Vous êtes sur la page 1sur 5

International Journal of Industrial Ergonomics 25 (1999) 97}101

Seven common reasons to not implement ergonomics
Martin G. Helander*
Graduate School of Human}Machine Interaction, ¸inko( ping Institute of ¹echnology, ¸inko( ping, Sweden

Received 19 November 1997; received in revised form 5 December 1997; accepted 1 September 1998

Introduction 1. Ergonomics is for design of chairs

In our pursuit of daily activities we interact with Is ergonomics teaching persons how to sit? How can
environments, machines and tools. No wonder any discipline with such a limited perspective be taken
everybody is a subject matter expert of ergonomics, seriously? =ho cares about chairs?
and no wonder our science is not taken seriously by
some. Ergonomics professionals are likely to hear Response: Some years ago the author visited an
many trivial comments and many objections to our air tra$c control center to survey ergonomic prob-
science. This paper is intended to expose some lems. Several controllers told us that they needed
lighter sides of ergonomics, but there is a serious new chairs. This was not what we had expected.
undertone with implications for our discipline. We Were there no problems in design of displays for
will dicuss seven common objections to ergonomics scheduling and routing and human}computer in-
and how we can project our knowledge so that teraction? Of course there were, but somehow ergo-
ergonomics/human factors is better used and com- nomics has become associated with chair design.
monly accepted in design. This is not necessarily the fault of ergonomics pro-
1. Ergonomics is for design of chairs fessionals but rather news media and shallow ad-
2. Ergonomics is only common sense vertising. It is unfortunate since there is a realm of
3. The research in ergonomics is too abstract to activities in ergonomics. Chair design is only one of
be useful them, and it is not very representative for the
4. People are adaptive, so there is no need for breadth of activities of our profession. When chair
ergonomics users test out chairs, they typically sit in a variety of
5. Information in handbooks cannot be used for di!erent chairs and unfortunately most users can-
design not tell the di!erence. If ergonomics design of
6. Laboratory and "eld experiments take too chairs makes no di!erence, this must mean that
long and are too costly ergonomics is not very important? If so, can the
7. Let us "rst design the technical system. Then unimportant deserve to be a science?
we consider ergonomics. Because they are easy to think of, physical as-
pects may always dominate the discussion of ergo-
*E-mail: mahel@ikp.liu.se. nomics. Information processing and decision

0169-8141/99/$ - see front matter ( 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 9 - 8 1 4 1 ( 9 8 ) 0 0 0 9 7 - 3
98 M.G. Helander / International Journal of Industrial Ergonomics 25 (1999) 97}101

making are much more illusive. Chair design and one must "rst understand mathematics. Since
noise exposure are good examples of physical aspects, people do not understand } they do not question
that users have opinions about. They may complain mathematics.
about chair design but will not consider the di$culty As it turns out psychology and ergonomics are
in making decisions as they sit in the chairs. Users will no more common sense than mathematics. The
complain that the noise level is too high, but there common sense about ergonomics is usually quite
is no mention of increased arousal levels and nega- super"cial, but as professionals we must be on our
tive e!ects on human information processing. guard and we must be able to explain our science. It
Users are unlikely to complain about informa- is a young science and explanations are necessary
tion design. For example, users will rarely complain } HCI is only 20 years and biomechanics 50 years.
about display design, because the implications of
display design are di$cult for users to analyze. We 3. The research in ergonomics is too abstract to be
(the users) have limited capability for understand- useful
ing how long it takes to process information; our
time perception is very poor. In addition, we have I cannot understand the articles in Human Factors
di$culties in thinking of alternatives for how the and Ergonomics journals. I cannot use them for
information could be displayed. One might assume anything practical.
that we would be well aware of our vulnerability
} that our personal experience of problems with Response: Wait a minute! Who says that you will
limitations in short-term memory and inattention have instant understanding of a "eld that has pro-
} would make us prioritize information design duced a million scienti"c articles.
} but that is not how it is. Instead our inability to There is room for basic research and applied
consider information design makes it less likely to research in every "eld. We must not qualify re-
report the need for information design. What is search by how applied it is. The purpose of journal
di$cult to see or di$cult to feel is also di$cult to articles is to advance our knowledge.This is a grad-
think of and will not be complained about. ual process, and it will take time before we under-
In our desperation we asked an air tra$c con- stand which basic research "ndings are important.
troller if he forgot to mention information over- There are many examples of scienti"c contributions
load? And the controllers responded: ` Sorry, what that must have seemed unimportant at the time
are you talking about? What did we forgeta The they were published, but turned out to be very
very limitations of the attention span and the important in the longer perspective } because they
short-term memory that makes information design were eventually applied in systems design.
so important, also makes it di$cult to report un- About 40 years ago research was published on
satisfactory design. engineering models for manual control of vehicles.
These models are today used in simulation analyses
2. Ergonomics is only common sense of aircraft handling and to investigate the steering
and stability of automobiles. The models are highly
=hy would anybody like to hire a consultant of abstract and their success could not have been
common sense? =hy would anybody like to take predicted in the 1950s.
a class in common sense? Vannevar Bush (the president of MIT) in 1948
published a paper on models for human associ-
Response: We are faced with the common prob- ation. This paper is quoted as the inspiration of
lem of common sense. It is not only in ergonomics. Hypertext, which 40 years later became the stan-
All behavioral sciences that deal with everyday dard format for Web information.
things are exposed to the same oversimplifying Many innovations in HCI, such as the mouse and
accusations. The physical sciences are better o!. direct manipulation interface were documented in
They shroud themselves in quantitative abstrac- the 1960s at Xerox Parc in California and imple-
tions. To have serious claims against mathematics mented about 15 years later.
M.G. Helander / International Journal of Industrial Ergonomics 25 (1999) 97}101 99

Signal detection theory has not only changed improves: it takes less and less time to perform the
considerably our view of sensory measurement, but task, and they commit fewer errors. However, sys-
the theory is now used for many important applica- tems with compatible, ergonomics design is always
tions, such as evaluation of radiographic imagery. better } in the short-term as well as in the long
Anthropometry has been used for years, but it is term. Even after 5 years use a compatible design
only recently that full body models have been de- will be faster than a non-compatible design. How-
veloped. These provide complex modeling tools ever, users do not complain, so the problems ex-
} that are di$cult to understand for the public perienced due to poor design are not emphasized.
} but extremely useful for design. There are several reasons. Sometimes users are re-
Research in human decision making has been used luctant to complain because of loyalty to the com-
to develop computerized tools that can aid human pany, sometimes they do not realize that there are
performance in complex environments such as nu- better design alternatives, and sometimes they
clear power plants, process industries, and aviation. think it is a job requirement to always perform well
Basic research in organizational management and handle all situations.
has been used to develop new models for commun- A classic story told by Dr. Richard Pew of an
ication in organizations, which can be used to es- experiment reportedly conducted at the Bell Labor-
tablish virtual companies. atories during World War II illustrates this:
As these examples show the critique to become The Telephone Company was concerned about
more applied is unwarranted, and this seems parti- ways to save copper, and one use of copper was in
cularly true for ergonomics. Our research typically the telephone cord. The experimental question was:
grows out of practical needs, and researchers in How long does the telephone cord need to be? The
ergonomics are, in fact, in a much better position to investigators identi"ed a series of test telephones
produce useful applied research than many other around the laboratory. Each night they went to
disciplines. The problem today in ergonomics is those telephones and cut o! the length of the cord
maybe the contrary } an overemphasis on short- by 1 in and replaced the telephone in its normal
term practical applications and not enough long- position on the desk. They also designated a special
term basic research. Government and industry re- telephone operator to receive complaints about the
quire instantaneous applicability; with the e!ect telephones. Day by day the lines got shorter, and
that super"cial development is promoted and not one by one the users began to complain to the
basic understanding. special operator. At the end there was one person
We need both competent basic research and ap- who had not complained. The investigators decided
plied research that can explore the practical use of to check up on him. One of them called while the
ergonomic concepts. There can be no controversy other visited his o$ce. When the telephone rang, he
between the two } both are necessary. leaned over his desk so that his ear could reach the
receiver. It was a very awkward posture. When
4. People are adaptive, so there is no need for asked later if there was anything funny about the
ergonomics in design of systems telephone he said: `Oh, the cord is a little short, but
that does not bother mea.
People are adaptive and they learn to operate even People are very adaptive and many will never
the most complex system. ¹hink of airplane pilots complain, but this does not justify poor design.
} they are trained to use the system!
5. Information in handbooks cannot be used for
Response: Maybe so, but with better design they design
would learn it even better, and there would be less
This handbook is far too theoretical and does not deal
errors in operation and less accidents.
with my problem!
There are many examples of excellent basic re-
search that invalidates the argument. As people Response: For once you may be right! In many
learn a system and work with it their pro"ciency other "elds it is far easier to use theories and
100 M.G. Helander / International Journal of Industrial Ergonomics 25 (1999) 97}101

models to predict behavior. In ergonomics there modi"cation of an old product should not come as
are not very many models. But there are concepts, a surprise to the employees. There is usually ample
principles and two (!) laws. Fitts' law makes it prewarning for the ergonomist to schedule ahead
possible to predict human movement time to a tar- and also do experimentation ahead of time. But the
get. From Hick's law we understand that decision experimentation must be wisely planned. We must
making takes longer if there are many choices investigate issues that are of practical signi"cance.
rather than few. This type of information is useful If there are di!erences that are practically signi"-
for handbooks, but it is not enough. Much informa- cant we can also count on statistical signi"cance.
tion is missing. For example, given that I want to The trained experimenter may insist on a complete
apply Hick's law, how do I calculate the number of symmetric factorial design of the experiment. These
decision alternatives? are, however, costly and time consuming. Some-
We sorely need to develop more models for top- times there is time for only one quick experiment. If
down design of systems. To be useful for systems so, let it be quick but not so dirty! With careful
design, these models must be stated in a general planning, thought and analysis one can reduce the
form, so that they are applicable to a range of complexity in experimentation. The goal, as we
design problems. mentioned, should be to obtain results that make
Guidelines are in the same category as hand- a practical di!erence in design.
books } they are usually di$cult to use. The prob- Experimentation may however be the wrong ap-
lem here is primarily with guidelines formulated at proach. Designers are often more interested in un-
a low level of abstraction. As an example Smith and derstanding options for design and will typically
Mosier published in 1988 Guidelines for Designing ask questions such as:
;ser Interface Software. This contains 944 guide- f What are the constraints in design and is there
lines for design of user interface software. Users of room for negotiation of the constraints?
these guidelines typically have di$culties in under- f What are the experiences from the old system?
standing how to apply the guidelines to their prob- f How much training will be necessary with the
lem. It is like looking at photographic slides new system?
photographed by someone else. One can see what it f Why did users not like the previous system?
is } but one cannot understand the context in which f How can we avoid user errors with the new
it was taken and one cannot understand the rel- system?
evance for the present design. It is probably better if These are questions of why, what and how. They
guidelines can be formulated at a high level } prin- are not answered by collecting data, rather by quali-
ciples of design rather than low level guidelines. For tative assessments and comparisons of systems.
example: Computer users need to receive feedback on We must, however, be prepared for serious criti-
their input. cism from individuals, who take little interest in
This design principle has to be interpreted consid- experimental studies } and ergonomics. Few out-
ering the peculiarities of task and the user. The inter- siders have expressed their dissatisfaction as vo-
pretation and implementation is up to the designer. cally as the late US Admiral Hyman Rickover, who
concluded: Human factors are about as useful as
6. Laboratory and 5eld experiments take too long teaching your grandmother to suck an egg!
and are too costly
7. Let us 5rst design the technical system + then we
The study you propose takes too long and is over will consider ergonomics
budget. I need the results by tomorrow!
We must xrst know whether the system can be
Response: This comment is common in industry designed at all. ;sability will have to wait!
where the development of new products imposes
deadlines for product design and evaluation. How- Response: Too little too late } as usual. You
ever, the development of a new product or the engineers need also to understand about concur-
M.G. Helander / International Journal of Industrial Ergonomics 25 (1999) 97}101 101

rent engineering and one important aspect is ergo- may sound it seems like a very good idea. Let us
nomics. Ergonomics design must be developed in develop systems that are so intuitive and self-explana-
parallel with other design considerations, such as: tory that they can be handled without a manual.
manufacturability, reuse, maintainability and so
forth. Ergonomics design is crucial and must be
considered at an early design stage. I do not agree with some of your points!
For design of computerized systems there are
clear indications from the research literature, that Response: Don't worry. We may have di!erent
the user must be considered at the earliest stage experiences. Maybe we can agree on the basic
possible, when the system is conceptualized. In de- issues } the di$culties and intricacies in promoting
velopment of software, the user interface usually and justifying ergonomics. To us, the professionals
takes us about half of the programming e!ort. The in ergonomics, ergonomics will always be neces-
ergonomics/usability expert must therefore partici- sary. It is a compassionate, humanistic approach,
pate in the concurrent e!ort to produce concepts which produces high quality design with high bene-
for systems design. Often, however, the ergonomist "t/cost, greater productivity, greater user satisfac-
is not invited, which has expensive rami"cations. tion an a reduction of injuries.
Thomas Landauer has noted that 60% of software
maintenance is devoted to removing `usability
bugsa, and there are clear economic rami"cations You missed many points!
of not "xing the usability bugs.
To obtain a clear user focus in the early systems Response: The author will not be defensive. As the
development, John Gould proposed that one title indicates the paper gives as many examples as
should "rst write the user manual and then the he could think of in a #ash } given the limitations
software. This would achieve three important of the short-term memory. Please write down
things: early involvement of ergonomists, improved your experiences and comments and send to
usability and a short manual. As provocative as this mahel@ikp.liu.se.