COMPLEX PROBLEMS CLASS 2 I THINK, THEREFORE I SOLVE Lessons from Analytical Methods Analytical Disciplines Math, Physics, Operations Research, Economics, Finance, … » Utilize modeling techniques and tools (math, logic, abstraction) for well-structured problems » Overlap in procedures used » Borrowing methods for ill-structured problems Solving a “Word Problem” Problem: » In the U.S., temperature is typically reported in degrees Fahrenheit where boiling point of water is 212 and freezing point is 32. Most other countries and scientific endeavors use degrees Celsius where the boiling point is 100 and freezing point is 0. If the temperature in Rome is 7 degrees Celsius, what is it in Fahrenheit? Steps? » » » » Goal: Need to find conversion formula (C to F); plug in 7C Relevant information: 32F=0C; 212F=100C Illustration: Draw graph depicting Celsius on x-axis, F on y-axis Math concepts (words to equations): linear relationship so y=mx+b; use known points (0,32), (100, 212); “b” or “y-intercept” = 32F; slope = (y2-y1)/(x2-x1) or (212-180)/(100-0) = 1.8; F = 32 + 1.8*C; 7C = 44.6F. Generalizing Steps in Analytical Problem Solving Basics » Explicitly identify (write out) objective » Simplify (Abstract) – Eliminate extraneous-incidental information – Explicitly identify key information (objective; variables, values, …) » Organize Key Information – Mathematical representation, equation, table, illustration, lists, … » Perform appropriate math/logical operations If problem with multiple solutions » Assign probabilities or weights to each possible outcome » Calculate expected value or weighted value of each outcome » Compare values A Business Example For a (Relatively) WellStructured Problem Executive management must determine the best location for a new unit of a multinational company. Return on Investment and how well the new unit will fit organizationally should be the most important factors with the ability to attract and retain a suitable workforce a secondary consideration. The Capital Investments Committee has determined a short list of possible cities that includes Bangkok, Chicago, Sydney, Singapore, and Shanghai. Their ROI estimates for each city (in order) are 12%, 12%, 10%, 15%, 25%. Human Resources has assigned staff retention rate scores and organization fit scores on a scale from 1-10 (10 best) for each city. For staff these are 10, 8, 6, 4, 2 and for fit these are 2, 6, 10, 2, 4. The Travel Office has also calculated the following travel multipliers using Chicago as the base of 1.0. These multipliers are 2.0, 1.0, 1.8, 1.8, 2.2. Thinking about the problem What are the “well-structured” aspects? What are the “fuzzy” aspects making it just “relatively” well-structured? A Possible Solution ROI Weights Staff Organizational Retention Fit (0.4) (0.2) (0.4) 5 5 4 6 10 10 8 6 4 2 2 6 10 2 4 Total (1.0) Alternatives Bangkok Chicago Sydney Singapore Shanghai 4.8 6.0 6.8 4.0 6.0 Notes on Solution 1. The criteria used here are ROI, Staff retention and Organizational fit. Your criteria would reflect your values for this decision (this is a not-so-well structured part of problem) 2. Weights reflect the relative importance assigned to each of the criteria. This is another value judgement. 3. The scores assigned to the alternatives for each of the criteria should use the same range. In the above example, we have used a score out of 10 for each criterion. This required converting the ROI estimates. A simple ranking of alternatives on each criterion could have been used. 4. The weighted total is the sum of the alternative scores X the weights. For example, Bangkok’s total is given by the following calculation. *Weighted Total = (5)(0.4) + (10)(0.2) + (2)(0.4) = 2 + 2 + 0.8 = 4.8 Potential Problems The “fuzzy” parts of the problem » » » » Inappropriate limits on alternatives Weights/Probabilities are rarely known or known with precision Values (preferences) behind weights may be unclear or in conflict Data quality Analytical Techniques for Solving Harder Problems (including ill-structured) Analogy Solve in Parts Backward-Forward Transformation into Known Problem Solve for Simplified Case -- Generalize Problem-Solving by Analogy General example » X+Y Y+X Business Example » Managers have opened a store in Bowling Green, KY and use it as a template for store in Jackson, TN Solving by Parts General Example » Integration by parts Business Example » Large construction project such as Channel Tunnel -determine sequence of tasks (land; tunnel; rail; rail cars; ports of entry)… Backward-Forward (if needed) General Example » Detective working backward from evidence to criminal as well as from interviews of criminal to evidence » Proving right triangle XYZ with area z2/4 has 2 equal sides » Backward: Solution means x=y; so (x-y)=0; so (x-y)2 = 0; so x2-2xy-y2 » Forward: area = xy/2 = z2/4; x2+y2= z2 (Pythagorean); so xy/2=(x2+y2)/4; x2-2xy-y2 =0; » Business Example » Strategic Games -- looking ahead to rival’s best options – Stage 1: Company1 Innovate/Not Innovate – Stage 2: Company 2 Response: Aggressive, Moderate, Mild – Company 1 looks ahead to Stage 2 decisions for company 2 best on company 2 best action; trim decision tree Transform into Known Problem General Example » Stats: Male height is normally distributed with mean of 70” and s.d. of 2”, what is the probability of male > 74” -- transform into standardized units (mean = 0, sd =1) and use standard normal distribution » Differential Equations Business Example » Contemplating a contract regarding several contingencies based on performance or exogenous conditions -- transform into option pricing model using information or guesses about distribution of relevant contingent variables Solve for Simplified Case & Generalize General Example » Celsius-Fahrenheit example: solve for two points on line; extrapolate (generalize) to any points Business Example » Method for resolving inter-unit disputes developed for two units, expanded to entire company Additional Pitfalls in Analytical Methods for Ill-Structured Problems Analytical (Cognitive) Biases » Limited capacities confronting complex worlds » Not always clear how we are really thinking – mental shortcuts – limited introspective abilities rendering perceived analysis as little more than “rationalization” » It is not easy to change how we think – preconceptions; self-serving Examples of Cognitive Biases Strong Priors or Anchoring Bias Relying almost exclusively prior beliefs about the relationship between variables; not updating beliefs in the face of new or contradictory evidence A manager believes that firms that moving quickly always wins will keep doing so even when the firm is not doing well Analogy Bias Using an example gained from one situation to apply to another situation that appears similar but overstating and understating differences Companies that diversify into new markets often assume that the policies-strategies that worked in one setting will work in another Representative Bias Assuming that a result from a small sample is representative of a larger group or time period An investor who made a 200% return between 1995-2000 invests expecting this to hold into the future Mean Bias or Stereotyping Assuming that the average result holds for a specific individual case Control Bias Overconfidence in one's ability to control outcomes Thinking that market influences can be ignored with no Cognitive Biases (con’t) Framing Bias Making different decisions or giving different answers when the same problem or question is stated differently Choosing decision with a 95% chance of success; rejecting one with a 5% chance of failure. Escalation Bias Continuing with an action when it is rational to stop. Companies competing in bidding for an acquisition target will sometimes bid well beyond the rational value of that target Attribution Bias Improperly understanding factors contributing to your own or others decisions or outcomes (especially in self-serving ways) We’re successful because of strong management; We’re failing because of a poor market Availability Bias Making judgments based on how easily you can think of information that is relevant to the judgment Confirmation Bias Valuing information that supports belief & rejecting contrary Critical Lessons Analytical Thinking is Powerful » » » » clarifying objectives simplifying identifying & converting key information using logical/organizational tools “Tricks” of Solving Hard Analytical Problems » Analogy; Break into Parts; Backward-Forward; Transformation; Generalizing from Special Case Analytical Biases are Also Powerful » Self-Awareness Critical Mini-Assignment Come to class with » a workplace example of a problem solvable with analytical methods » a workplace example of a cognitive bias