Mistake Mitigation Tool

1
Mistake Mitigation Tool

• Douglas M. Stewart, Ph.D.
• Anderson Schools of Management
• University of New Mexico

2
Error Proofing in Services

• Services are less amenable to traditional
  variation-based quality improvement techniques.
• Errors are a prime cause of service failure.
• One successful approach has been Service
  Failsafing.
• Chase and Stewart, 1993, 1994, 1995 Stewart and
  Chase 1999, Stewart and Grout 2001.
• What Failsafing has been unable to address are
  decision making errors.

3
Known Biases in Decision Making

• General Biases
• Representativeness - Use of stereotypical traits
• (Tversky and Kahneman, 1974, Kahneman, Slovic and
  Tversky, 1982)
• Availability Emotional, vivid, recent, more
  frequent
• (Tversky and Kahneman, 1974, Kahneman, Slovic and
  Tversky, 1982, Johnson et al. 1993)
• Joint and Separate Evaluation Easy to evaluate
  attributes will be more important in separate
  evaluation.
• (Hsee, 1996)
• Anchoring and Adjustment
• (Slovic 1967, Slovic and Lichtenstein, 1968,
  Lichtenstein and Slovic 1971, Tversky and
  Kahneman 1974, Griffin and Tversky 1992, Chapman
  and Johnson 2002)
• Prospect Theory Decreasing returns to losses or
  gains, but losses hurt more. High probability
  gains (losses) produce risk aversion (seeking).
• (Kahneman and Tversky 1979, Tversky and Kahneman
  1992)

4
Known Biases Continued

• Preferences for Sequences of Outcomes Utility
  levels that increase over time.
• (Loewenstein and Prelec 1993)
• Anticipatory Regret dont sell that lotto
  ticket
• (Janis and Mann 1977, Bell, 1982, 1983, 1985,
  Kahneman and Tversky 1982, Fishburn 1983, Miller
  and Taylor, 1995)
• Ambiguity Aversion People prefer known rather
  than unknown probabilities.
• (Ellsberg 1961, Fox and Tversky 1991)
• Hedonic Framing Segregate gains, integrate
  losses, segregate smaller gains from larger
  losses, integrate smaller losses with larger
  gains.
• (Thaler 1999)

5
Known Biases Continued

• Endowment Effect I wouldnt pay what Id need
  to part with it.
• (Knetsh, 1989, Thaler 1980, Kahneman, Knetsch and
  Thaler 1991)
• Sunk Costs are not ignored (e.g. shoes)
• (Thaler 1980, 1999)
• Opportunity costs Foregone gains that are
  weighed less than out of pocket losses.
• (Thaler 1980, 1999)
• Transaction vs. Acquisition Utility The deal
  matters
• (Kahneman and Tversky, 1984, Thaler, 1985, Thaler
  1999)

6
Known Biases Continued

• Diversification Heuristic Give them three
  options and theyll take 1/3 of each.
• (Thaler 1999)
• Reluctance to close a mental account in the red.
• (Odean 1998)
• Payment Decoupling hide link between payment
  and consumption, pay first.
• (Prelec and Loewenstein, 1998, Thaler 1999)
• Wealth accounts (cash on hand, current wealth,
  home equity, future income) tend to be analyzed
  separately
• (Thaler 1999)

7
Known Biases Continued

• Money Illusion Tend to think in nominal (not
  real) dollars and rates of return.
• (Shafir, Diamond and Tversky, 1997)
• Timid Choices Tend to evaluate choices
  separately when risk is pooled
• (Kahneman and Lovallo, 1993)
• Planning Fallacy Tend to be overly optimistic
  (focus on task and ignore those things that can
  interfere.)
• (Kahneman and Lovallo, 1993, Buehler, Griffin and
  Rose, 1994)

8
A Proposed Method for Mistake Mitigation

• Flowchart
• Questionnaire for decision points
• Force Field Analysis
• Bias Removal
• Knowledgeable Correction

9
The Questionnaire

• Questions for each bias
• Scaled so total for each bias runs -1 to 1
• Relative strengths are unknown
• All questions for each bias considered as a
  single bias.
• See spreadsheet
• Difficulty 1 Handling single value/probability
  selection vs. options
• Currently Use two options higher value lower
  value
• Difficulty 2 Handling more than 3 options may
  make bias less clear

10
Concluding Thoughts

• A preliminary approach
• Evaluate more decisions to refine and modify.
• Strengths of the approach
• Wealth of knowledge about biases should be tapped
• Approach appears workable
• Provides a framework to structure problem
• Makes more obvious biases that are concealed
• Works for both customer and provider decisions

11
Concluding Thoughts (continued)

• Weaknesses
• Do not know relative strengths of effects
• Can not trade off positive for negative biases to
  cancel them out.
• Familiarity with theory necessary to interpret
  questions
• May be possible to refine questions further.
• Not all biases included
• Known Fairness biases
• Unknown
• Some biases may be removed
• Different decisions made often vs. similar
  decisions made infrequently
• May be able to segment questions into those
  biases that pertain across all decisions (and
  de-bias these) and a shorter list for the unique
  aspects of each decision.
• May be able to identify commonality of biases
  (for removal) even though decisions are unique.