Lecture 1: Common Cognitive Bias and Problem Complexity Flashcards
Common cognitive biases
Systematic patterns of deviation from norm or rationality in judgment
Criteria for a good decision process
- Procedural rationality: procedure that leads to the final decision can be more or less rational
- Rational preferences: consistent assumptions in the decision scenario
Procedural rationality
- tackle the right problem
- put the appropriate effort in information search
- derive proper expectations using relevant objective data
- think carefully about objectives and preferences
Rational preference
- Prospective orientation
- Completeness
- Transitivity
- Independence of irrelevant alternatives
- Invariance
Prospective orientation
Decisions should be forward looking -> ignore sunk costs
Completeness
Decision maker has a preference for any pair of alternatives (indifference is allowed)
Transitivity
If a is better than b and b is better than c, then a should be better than c.
Independence of irrelevant alternatives
Given the set of alternatives {a, b}, if a is better than be, then for the set {a, b, c}, a should still be preferred over b.
Invariance
Preferences should not be affected when the same problem is presented in a different way. (Example: 200 are alive or 400 die)
Why is it important to characterise a good decision?
- Check whether you have done the best job given the resources spent
- Can you justify the decision even if a bad outcome was realised?
Decompose complex decisions (4 Steps)
- Objectives (what do you want?)
- Alternatives (what can you do?)
- Uncertain events (what do you (not) know?)
- Outcomes (what is the impact of choices?)
Do we know our objectives?
No, people tend to generate an inadequately narrow set of objectives
Where to look for objectives?
- Obvious insufficiencies: Things your are currently not happy with
- Comparison of alternatives: Differences between alternatives might help you to realise what is important to you
- Strategic goals: Some objectives are not directly linked to your current decision but of overall interest (e.g. company reputation)
- External guidelines: Guidelines should indicate an important objective
- Impact of other people: Consider the objectives of other people who might be affected by the consequences of our decision
Fundamental objectives
- relevant for their own sake without further justification
- Example (vehicle regulations): maximise safety, minimise loss of life, …
Means objectives
- relevant because they help to achieve a more fundamental objective
- Example (vehicle regulations): Educate public, more traffic lights, …
Objectives hierarchy
- Diagram of relationships between objectives, sub-objectives and attributes
- All objectives in a hierarchy should be fundamental
- Most general objective at the top
- Higher levels represent more general objectives
- Lower levels represent more specific statements
- Lowest levels represent attributes based in which the alternatives will be compared
- Attributes are used to characterise performance in relation to an objective
Criteria for an ideal set of fundamental objectives
- Completeness: all fundamental objectives are included
- Simplicity: Small set of objective is easier to communicate and estimate
- No redundancies: Avoid double counting or overlapping objectives
- Preference independence: preferences of a subset of objectives should be independent of the attribute levels of the remaining objectives
- Measurability: achievement of objectives should be measurable
Attributes
- unambiguous ratings of how well alternatives do with respect to each objective
- quantitative or qualitative
- useful for describing consequences of alternatives and making value tradeoffs between objectives
A good attribute should be … ?
- Direct
- Operational: low cost of information collection
- Comprehensive: attribute levels cover the full range of possible consequences
- Understandable: units should make sense to the DM
- Unambiguous: we should have the same interpretation of the same attribute level
Types of Attributes
- Natural
- Zero or One Measures
- Constructed Measure
-> Prefer natural attributes!
Natural Attribute
- Is in general use with a common interpretation by everyone
- E.g. cost in dollar, weight in kg, time in hours
Zero or One Measures
- Describe whether the alternatives have a specific attribute or not
- E.g. Diploma (yes or no)
Constructed Measure
- Is developed for a particular decision problem
- E. g. five point scale to measure ease of use
- Group natural attributes to obtain a discrete attribute when the natural attribute is to specific (e. g. excellent, good, ok, bad)
Goal of decision analysis
The goal is to make better, more rational decisions.
- Prescriptive decision theory aims to teach the method and concepts that support and improve rational decision making
- Descriptive decision theory aims to raise your consciousness about decision-making by pointing out the errors and biases in it
