INSTRUCTIONAL SYSTEMS DESIGN
Island
MultiMedia and its Associates have extensive expertise and experience in the
application of Instructional Systems Design procedures and approaches to
performance problems. The rigor and productivity of the ISD approach, when
correctly and appropriately applied, ensure results, control costs and reduce
risk. ISD as a discipline is the central content of many graduate degrees, but
some of its roots and essential characteristics are outlined in the following
article. Island MultiMedia Associates have been primary contributors to several
of the most successful of the major ISD models.
AN INTRODUCTION TO INSTRUCTIONAL SYSTEMS DESIGN,
O'Neal, A.F.,
Fairweather, P.G., and Huh Y.H.
ILO Asian and Pacific Skills Development Program
United Nations, Goa, India, 1988.
Historically,
instructional development has tended to be an artistic endeavor, carried out in
a cottage industry setting (Molnar, 1971). All aspects of development
responsibility including analysis of the training or educational problem,
design of the instruction, development of the materials, and in many cases,
production of the media, evaluation, and revision were concentrated in the
individual instructional developer. The developer's approach to each
instructional problem tended to be ad hoc and subjective. The developer
himself, more often than not, tended to have expertise in the subject matter
being addressed, rather than formal training in instructional design science.
This model of
the Renaissance man as instructional artist, solving each training problem as
it arose with a combination of experience, intuition, and personal insight,
began to become suspect as instructional development activities grew larger and
more complex and as the consequences of inadequate training grew more and more
expensive. Figure 1 represents three points on a continuum of instructional
development methodology. The artistic-intuitive approach, with its dependence
on the judgement and multi-disciplinary skills of the individual developer
gradually has yielded to more systematic approaches to the design and
development of instruction and training systems and materials.
As shown on
the figure, the search for better and more systematic ways to handle
instructional problems led to the development of some important tools,
including task analysis, the use of well defined behavioral objectives, and
sophisticated measurement and evaluation methods. To protect against the
consequences of poor training, more reliance was placed on empirical methods,
which involved repeated tryout and revision of materials (Merrill &
Boutwell, 1973). This stage in the evolution of a technology of instructional
design is represented by the empirical phase of Figure 1.
Although these
methods were expensive and time-consuming, they did at least help pinpoint
inadequate instructional segments and make possible the improvement of
instructional effectiveness. However, once inadequate segments were identified,
the job of revision of the instructional materials was generally still left to
the Renaissance man. His activities were still basically artistic in nature,
his solutions to instructional problems were still idiosyncratic, and the
procedures that he might apply tended to be based on his own experience and
were often not generalizable to other developers. Notice from the figure that
analyzing the results of the training could lead to changes in the experience
and intuition of the developer as well as, hopefully, to the improvement of the
instructional product through revision. In this mode, if the project had
unlimited time and money, and could afford to iterate enough times, very good
training might eventually emerge. Unfortunately, in the real world, there never
seems to be enough time or money. There was clearly a need for further
evolution of instructional design and development procedures.
In the early
1970's a growing trend began in training circles away from this artistic
approach to development, and towards the rigorous application of theory- and
research-based models in an instructional engineering approach to development
of instructional programs. This class of procedures has been variously called
instructional systems design or development (ISD), as a systems approach to
training (SAT), or by other such designations. As shown in Figure 1, the
approach allows for inputs from experience, intuition, and most importantly,
from research.
Notice that an
important aspect of the systematic model is the formulation of different levels
of definition of practice. First, scientists synthesize principles from
learning research and the experience and intuition of expert training
practitioners. These principles represent high level abstract generalizations
such as "Distributed practice is superior to massed practice in long term
skills maintenance", and "Increasing imagery improves
retention". They should be observable in training practice and supported
by research findings. However, at this level of abstraction they offer little
guidance to the instructional designer.
The high level
principles must be documented and developed into operational procedures. These
procedures are designed to be applied by technicians to generate instructional
products. Again results are analyzed and may lead to improvements in the
product, changes in the experience and intuition of the practitioners, or they
may indicate the need for further research. Most important of all, however,
these results may lead to improvements in the procedures, leading to better
practice!
Figure 2 shows
a simplified view of the ISD process. It starts with the analysis of the
problem, the context, and the intended learner population. The analysis
products (such as the job/task analysis, the entry population analysis, the
needs/goals/constraints analysis, etc) provide inputs to the design phase of
the project. Here the learning objectives are refined, the training media
specified, the syllabus is generated, and the individual lesson designs are
specified. The design documents form the basis for the development phase, and
the implementation and evaluation phases are carried out based on the
evaluation and implementation plans developed in the analysis and design
phases. Notice that evaluation in a systematic model of development has a
quality control aspect. Since the process proceeds according to well specified
and documented procedures, with well defined products at each stage, it is
possible to evaluate the emerging training product at each step, detecting
problems as they emerge, instead of discovering them much later, in a training
product which doesn't work.
Most large
scale approaches will be similar at this general level. Many will recognize
that at this level, ISD is nothing more than the application of well proven
systems development techniques to the problem of training development. At finer
levels of detail, where the individual procedures are defined, strong
institutional, and even personal influences, may be clearly evident in
different ISD models. Traditions and philosophies of ISD practice may often be
clearly traced by such distinctive characteristics (Gibbons 1988). It is,
however, at this finer level of detail where the real benefits of ISD begin to
emerge.
When closely
examined, good ISD is more engineering than art. Its important benefits come
from well documented procedures, a differentiated staff team development
approach, separation of instructional content and strategy, and the continuing
evolution of a prescriptive, analytical, research-based model.
In order for
teams of specialists to work efficiently and effectively together, procedures
must be well documented at all levels of the ISD process. Documented procedures
allow for peer review, process control, and the possibility for improving practice
over time. They help to standardize the output of different team members doing
the same task and they help alleviate the training burden imposed by new team
members arriving during the project. They serve as quality control as well as
development tools. While documentation has long been an essential ingredient of
systems design in general, it has extreme value in ISD in particular.
An essential
ingredient in any large scale instructional development activity is the
development of some form of team organization where specialized expertise can
be most effectively utilized and where personnel training problems can be
minimized through specialization. This tends to result in an industrial
revolution, or factory approach to instructional development. Major areas of
specialization include (but are not limited to) instructional content (subject
matter experts), instructional process (instructional designers), and media
technicians.
One of the
most important contributions of ISD is the separation of content and strategy.
Content is described in terms of well defined and documented, discrete
instructional components. Strategies are defined in terms of well specified
sequences of these components set in the context of the particular media
selected for each instructional module. This approach has particular value in
the specification of frame-oriented media and in complex logical environments
such as computer based training. Sets of well defined, strategy independent
structures also allow the easy and economical construction of learner
controlled training environments where desirable.
Perhaps the
greatest strength of the ISD process is the evolutionary nature of the
prescriptive, research-based model itself. While the practice of ISD still
retains the strengths of the empirical evaluation and revision cycles, to the
extent research and experience permit, it is prescriptive. That is, rather than
depending extensively on the test-revision cycle to generate effective
instruction in an iterative manner, every attempt is made to incorporate
research findings and past experience into the detailed procedures and
supporting ISD documentation to ensure that the instruction developed comes as
close to the mark as possible the first time. This improves the validity of the
process while also improving reliability. This has proven to be a powerful tool
in large scale ISD. In addition, as the process provides more data from the
constant evaluation process, the procedures can be continually improved.
At the
detailed level the particular procedures developed for each ISD model may
differ considerably. Figure 3 shows some important sub procedures or acivities
for a typical industrial training ISD model. There are strong dependencies
between the activities.
1) Task
Analysis is dependent upon a clear identification of the problem as a training
problem. If the problem is NOT a training problem, no amount of training or
training related activity will solve the problem. Therefore a task analysis
should not be undertaken until there is clear identification of the problem as
a training problem.
2) Similarly,
the Needs, Goals and Constraints Analysis (NG&C) should not be completed
until there is clear identification of the problem as a training problem.
However, since this may sometimes require some of the activities normally
completed during a NG&C in order to fully determine, the dependency is
partially mutual.
3) The Entry
Population Analysis (EPA) is intended to identify the important characteristics
of the intended population(s) for the training. While this may seem on the
surface to be straightforward, there are sometimes surprises uncovered during
the NG&C in terms of who wants (or doesn't want) the training, and for
what. Therefore the EPA is dependent upon the NG&C to at least the degree
that it should not be finished until all the information from NG&C is
available.
One major
purpose of the EPA is to help with the selection of tasks for training
(sometimes called out as a separate ISD activity). The premise is that there is
no use spending time, money, and effort on developing instruction for
skill/tasks uncovered on the task analysis, if the intended population already
has these skills etc in their repertoire. Therefore the EPA is dependent on the
Task Analysis being completed before the EPA can be completed. On the other
hand, it is more economical to compare the entry capabilities of the intended
population against the restricted set of things identified in the task analysis
than it is to ascertain/describe the entry repertoire in general of the
population.
4) Once the
EPA is completed and the tasks to be trained have been selected, it is possible
to define the Evaluation and Implementation Plans. This plan must accommodate
the needs and goals of the various groups identified in the NG&C and must
conform to the availability of samples of the population identified in the EPA
for formative evaluation. In addition, the implementation planning must
accommodate the constraints (personnel, time, resources, budget, and traditions/corporate
culture) of the organizations impacted. Failure to plan UP FRONT for both the
evaluation and implementation of the training can result in serious problems
later in the project.
5) Once the
tasks to be trained have been selected as part of the EPA, it is possible to
expand the objectives derived from the surviving portions of the task analysis
into an Objectives Hierarchy by supplying the supporting and enabling objectives
that this student population will require. For example, an experienced,
sophisticated population learning a new variant on an already mastered set of
skills may not require many intermediate levels of partial training objectives.
A naive, unsophisticated population may need lots of levels of successive
approximations of the more complex tasks, and they may even need considerable
training on how to use the training system and/or on "how to learn"
in general, not just on the training content.
6) As the
Objectives Hierarchy is completed, it is possible to begin examining the media
requirements the objectives imply in terms of stimulus, response, control,
record keeping, and other dimensions. This is only one part of the final media
decision and at this point should be concerned only with the instructional
requirements of the objectives, unfettered by real world considerations of
cost, availability, etc.
As basic
instructional requirements are established, the media choices for each
objective must be qualified by cost, availability, and practical considerations
of implementation within the syllabus context. For this reason the final media
selection is mutually interdependent with the syllabus development process.
Sometimes you may have to make changes in the media selection based on
practical considerations from the syllabus, and other times you may choose to make
alternative decisions on the syllabus definition based on media considerations.
These decisions must NOT violate considerations of minimum instructional
requirements for the media for an objective, or prerequisite sequencing in the
syllabus, however. That is, the media you finally settle on must be able to do
the job called for in the objective, and the syllabus sequence you end up with
must never have a student trying to accomplish an objective for which he has
not had the prerequisites.
8) As the
Objectives Hierarchy is completed, it becomes possible to Classify Objectives
in terms of the category of instructional problem they represent.
To a certain extent, this will affect the media selection in that certain
instructional strategy dimensions in terms of control, manipulation of stimuli,
and numbers of instances may imply certain media requirements. The indirect
nature of this interaction is implied through the route from the Objectives
Classification to the Media Selection through the Syllabus Development. It may
be that a more direct connection on the diagram would better represent the
relationship.
Similarly, resource constraints identified in the NG&C may affect the final
media selection, by reducing the suite of candidate media realistically to be
considered.
9) The mutual
interdependence of the Syllabus Development and Media Selection processes has
been identified above, as has the requirement that the syllabus derived must
not violate prerequisite relationships between objectives identified in the
Objectives Hierarchy Development.
10) The
syllabus should accomodate the Objective Classification in that an "early
hands on" approach should be encouraged as opposed to a "do all low
level objectives first before advancing to the next level of the
hierarchy" approach. That is, if possible, the syllabus should encourage
the learner to advance quickly up a leg of the hierarchy until a significant
skill, representing some identifiable subset of the job being trained, can be
mastered, before returning to the lower levels of the hierarchy and attempting
another leg of the hierarchy. The alternative of doing all low level objectives
first and then advancing leads to syllabus definitions which result in
days/weeks/months of low level learning objectives such as memorizing terms,
locations, and functions, before advancing to simple procedures, and finally,
late in the course, actually beginning to master some of the high level
objectives which are identifiable as approximations of the job being trained.
This latter approach will severely impact motivation.
11) Further,
of course, the syllabus must accomodate the real world concerns and constraints
identified in the NG&C. Shortages of critical resources such as simulators,
instructors, and certain media or facilities may lead to quite a different
syllabus design than would otherwise be the case. The syllabus must accomodate
these considerations as anticipated in the Evaluation & Implementation
Plan.
12) The
Author Management System which manages/tracks development of the training
materials is dependent on the syllabus in several ways. First, the syllabus may
be implemented in part while development of the later materials is still
underway. This is often the case where time is a severe constraint. In that
case the syllabus will identify the order in which development should proceed
as well as the order in which learners will progress. In addition, the syllabus
order must always be accomodated in the instruction in the sense that until you
know the syllabus order, you cannot assume what the learner will already know
at any given part of the course. This confusion often leads inexperienced
developers (especially SME's) to essentially try to "teach the entire
course" in each instructional segment.
13) The
Lesson Specifications are generated according to the instructional strategy
most appropriate for the instructional classification of each objective. They
constitute the "micro-design" and initial content capture for each
lesson.
14) The
lesson is then developed in full and the media/materials are produced.
15)-16) The
course is then implemented and evaluated according to the Implementation and
Evaluation Plan developed in (6) above.
This is only
one example of an elaborated ISD model. Each project and institution will
develop their own detailed model. The ISD approach, with its potential for self
improvement and its engineering discipline has taken instructional development
from the productivity of the individual artisan to the productivity and quality
control of the factory. It is limited only by the imagination and discipline of
the practitioners.
Molnar, A.R. the future of educational technology research and development.
Washington, D.C. : National Science Foundation, 1971. (ERIC Document
Reproduction Service No. ED 054 642)
Merrill, M.D., & Boutwell, R.C. Instructional development: Methodology and
research. In F.N Kerlinger (Ed) Review of research in Education (Vol 1).
Itasca, Ill. F.E. Peacock, Publishers, 1973.
Gibbons, A.S. The influence of instructional systems development (ISD) on
simulator design. Wicat Systems, Orem, Utah, 1988.
Copyright © 1996 Island MultiMedia. All rights reserved. Reprint of this
article is allowed with copyright acknowledgement.
Last Updated January 2003
Comments,
suggestions to fred@whidbey.com