Lecture 10: Human Error I

1
Lecture 10 Human Error I
09/10/99
2
Goals

• Introduce types of errors
• Understand differences between error types with
  examples
• Motivate user-centred design to reduce error
• Reading Recommended text (Norman)

3
Introduction

• People make errors all the time.
• Dial wrong numbers
• Forget peoples names
• Turn on wipers instead of turn signals
• Misplace things

09/10/99
4
Errors Causation

• Errors arise from wrong action
• Problems at different stages in the action
  sequence lead to different types of error

5
The Action Sequence

• Forming the goal 
• Forming the intention 
• Specifying an action 
• Executing the action 
• Perceiving the state of the world 
• Interpreting the state of the world 
• Evaluating the outcome 

6
Actions as dialogues with the world
Outcome
Goal
State
Human
World
Intention
Action
Specification
7
Two Main Types of Error

• Slips
• Mistakes
• Slip Correct Goal, Bad Execution
• Mistake Wrong Goal

8
What is a Slip?

• "Slips result from automatic behavior, when
  subconscious actions that are intended to satisfy
  our goals get waylaid en route
• E.g., someone who unbuckled his watch instead of
  his seatbelt when about to get out of a car.
• Slips are usually easy to spot

9
What is a mistake?

• Wrong goal
• Arising from conscious deliberation
• E.g., False generalizations (plenty of gas in
  tank when my old car said empty)

10
Different Types of Slip

• Capture errors 
• Description errors 
• Data-driven errors 
• Associative activation errors 
• Loss of activation errors 
• Mode errors 

11
Capture Error

• Frequent activity captures intended action
• E.g., , you get into your car to go to the store
  and you end up going to your office.

12
When and Why do capture errors occur?

• two different action sequences have their initial
  stages in common
• one sequence is unfamiliar, the other well
  practiced.
• start off with the unfamiliar sequence and switch
  over to the familiar sequence unintentionally.
• Switching from the familiar to the unfamiliar
  sequence hardly happens, if ever. 

09/10/99
13
What is a description error?

• the internal description of the intention is not
  sufficiently precise.
• correct action on the wrong object.
• Similarity between wrong and right objects
  increases Pr(error).
• Physical proximity also increases Pr(error)

14
Examples of description errors

• Put lid on coffee cup instead of sugar bowl
• Pour oil into measuring cup instead of rice
• Pressing release instead of hold (similar
  buttons) drops call
• Poor design -gt long rows of identical switches
• When different actions have similar descriptions,
  there is a good chance of mishap

09/10/99
15
What is a data driven error?

• automatic action intrudes upon an ongoing action
  sequence.
• E.g., reading a number on the wall in front of
  you
• -gt the (different) phone number you are dialling
  has changed based on the number just read

16
What is an associative activation error?

• internal thoughts and associations trigger an
  inappropriate action.
• E.g., saying "come in" when you pick up the
  telephone
• (confusing the normal response to a knock on the
  door with what you should do when you pick up the
  telephone). 

09/10/99
17
What is a loss of activation error?

• when you leave out part of the action sequence.
• E.g., switching on the computer and then
  wondering what it was you wanted to do

18
Example of loss of activation error?

• I have to go to the bedroom before I start
  working in the dining room.
• no idea why I am going there.
• I get there but still cannot recall what I
  wanted
• go back to the dining room.
• realize that my glasses are dirty.
• go back to the bedroom,
• get my handkerchief,
• wipe my glasses clean." 

19
What is a mode error?

• when devices have different modes of operation,
• action appropriate for one mode has different
  meanings in other modes.
• Occur when equipment has more possible actions
  than it has controls or displays
• so controls must do double duty.
• E.g., using a mail system and you want to reply
  to a message. You forget you are already in reply
  mode.
• You type Command-R and nothing happens.

20
Detecting Slips

• Usually easy, because feedback is obvious
• Where feedback is unclear people may persevere
• E.g., trying to unlock someone elses car
• (thinking it is your own)

21
Mistakes as Errors of Thought

• Mistakes are not as clearcut or easy to
  characterize as slips.
• Problems in thinking lead to mistakes.
• Applying good principles of design reduces
  likelihood of mistakes and slips.
• human error causation outside this course
• (see book by John Reason,cited in the references
  for the Web lecture).
• accidents result from complex mix of mistakes and
  unexpected events.

22
Design as Culprit in Human Error

• Bad Design makes it easy to make wrong settings,
  misread an instrument, misclassify an event.
• Poor Design of social structure makes false
  reporting of danger punishable.
• E.g, Turn a nuclear power plant off by mistake
  costs hundreds of thousands of dollars
• Fail to turn it off when there is a real
  incident, and you might lose your life.
• refuse to fly in bad weather, company loses lots
  of money and passengers get angry.

23
Designing for Error

• Error is natural in human behavior.
• Understand the causes of error and design to
  minimize those causes. 
• Make it possible to reverse actions - to "undo"
  them - or make it harder to do what cannot be
  reversed. 
• Make it easier to discover the errors that do
  occur, and make them easier to correct. 
• Change the attitude towards errors.
  (approximations of required actions)

24
Errors are often "sensible

• incomplete or misleading information (mistake) 
• poor design or distraction (slip) 

25
User-centred design for Error

• Put the required knowledge in the world
• Use the power of natural and artificial
  constraints
• Narrow the gulfs of execution and evaluation.

26
Put the required knowledge in the world.

• Don't require all the knowledge to be in the
  head.
• Allow for more efficient operations after
  knowledge is in the head. 

27
Use the power of natural and artificial
constraints

• physical, logical, semantic, and cultural.
• Use forcing functions and natural mappings. 

28
Narrow gulfs of execution and evaluation.

• Make things visible, both for execution and
  evaluation.
• Make options (execution) readily available
• make the results of each action (evaluation)
  readily apparent.
• Make it possible to determine the system state
  easily
• In a form consistent with the person's goals,
  intentions, and expectations.

29
Lesson

• Design for Error
• Apply user-centred design principles
• Design for error in your projects