bowers and Seashore found significant positive

correlations between measures of these characteristics and job satisfaction

and performance. Yunger and Hunt (1976) found these characteristics similar

to the characteristics identified in the Ohio State LBDQ leadership scales.

Page 19



In addition, differences in the background of the project leaders and
differences in the amount of time the team has worlced together - team
longevity - could affect could affect external validity. Katz (1979) has
found that group longevity had a significant effect on performance in
research and development groups.

Project team differences between the treatment and control groups would
have the same effect on external validity as individual differences. We
would be uncertain whether to attribute changes in the dependent variable to
the independent variable or, to the interaction of the independent variable
and the project team differences.

Research site characteristics - The characteristics of the research
site could affect both internal and external validity. For example, the
target organization could select only the most satisfied programmer/analysts
to be used in the treatment group. If this were the case, we would expect
the satisfaction of this group to regress to the mean between the pretest
and the posttest - independent of the introduction of structured methods.
We could then falsely attribute their change in job satisfaction to the use
of structured methods. Cook and Campbell call this a threat to internal
validity due to statistical regression.

Differences in how structured methods were used at the research site
could also threaten the validity of the quas i -exper iment. If the
programmer/analysts in the control group determined that structured methods
were a valuable tool, they could start imitating the treatment group by
using some of the structured techniques. The differences in job
satisfaction between the treatment and control group could then be moderated
by the imitation of the treatment. A similar problem would occur if the
treatment group did not become proficient in the use of structured methods

Page 20



between the pretest and the posttest.

Organizational characteristics of the research site could also affect
external validity. For example, if the d. p. organization has had a number
of changes in top management in the last year, if they have introduced a new
compensation scheme, or if they have had a large number of new hires or
layoffs this could affect programmer/analyst's reaction to the introduction
of structured methods. This would limit the general i zabi 1 i ty of the
quas i -exper iment to other organizations.

QUASI-EXPERIMENTAL DESIGN

The model described above can be tested using a pretest-posttest with

control group design. Cook and Campbell (1979) diagram this design as:

0X0


The design can be implemented by finding one company that is about to

introduce structured methods to some, but not all, of its

programmer/analysts. We can then administer a questionnaire to all the

programmer/analysts in the company right before structured methods are

introduced to measure the independent, intervening, and environmental

variables (the pretest). The questionnaire can then be readministered six

months later (the posttest). The group of programmer/analysts who do not

use structured methods can serve as a control group at the pretest and

posttest. This section outlines the quas i -exper imental design. It

discusses the requirements of the research site, the experimental procedure,

the measures, and the method of analysis. Goldstein (1982b) provides a more

detailed discussion of the design.



. Page 21



Research Site - The research site should be a large organization that
is about to introduce structured methods on some of its project groups.
Ideally, the organization should have 100 or more programmer/analysts. It
should develop mostly large transaction processing and management
information systems that span boundaries within the organization. These are
the types of systems most suited for development with structured methods.
The organization should not have suffered any recent organizational
disruptions, such as the ones discussed in the previous section, that could
overshadow the effects of the introduction of structured methods. We should
be able find the organization through a company involved in marketing
structured methods.

Experimental procedure - The questionnaire should be administered to
all the programmer/analysts in the organization and their project leaders
right before the introduction of structured methods and six months after
their introduction. The structured methods should be used by at least 30
prog rammer /anal ysts.

Measures - Measures have already been developed for many of the
variables in the model. For example, the scales developed by Rizzo, et
al. (1970) and modified by Bostrom (I98O) can be used to measure role
conflict and ambiguity. Hackman and Oldham's (I98O) Job Diagnostic Survey
contains scales that measure skill variety, task identity, and autonomy.
The JDS also contains items to measure job satisfaction, level of
motivation, and satisfaction with job contexts. Bowers and Seashore's
(1966) measure can be used to measure the leadership characteristics of each
programmer/analysts' peers and project leader. Other items can be added to
the questionnaire to measure task achievement, the other characteristics of
programmer/analysts identified above, and the programmer/analysts'

Page 22



proficiency with structured methods.

Project leaders should be given a separate questionnaire. It should be
used to obtain information on productivity and quality for their
programmer/analysts, to obtain information on their team and on their own
background, and to verify the intervening variable measures obtained from
their programmer/analysts.

Method of Analysis - Simple gain score analysis can be used to test the
hypotheses. This is an analysis of variance technique that examines the
differences in changes in a variable from the pretest to the posttest. The
assumption is that the treatment will lead to more (or less) change in the
treatment group than in the control group (Cook and Campbell, 1979).

The characteristics of the treatment and control groups can be examined
to determine the validity of our results. Any significant differences
between the treatment and control groups on the background, individual
differences, and team differences measures could affect the internal
validity of our results. Further, the characteristics of the entire
population will give us some idea of the general i zabi 1 i ty of our results.

DISCUSSION

This research will provide some insights into the effects of the use of
structured methods on the job satisfaction of programmer/analysts. Using a
pretest-posttest with control group design allows us to isolate the effects
of structured methods. The intervening variables provide further
explanation as to how the use of structured methods affects job
sati sfaction.



Page 23



There are, however, some potential problems with the study. First, the
six month time frame might not be long enough for programmer/analysts to
become familiar with structured methods. However, a longer time frame would
create problems with experimental mortality. Turnover would cause people to
drop out of the experiment between the pretest and the posttest.

Second, differences between treatment and control groups and between
the experimental groups and the population of programmer/analysts could
affect the general izabi 1 i ty of the results. Although this is hard to
control for in a quas i -exper imental design, the presence of multiple
measures of individual and team differences should point out any potential
problems.

A larger sample survey could eliminate some of the problems. The
instruments developed in this quas i -exper iment could be used to compare the
programmer/analysts that use structured methods to those that do not use
structured methods in several companies that have partially implemented
structured methods. If the results of the pretest-posttest quasi -exper iment
were confirmed with this posttest only design, this would provide stronger
evidence that the hypotheses were true.

Finally, this research should lead to further research in two areas.
First, this research will provide some insights into the determinants of job
satisfaction in programmer/analysts. Further research could explore the
effects of the factors identified here, as well as other factors, on job
satisfaction. Research could also explore the effects job enrichment
programs or other work redesign efforts on programmer/analysts. This is an
important research, because of the relationship between job satisfaction and
job outcomes such as productivity and turnover. These are critical problems
in d. p. organizations.

Page 2U



Second, the research provides some insights into the effects of job
changes on job satisfaction. The introduction of new office technologies
and the more widespread use of computers will change the jobs of many
workers. Further research could examine the effects of these new
technologies on job satisfaction. This would give us a better idea of the
possible costs and benefits of office automation or other technologies.



Page 25



REFERENCES



Awad, Elias M., "Prediction of Satisfaction of Programmers, Analysts," Data
Management , January, 1977. 12-18.

Baker, F. T., "Programmer Team Management of Production Programming," IBM
Systems Journal , Volume 11, Number 1, 1972, pages 56-75'

Bergland, G. D., "A Guided Tour of Program Design Methods," Computer .
October, I98I, pages 13-37-

Bernstein, C. M., "Experience with Exxon's Implementation of the Jackson
Program Design Method," National Computer Conference Proceedings , 1978,
pages 636-637-

Bostrom, Robert, P., "Role Conflict and Ambiguity: Critical Variables in
the User-Designer Relationship," Proceedings of the Seventeenth Annual
Computer Personnel Research Conference , I98I, pages 88-112.

Bowers, David G. and Stanley E. Seashore, "Predicting Organizational
Effectiveness with a Four-Factor Theory of Leadership," Admi ni strat ive
Science Quarterly , 1966, pages 238-263.

Canning, R. G., Editor, "The Search for Software Reliability," EDP Analyzer ,
Volume 12, Number 5. May, 197'«a-

Canning, R. G., Editor, "The Advent of Structured Programming," EDP
Ana 1 yzer , Volume 12, Number 6, June, 197'»b.

Canning, R. G., Editor, "The Analysis of User Needs," EDP Analyzer , Volume
17. Number 1, January, 1979a-

Canning, R. G., Editor, "The Production of Better Software," EDP Analyzer ,
Volume 17. Number 2, February, 1979b-

Canning, R.G., Editor, "Application System Design Aids," EDP Analyzer ,
Volume 19, Number 10, October, I98I .

Cook, Thomas D. and Donald T. Campbell, Quas i -Exper imentat ion Design and
Analysis Issues for Field Settings , Houghton Mifflin Company, Boston,
1979.

Couger, Daniel and Robert Zawacki, Motivating and Managing Computer
Personnel , John Wiley and Sons, New York, I98I.

Goldstein, David K., "Exxon's Use of Structured Systems Analysis: A Case
Study," Unpubl i shed paper , I98I.



Page 26



Goldstein, David K., "The Impact of Development Aids on the Systems
Development Process," Proceedings of the Conference on Human Factors in
Computer Systems , 1982a, pages 130-13't.

Goldstein, David K., "The Effects of Structured Development Methods on the
Job Satisfaction of Programmer/Analysts: A Quasi -Exper imental Design,"
Unpubl i shed paper , 1982b.

Greenbaum, Joan M., In the Name Of Efficiency , Temple University Press,
Philadelphia, 1979.

Hackman, J. R. and G. R. Oldham, Work Redesign , Addison Wesley, Reading,
Mass.. 1980.

Hamilton, Kenneth and Arthur Block, "Programmer Productivity in a Structured
Environment," I nfosystems , April, 1979, pages 4i»-50 and May, 1979,
pages 62-68.

Inmon, Bill, "An Example of Structured Design," Datamation , March, 1976,
pages 82-86.

Jackson, Michael A., Principles of Program Design , Academic Press, 1975.

Jones, Martha N., "HlPO for Developing Specifications," Datamation . March,
1976, pages 112-125.

Kahn, R. L. , Wolfe D., Quinn R., Snoek J. D., and Rosenthal R.,
Organizational Stress: Studies in Role Conflict and Role Ambiguity ,
John Wiley, New York, I96i*.

Katz, Ralph, "Satisfaction and Behavioral Reactions of Employees to Task
Characteristics as a Function of Job Longevity. Age. and High-Order
Needs," Sloan Working Paper , 932-77. MIT, Cambridge, Mass, 1977.

Katz. Ralph, "The Influence of Group Longevity on Project Communication and
Project Performance," Sloan Working Paper . 1179-80. MIT. Cambridge,
Mass, 1980.

Kraft, Philip, Programmers and Managers . Spr i nger-Verlag, New York. 1977.

Locke, E. A., "The Nature and Causes of Job Satisfaction," in Dunnette,
M. D., Ed., Handbook of Industrial and Organizational Psychology , Rand
McNally, I976.

Menard, Jayne B., "Exxon's Experience with the Michael Jackson Design
Method," Data Base , Volume II, Number 3. Winter-Spring. I98O. pages
88-92.

Mendes, Kathleen S., "Structured Systems Analysis: A Technique to Define
Busines Requ i rments, " Sloan Management Review , Volume Ik, Number *♦,
Summer, 198O, pages 51-63-



Page 27



Rader , J. A., "Experience with an Application of Structured Design,"
'EB 9 n in National Computer Conference Proceedings , 1978, pages 633"636.

^^^"Ct Rizzo, J. R., R. J. House, and S. J. Lirtzman, "Role Conflict and Ambiguity

in Complex Organizations," Administrative Science Quarterly , Volume 15.
1970, pages 15O-I63.

Ross, Douglas T., "Structured Analysis (SA) : A Language for Communicating
Ideas," IEEE Transactions on Software Engineering , Volume SE-3. Number
1, January, 1977. pages 16-21.

Stevens, W. P., G. J. Myers, and L. L. Constantine, "Structured Design," I BH
Systems Journal , Volume I3. Number 2, 197'*. pages 115"139'

Wanous, J. P., "A Causal-Correlational Analysis of the Job Satisfaction and
Performance Relationship," Journal of Applied Psychology , Volume 59.
197'*, pages 139-Ki».

Warnier, J. D., Logical Construction of Programs . Van Nostrand, Reinhold,
New York, 1974.

Willoughby, T. C, "Staffing the MIS Function," ACM Computing Surveys ,
Volume U, Number I4, 1972, pages 2l»l-259.

Winters, Edward W., "Experience with Problem Statement Language: A Tool for

Structured Documentation," Appl i cat ion Deve 1 opment Sympos i urn

Proceedi ngs , 1979. pages ]}'J-]k].

Yourdon, E., and L. L. Constantine, Structured Design , Yourdon Press, New
York, 1975.

Yunger, G. W. and J. G. Hunt, "An Empirical Comparison of the Michigan
Four-Factor and the Ohio State LBDQ leadership Scales," Organ i zat i ona 1
Behavior and Human Performance , Volume 17, 1976, pages '♦S-^S-



Page 28



BASfiM^fe



^G0

SEP I '68



Z'SlB



Lib-26-67



HD28.M414 no.l330- 82

Goldstein, Oav/The effects of structur

745062 D*BKS 0Q136

III'




3 TDflD 002 DM? SM3