norman (01/01/83)
January 1, 1983
REQUEST FOR REPORTS OF MISTAKES
I am seeking reports of mistakes that you may have made or observed. I
believe that such reports can be of great assistance in determining the causes
of human error, perhaps thereby allowing us to design things so as to minimize
error. I need a source of naturally occurring events, but reported with great
detail and accuracy.
I classify errors into two major categories: slips and mistakes. A slip
occurs when the action that was intended does get done properly. (Example:
intending to empty the dirty glass while sipping a cup of coffee, finding
yourself sipping from the dirty glass and emptying the coffee.) A mistake
occurs when the intention itself is wrong. [Examples: turning off the auxili-
ary feedwater system when it automatically turned on (incident at Three Mile
Island; filling the "toner" bottle on a graphics printer with the wrong fluid
("premix").] This study covers mistakes only.
I want only first hand reports, written down or dictated as soon as pos-
sible after the incident. Second hand reports, and reports made after the
passage of time, tend to lose accuracy and to be incomplete and cannot be
relied upon. I would like the report to be as thorough and complete as possi-
ble. Details are essential. I especially need to know about the cir-
cumstances under which the error occurred: were there time pressures, stress,
emotions? Were there social pressures? Other factors? How was the mistake
discovered? How long after the incident? By whom? What were the effects of
the mistake?
Protection of Sources
---------- -- -------
Although I want your name and address so that I can check back with you
for details, I will protect the sources and names. I will destroy the name of
the contributor shortly after receiving the contribution unless we are still
corresponding. I will edit all contributions to disguise the identities of
the people involved. It is usually very embarrassing to be the cause of an
error, and so people are often reluctant to report them. However, be relieved
to know that everyone makes mistakes and that a full and complete report will
aid in our scientific understanding of how they come about. Your identity
will be protected. First person reports are encouraged.
Related Prior Research
------- ----- --------
(This is both to establish my credentials and to serve as examples of how
the results get used.) This study follows similar ones on slips that have
been useful in developing guidelines for human-machine interaction. See
"Design rules based on analyses of human error." Communications of the ACM,
1983, in press. See also my articles in Psychological Review, 1981, Vol 88,
1-15 [with a simplified version in Psychology Today, 1980, (April) and an even
more simplified version in Reader's Digest, 1980, Vol 117 (August), 103-106].
Form of the Report
---- -- --- ------
If possible, follow the following outline:
"The Time of of Both the Incident and of Report"
"Relationship of Reporter to the Incident"
"Background Material"
"Special Circumstances"
"The Incident"
"How the Error Was Found"
"Sources of Difficulty"
"Eventual Outcome, Including Damage, Repercussions, Etc."
(I am not including any sample reports because of the length: I will forward a
sample upon request.) Send the report to:
Donald A. Norman
Cognitive Science C-015
University of California, San Diego
La Jolla, California 92093
(Computer mail can be sent to: ucbvax!sdcsvax!norman or norman@nprdc)