Table of Contents
- Introduction
- Taxonomy of ISD Models
- Gerlach & Ely Model
- Kemp Model
- Dick & Carey Model
- Seels & Glasgow Model
- Rapid Prototyping Model
- Criticism of the Dick and Carey Model
- Links
- References
Dorsey (1997) suggests that a model is "intended to serve as a framework to guide the thinking and practices of designers in the field" (p.445). Dorsey implies that models should be judged based on their fidelity - i.e., the degree to which the model reflects reality in its attempts to describe, prescribe, predict, or explain (Andrews & goodson, 1980). This drive to improve the fidelity of instructional design (ISD) models, I suggest, has led to a proliferation of models over the past few decades. Nevertheless, these models seem to share more similarities than differences and can generally be described as variations on the basic ADDIE blueprint: analysis, design, delivery, implementation, and evaluation.
Arguably the most popular ISD models are the Dick & Carey (1985), Kemp (1985), Gerlach & Ely (1980), and Seels & Glasgow models. In contrast to these traditional models, we are currently witnessing the emergence of Dave Merrill's ID2 model (a derivative of component display and transaction theories) and Thiagi's Rapid Prototyping model. On this page, I have attempted to describe summarize each of these ISD models.
In his 1991 Survey of Instructional Development Models, Gustafson categorizes ISD models as classroom, product development, and systems development models. The following table, adapted from Gustafson's taxonomy, categorizes the ISD models that Gustafson studied.
|
Classroom
Models |
Product Development
Models |
Systems Development
Models |
|
- Van Patten
- Leshin, Pollock, and Reigeluth - Bergman and Moore |
Classroom Models - The classroom models assume that there is already an instructor, some students, a curriculum, and a classroom. The goal of the instructor, and in turn the model, is to improve the development and delivery of a set piece of content. These models are popular in K-12 and higher education.
Product Development Models - The goal of a product-oriented ISD model is to optimize the production efficiency and quality of one or more instructional products (e.g., CBTs, WBTs, textbooks, etc.). Usually, these products are replicable and the decision to develop them (i.e., the needs assessment) has already been made by higher powers. These models tend to have a corporate feel to them.
Systems Development Models - The goal of a Systems model is the development of a complete instructional system for managing learning needs. In other words, Systems models address issues of what instruction should and shouldn't be developed and how the process should be planned, implemented, and evaluated. In addition, they typically require extensive analysis of environment, learners, needs, and tasks. These models are commonly used in millitary and government institutions, mid-size to large corporate training departments, and increasingly in higher education.
The following matrix summarizes some of the characteristics. It has been reproduced from Gustafson (1991, p.8).
|
Hours of
Instruction |
Resources
Committed to Dev. Process |
Team or
Individual Dev. |
Emphasis on
Dev. or Select Materials |
Amount
Front End Analysis/Needs Assessment |
Amount Tryout
and Revision |
Distribution/
Dissemination |
|
|
Classroom
Instruction |
Hour of
Instruction |
Very |
Individual
|
Select
|
None |
Low
to Medium |
None
|
|
Product
Orientation |
Self
Instructional Package |
High
|
Individual
or Team |
Development
or Select |
Low
to Medium |
Very
High |
High
|
|
Systems
Orientation |
School, College,
or Military Course (s) |
High
|
Team
|
Development
|
Very
High |
Medium
to High |
Medium
to High |
Click here for a graphical depiction
(from Sue Braxton's site)
The Gerlach and Ely (1980) model is a classroom ISD model that is particularly
popular in K-12 and higher education. The model assumes that the need for
course content has been established (e.g., through a standard curriculum),
so it includes no needs assessment. Similarly, since it was designed with
teachers in mind, who implicitly know their students and classroom settings
better than anyone else, no learner or context analyses are included either.
Gerlach and Ely also recognize that teachers (and most other educators in
my opinion) think about their courses first and foremost in terms of content,
not objectives. Therefore the model describes its first step as an interactive
process of examining content and generating objectives. This is a distinguishing
feature, since most models do not recognize this content orientation (Gustafson,
1991).
The second step in the model highlights the importance of assessing entry behaviors, though, as Gustafson points out, Gerlach and Ely offer "few concrete procedures describing how to accomplish this step" (p.11). The third step consists of five interactive, simulatneous procedures: (1) determine strategy, (2) organize groups, (3) allocate time, (4) allocate space, and (5) select resources. These are followed by a performance evaluation and a feedback analysis step, which are similar to those in other ISD models.
Click here for a graphical depiction
(from Sue Braxton's site)
The Jerrold Kemp (1985) model is similar to the Gerlach and Ely model in several
ways. Both are classroom models that are designed to appeal to teachers in
particular. However, Kemp's claim to fame is its flexibility. You can start
anywhere you like. While the process steps are laid out in a clockwise order
and his book generally organizes discussion of the model in a manner that
suggests a sequential form, Kemp's intent is for the elements (not steps)
to be non-sequential (even simulatenous) and interdependent. In addition,
much like Hannafin
and Peck's ISD model (not covered here), evaluation and revision are considered
continuous and on-going processes. Lastly, Kemp's model pays some attention
to needs assessment by placing "learning needs, goals, and priorities
contraints" at the center of the model. Nevertheless, its focus in on
the content, not the analysis or evaluation.
Click here for a graphical depiction
(from Sue Braxton's site)
The Dick and Carey (1990) model could be considered both a product development
and systems development model, depending on the scope of the project and the
depth of the development steps. The model depicts a comprehensive and somewhat
linear process that is sometimes criticized for being overly-rigid and cumbersome.
Dick and Carey (1990) propose a comprehensive and systematic model for developing
instruction in a wide variety of contexts (corporate, education, military,
etc.) and at a variety of levels (curriculum, program, course, unit). While
it follows the 5 basic stages of the ADDIE model - analysis, design, development,
implementation, and evaluation, it offers more detail and guidance than other
models (e.g. Kemp, 1985; Seels & Glasgow, 1990). As a result of its comprehensiveness,
the Dick and Carey model has gained widespread acceptance and often comes
to symbolize ISD (instructional systems design). For these reasons, because
it is so comprehensive and also symbolic of established instructional design
practices, it has come under attack. I shall address this further in the second
part of this paper. First, let's look at the various components of the Dick
and Carey model through the lens of the ADDIE model.
Analysis
Front-end analysis is a critical component in the Dick and Carey model. A
characteristic step in this model is to perform an upfront needs assessment.
Here, instructional designers (IDers) look at things such as optimal and actual
performance, the feelings and perspectives of those involved, and the environment
which supports (or does not support) performance (Rossett, 1991). Depending
on the type(s) of problems (needs) identified, solutions are selected from
a variety of performance improvement strategies, including clarifying goals,
redesigning workflows, changing incentive systems, and providing job aids.
Among these strategies is developing instruction. If this strategy is appropriate
- and Gilbert (1978) argues that the potential benefit of the intervention
should be at least 1.5 times greater than the cost in order for it to be a
worthwhile endeavor (see Gilbert's work with performance improvement potential
- PIP) - then and only then does the instructional designer move to the next
steps in the analysis stage: goal, task, learner, and context analysis. These
steps produce a profile (a framework or skeleton) upon which the instruction
will be developed. Goals and task analysis lead to the development of behavioral
learning objectives, task outlines, behavioral algorhythms, semantic maps,
and other depictions of the content domain. The learner analysis guides decisions
about entry-level behaviors as well as those regarding with the actual design
of the instruction. Context analysis also reveals information about the environment
in which the mental or physical task being trained will actually be used.
This can help IDers keep contextual factors in mind as they design instruction.
To summarize, if an initial needs assessment warrants the development of instruction,
goals, tasks, learners, and context are analyzed. A critical outcome of this
stage is the hierarchical list of learning objectives.
Design
In the design phase, two sets of products will be designed: the assessment
items and the instructional content itself. The assessment items are derived
and should correspond directly with the criteria (learning objectives) written
in the analysis stage. This is what makes them criterion-referenced test items
rather than norm-referenced items (which are not directly linked to learning
objectives). Once the assessment items are written, the instructional content
can be designed. In particular, what are being documented here are the over-arching
and goal-level design strategies for the curriculum, program, course, or unit
as well as the design of the learner interactions (if any) that will take
place. The former is typically written up in the form of a design strategy
document, while the individual interactions are documented in a treatment
plan. In the case of computer-based instruction, interactive video, and other
forms of technology-based instruction, scripts/storyboards are developed to
guide the development of media and programming. To summarize then, by the
end of this design stage, a design strategy document (containing the findings
of the analysis stage along with design strategies), test items, a treatment
plan for all of the interaction types, and possibly a set of scripts/storyboards.
Development
In the development phase, the tests, handouts, instructor guides, presentation
slides, handouts, and all of the instructional media needed are prepared.
In computer-based instruction, audio and/or video may need to be recorded,
edited, and digitized; graphics and animations will need to be developed;
and lines of programming code will need to be written/authored. A common practice
here is to generate narration scripts for the audio and video producers; graphics
lists for the graphic artists, and full scripts/storyboards for the programmers/developers.
Implementation
In this phase, the instruction is implemented in the actual learning setting
with real learners. Sometimes, a pilot test of the instruction is done. In
other cases, the new program is simply rolled out to the learners. During
this phase, typically instructors, tech support, facilitators, and other stakeholders
of the instruction need to be briefed, if not trained themselves. For example,
trainers would need to be familiarized with how to present a new course, how
to use the instructor guides, and how to run the activities for example.
Evaluation
The Dick and Carey model put a good deal of emphasis on evaluation and quality
control, distinguishing formative and summative evaluation. Formative evaluation
is an ongoing process of reviewing outcomes, often with the help of subject-matter
experts and/or clients. For example, at the end of the design stage, the IDer
might send the client the design strategy document for review. If the client
finds something wrong, then the IDer would go back and revise the strategy
or might even go back and revise learning objectives. Nevertheless, formative
evaluation in the Dick and Carey model seems to center on the alpha, beta,
and pilot tests. While the definitions seem to change depending on the context,
alpha testing is typically a dry run that involves no learners and is run
off-line from the actual learning context. For example, in a web-based instruction
context, this might mean the testing done before the course is moved to the
server that will house it. A beta test typically involves a live test and
may or may not involve actual learners. Sometimes, the trainers themselves
or other people will run through live instruction to a true dry run. A pilot
test could be considered a beta test, however, it typically occurs after the
development and basic testing of the instruction is considered finished. A
pilot test involves actual learners, sometimes only a sample of them. Again
the results of these formative evaluation procedures are documented through
variance tracking methods ("bug lists," is a common term for them)
and then the instruction is revised.
After the instruction has gone through formative evaluation, it is rolled
out to the entire population of learners. Once some time passes, Dick and
Carey suggest, the effectiveness of the course should be assessed through
summative evaluation procedures. These can involve smile sheets (level 1 on
Kirkpatrick's evaluation scale), assessment of learning outcomes (by analyzing
post-test scores for example), evaluation of transfer rates (finding out how
well the skills have transferred to actual work settings), and/or assessment
of impact (e.g., return on training investment… was it worth doing?).
In their book, Exercises in Instructional Design, Seels and Glasgow (1990) analyze a number of different ISD models and then proceed to discuss their own model. Actually, there are two different Seels and Glasgow models floating around. The first one is rather linear in form and offers little that is new (basically a recap of ADDIE) except for one thing: the ISD process occurs within the context of project management. This is an interesting change of pace insofar it recognizes the fact that instructional designers often play the role of project managers or at the very least that they work within a broader project management model. In my opinion, this seems like a facet that should be expanded upon both in the Seels and Glasgow model as well as in the ISD literature in general. Most other models treat instructional development as an isolated actvity rather than as something that involves a much broader context and set of activities.
Click here for a graphical depiction
(from Sue Braxton's site)
Rapid prototyping is both a systems and a product development model. It is
often seen, in some variation, within corporate settings. In particular, it
is increasingly popular in software development contexts due to its perceived
benefits of improving quality while simultaneously reducing development costs.
Generally, rapid prototyping models involve learners and/or subject matter
experts (SMEs) interacting with prototypes and instructional designers in
a continuous review/revision cycle. Developing a prototype is practically
the first step, while front-end analysis is generally reduced or convereted
into an on-going, interactive process between subject-matter, objectives,
and materials (similar to Gerlach and Ely's treatment of objectives). Evaluation
is a recurring event that moves from global issues (navigation, structure,
design, motif, color scheme, etc.) to fine tuning of instructional content
(e.g., semantics). generally speaking, this model makes course development
and analysis simulatneous processes with evaluation loops recurring frequently
throughout the life of the project.
With Rapid Collaborative Prototyping, a model put forward by Dorsey, Goodrum, and Schwen (1997), the learners, SMEs, and instructional designers (IDers) are equal partners in the development process, not only reducing the need for time-consuming and inaccurate (being subject to the IDers interpretation) learner, context, need, and task analyses but also improving the quality of the course. Instead of communicating the design of a course through words in a design strategy document, for example, ISD concepts are communicated through concrete examples (i.e., the prototypes), so reviewers are able to provide more precise feedback early enough in the process when revision costs are relatively low (especially where it concerns major design issues). Another benefit is that as the product is reviewed and revised over multiple cylces, the entire production team becomes more and more familiar with a client's needs and is therefore improving itself just as much as the product is being improved.
The downside, of course, is the wear and tear on the production team. Project managers will try to control a liquid and at times chaotic process; instructional designers will make strategic decisions before much in the way of analysis has been completed and work with lay people to fine-tune their designs; graphic artists and programmers will work intense hours revising their outputs repeatedly (something that drives them crazy); and learners/SMEs will consume a great deal of their time going through the course over and over again. In order to be successful, the entire team needs to be comfortable with an iterative process where you are presenting something to the client that is incomplete and of relatively low quality. In my experience, individuals working in development teams are perfectionists and not entirely patient. They want to do a great job coming out of the gates. With rapid prototyping, this is neither possible nor necessarily desirable.
Criticism of the Dick and Carey Model
In my specific context, the Dick and Carey model represents
a fairy-tale, describing how instruction could be developed if we had unlimited
time, knowledge, and resources. Much of my work in the last few years has
been in the area of computer-based and web-based instruction. In particular,
I have worked as a project manager and lead instructional designer serving
as a consultant for Fortune 500 clients. In this capacity, I had to lead a
team of designers, artists, programmers, and sub-contractors in an effort
to develop instructional modules within a given budget and timeframe to the
client's satisfaction. We were evaluated by the client based on the quality
of the instruction, the seamlessness of the development process, and our ability
to hit target dates. We were evaluated internally based on out ability to
make the client happy while staying within budget. Subject-matter expertise
was typically provided by the client in the form of subject-matter experts
(SMEs) and documentation. The SMEs operated outside of my sphere of control.
Needless to say, in most cases, my design team did not have very much subject-matter
expertise of its own. Given these constraints on time, expertise, and resources,
I found the Dick and Carey model unrealistically constraining, arrogant, and
significantly prone to problems and at times irrelevant.
The model emphasizes the role of needs analysis, implementation,
and summative evaluation, but these components were well out of my hand. If
a needs assessment was conducted at all, it was done by the client. Implementation
and any potential summative evaluation was also the realm of the client. As
a result, I often found myself developing instruction for things that I knew
instruction would not help. Similarly, without the opportunity for summative
feedback, it was often difficult to tell what strategies worked well and which
ones did not. Also, without having much input on implementation, we often
saw our courses (which were the product of thousands of man-hours being implemented
in a slipshod manner that we knew would lead to failure). In spite of our
warnings to these clients - all of whom used the Dick and Carey model also
- the clients still did not perform these critical functions consistently
and properly. What I conclude from this is that either training departments
need to align their practices more consistently with the Dick and Carey model,
or they need to select a model that factors in the real-life, systemic factors
that guide how they do things. In other words, perhaps we need to get real.
I find that the model is somewhat arrogant in its assumption
that we can predict with accuracy at the beginning of a development process
what will work and what will not work (with the client, the SMEs, the learners,
and the learning context). It assumes that our analysis will be correct, and
if not correct, we will find out 6-12 months later when we begin formative
evaluation. Making such assumptions in a high risk, high-pressure environment
(where hundreds of thousands of dollars are involved developing a few hours
of instructional content) is dangerous and, as I have observed, prone to nasty
surprises.
So what are the alternatives? In my experience, rapid prototyping
(Thiagi, 1995 and others) offers tremendous opportunities to improve on the
basic elements of Dick and Carey. In this relatively new ID model, analysis,
design, development, and evaluation are ongoing, recursive processes. A client
comes to me and says, 'I need instruction for this.' I assemble a team, develop
a working concept, develop a working prototype or modify a pre-existing one,
and show it to the client. The client provides feedback and we negotiate changes.
Then, we go back, modify, develop, and go through the process again with the
client. At each interval, it is likely that more and more SMEs, documentation,
learners, and stakeholders will get involved, each fine-tuning the instruction.
The ID team and the client organization develop a close working relationship
where constant improvement is… well… a constant. In the Dick and
Carey model, small mistakes early in the process can slip through the cracks
and lead to expensive revisions in post-production (e.g., it could cost thousands
of dollars just to have a professional go back to the studio and record revised
narration). The Dick and Carey model does not allow for the fact that mistakes
commonly occur and people change their minds. We need to be more humble out
the development process. As chaos theory warns, stuff happens! In the rapid
prototyping model, mistakes are assumed to be there and are constantly being
uncovered and revised. Flexibility is built into the process so it is okay
for clients and even designers to change their minds (as long as they don't
mind fighting with the graphic artists and programmers :-).
One element of the D&C model that I would keep is the
idea of writing of test items before designing instruction. I would necessarily
hold up the prototyping process by insisting on writing the test items before
designing the instruction. However, I would advocate the conceptualization
of test items at the beginning of the process. In the absence of front-end
analysis (it's there but it's ongoing and trickling in throughout the process),
it may not be feasible to write learning objectives before meeting with the
client and SMEs, etc. However, I believe that it is both feasible and practical
to begin discussing the course in terms of how we are going to assess performance.
This question cuts through abstract discussions about goals, objectives, and
tasks, and gets right to a concrete and important design decision. I find
that SMEs are able to talk about testing more readily than they are able to
describe their knowledge. I often start with questions like, 'if I told you
that I knew how to do this job, how would you know if I was telling the truth?'
I find that giving SMEs a concrete scenario like that teases out the goals
and learning objectives more effectively than analyzing a bunch of documents.
From their answers, I am able to come up with an assessment concept. Typically,
it is performance oriented due of the nature of the SMEs response. If I used
a traditional task analysis approach, I find that IDers are more likely to
fall into the trap of always using multiple choice, true-false, matching,
etc. as a way of fitting the learning objective to the test, rather than fitting
the test to the learning objective. As you can see, this model is still a
work in process in my own mind, but I strongly believe than an assessment-first
mindset is an effective way of ensuring alignment between the instruction
and real-life performance goals (not just abstract behavioral objectives).
The rapid prototyping model does not address this issue and unfortunately,
Dick and Carey's attempts to place test items in front of content design have
been largely ignored. To summarize, thus far I have advocated the integration
of analysis, design, development, and formative evaluation into a recursive,
ongoing process that creates high stakeholder involvement, increases the level
of responsibility that clients will have in their courseware (hopefully, resulting
in better implementation), and creates flexibility in the process, acknowledging
the fact that stuff happens and we need to be able to respond quickly when
it does.
Lastly, while I would emphasize the importance of integrating
analysis, design, development, and formative evaluation, I recognize that
at some point the process must yield a "final product." I would
like to see summative evaluation broken down into more detail the Dick and
Carey model. Practitioners often try to get away with only worrying about
levels 1 and 2 of Kirkpatrick's model (1981?), ignoring transfer and impact.
If IDers, and more generally, the field of instructional design are to be
taken seriously in organizations, we must begin to emphasize the latter over
the former. I would flesh out the Dick and Carey model's treatment of summative
evaluation so that organizations can see the distinction and feel a little
guiltier about under-evaluating their programs.
In general, then, I advocate a model that is more realistic, in the sense that it recognizes the limitations of time, resources, and expertise (we aren't perfect) and promotes more emphasis on evaluation than on front-end analysis. If it is true that humans are goal-oriented, then it makes more sense to focus ID efforts generating frequent feedback (through rapid prototyping), focusing on testing first as a way of clarifying goals, and elaborating on summative evaluation.
For additional perspectives, visit some of the following
sites:
http://www.ittheory.com/timelin2.htm
Robert Maier's Instructional Technology Timeline
http://www.trainingsupersite.com/publications/archive/training/2000/004/004cv.htm
Gordon, J. & Zemke, R. (2000). Attack on ISD. Training.
http://itech1.coe.uga.edu/itforum/extra2/extra2.html
Merrill, D.Reclaiming the Discipline of Instructional Design. ID2 Research
Group, Utah State University
http://itech1.coe.uga.edu/itforum/extra2/disc-ex2.html
ITForum discussion dealing with Merrill's article: Reclaiming the Dicipline
of Instructional Design
http://www-cscl95.indiana.edu/cscl95/wiburg.html
Wiburg, K.M. (1995). An Historical Perspective on Instructional Design: Is
it Time to Exchange Skinner's Teaching Machine for Dewey's Toolbox?
http://home.istar.ca/~djcote/epss/software.htm
Discusses EPSS model as an alternative to traditional ISD and software design
models
http://www.personal.psu.edu/users/s/b/sbf116/2.htm
Sarah Fitzpatrick's web site: Description of Dick & Carey; Gerlach
& Ely; Leshin, Pollack, & Reigeluth Models
http://www.student.seas.gwu.edu/~sbraxton/ISD/id_models.html
Quick summary of several ID models, including: Hannafin & Peck; Jerrold
Kemp; and Rapid Prototyping
http://web.mala.bc.ca/lizhk/IDesign/ISDresources.htm
ID Resources (mostly constructivist)
http://www.thiagi.com/article-rid.html
Sivasailam "Thiagi" Thiagarajan's site... discusses Rapid Prototyping
model
http://carbon.cudenver.edu/~mryder/itc_data/idmodels.html
* University of Colorado: ISD Models
http://tip.psychology.org/
* TIP - Explorations in Learning & Instruction: The Theory Into Practice
Database
http://www.pcd-innovations.com/
* EdMAP is a large scale, enterprise-wide database specifically designed for
the development and management of instruction and assessment
http://www.bookstoread.com/e/et/id.htm
Huge list of instructional technology books (good for tracking down references)
http://www.dlrn.org/library/dl/guide4.html
Bloom's Taxonomy
http://www.fbe.unsw.edu.au/learning/instructionaldesign/materials.htm
ADDIE Model
http://www.indiana.edu/~idtheory/
http://www.indiana.edu/~idtheory/yellow.html
Charles Reigeluth's Green and Yellow Books - home pages
http://www.ittheory.com/timelin2.htm
Instructional Technology Timeline
Andrews, D.H.& Goodson, L.A. (1980). A comparative analysis of models of instructional design. Journal of Instructional Development, 3(4), 2-16.
Dorsey, L.T., Goodrum, D.A., Schwen, T.M. (1997). Rapid collaborative prototyping as an instructional development paradigm. In C.R. Dills & A.J. Romiszowski (Eds.), Instructional Development Paradigms (pp. 445-465). Englewood Cliffs, NJ: Educational Testing Publications.
Gerlach, V.C. & Ely, D.P. (1980). Teaching and Media: A Systematic Approach (2nd Edition). Englewood Cliffs, NJ: Prentice-Hall.
Gustafson, K. (1991). Survey of instructional development models. Syracuse, NY: Eric Clearinghouse on Information Resources (ED 211 097).
Kemp, J. (1985). The Instructional Design Process. New Yrok, NY: Harper Row.
Seels, B. & Glasgow, Z. (1990). Exercises in Instructional Technology. Columbus, OH: Merrill Publishing Co.
