Discussion

Decisions - Heuristics.html

Decision Making - Heuristics Dr. Karen Klisch

edited 5/2017

Except where noted, all of the information in this lecture is from: Bazerman, M. H. (2006). Judgment in managerial decision making. (6th ed.). New York, NY: Wiley & Sons, Inc.

Bounded Decision Making

What does "bounded" decision making mean? Simply put -- it means that there are limits -- boundaries -- to our decision making that may keep us from making the best most logical most rational decision. Even when we think we are being completely rational and logical we often are not. Even when we make a concerted effort to be logical and rational -- we are not. How can this be? What can we do to make this not so? What do we do that makes this happen? Ideally, we should make the most logical rational decision when we have issues to resolve. Let us first deal with some basic definitions from Bazerman.

• Rationality -- is the logical decision making process that is logically expected to lead to the optimal result.
• Bounded rationality -- is the attempt one makes at this rational logical decision making but does not succeed. Most decisions have some degree of boundedness even when one claims to be making a completely rational decision.
• Satisficing -- is the result of bounded decision making. People simply forgo the best most rational logical decision for the first acceptable or reasonable solution.

Reasons Bounded Rationality Occur:

• Important information is not available or simply lacking
• The definition of the problem is not completely clear
• Relevant criteria are lacking
• Lacks time
• Cost is prohibitive
• The individuals making the decision have limited memory or intelligence

There are two other "bounds" -- both of which will be dealt with later in the course, but will be mentioned here. Bounded Ethicality The degree to which we behave ethically is limited  -- that is -- bounded. Bounded Awareness We easily fail to notice important information that is easily available to us, that results in limited decision making. System Thinking It would be wonderful if people actually followed the format described above when making decisions. The fact is --- sometimes we do use this process and the decision is a very logical one, but most of the time we do not. The following two approaches to decision making explain the "reasoning" we tend to use. System I This way of making a decision emanates from our intuition. The decision is made fast. It is "automatic, effortless, implicit and emotional" (Bazerman, 2006, p5). Many of the decisions we make in life are made using this process. If we are stressed, super busy, distracted -- we use this process. We do not stop and give the matter much thought. We simply react. In many situations this process is OK. Certainly things like making the bed, grocery shopping, selecting a TV show are things that will not be detrimentally affected by not giving the decision much thought. But if we want to make good logical decisions about more challenging issues we need to use another process. System II System II thinking is very much the opposite of System I thinking. It is reasoned, conscious, slower, effortful, explicit and logical. It is important for managers and leaders --- (everyone for that matter)  -- to know when to move from System I thinking to System II thinking. Using System I thinking for complex decisions will increase the probability that the decision will be fraught with bias. There is no guarantee that bias will not creep in when using System II, but the likelihood of at least reducing the bias and perhaps tremendously minimizing the bias is greater if one stops and determines when System II is best utilized. HEURISTICS We use heuristics to help us make decisions. Heuristics are patterns of behavior and thought that allow us to make quick decisions simply, when we are dealing with a complex world.  They can help us make very good decisions --- and they can lead to horrible mistakes. Heuristics often result in a decision that is terribly biased. It is unlikely that too many people, if asked do they make biased decisions , would say yes. Generally people like to think that biases are not a part of their usual behavior. However, we all have many biases, we do make biased decisions, and we are, almost all of the time, completely unaware that we are doing it. It happens because we rely on  heuristics. Let us take a closer look at some of these heuristics.  Availability Heuristics Ease of Recall Bias  (vividness and recency) This heuristic explains how much, recent and vivid events affect our decision making process. The examples Bazerman uses are excellent.  His question asking the rank order from most deaths to least, of a list of causes of death  --- tobacco, poor diet and physical activity, motor vehicle accidents, guns, and illegal drug use typically resulted in people choosing motor vehicle accidents as the leading cause of death. In fact tobacco is by far the leading cause. However we hear about auto accident deaths way more frequently than direct information on tobacco deaths so we fall prey to the ease of recall heuristic.      An example of this heuristic occurred several years ago between me and my business partner. She had been asking me to get a cell phone. I was not interested in having a phone with me. I did not like the idea of others having such immediate access to me. (Of course, now I do have one -- almost unavoidable, but back then I resisted). One day a young couple was driving around the beltway and as they headed north up the interstate into Maryland they ran off the road into a cluster of trees. The car was not visible to the road and they were trapped in the car. After a couple of days the man decided to light a fire -- thinking it would be seen and they would be rescued. Alas, they were seen, but too late to save them. They were killed in the fire. When that hit the news my partner said, "You have to get a cell phone. This could happen to you. Suppose you drove off the road and no one knew you were there? You need the cell phone so that you could call for help. If these two people had a cell phone they would be alive today."      The probability of me running of some road into woods where the car could not be seen has to be minuscule. But the recency of the event and the vividness of it made my partner think it was urgent for me to get a cell phone. It was not the most logical suggestion considering the actual likelihood of that happening to me. And for a moment I did consider getting a phone --- until a calmer thought process took over. I now have a cell phone, but NOT for that reason -- although if that should ever happen I am prepared. Retrievability Bias This heuristic is best illustrated with the location of gas stations. It is routine for several to be located very near to each other if not directly opposite each other. I have often wondered why they do this. How can they all make money with so many in one place? But consider --- when someone asks you for the location of a gas station, you are much more likely to think of the area where there are several, rather than remembering the lone station out on some country road. The retailers rely on this heuristic to draw people to their store. Presumed Associations Bias This bias takes some focused thinking to really understand. It involves the connections between two separate events.  When the two are connected they can create a quick and believable bias that is totally wrong. The easiest example is the one Bazerman provides. It connects drug use with delinquency. Question -- Does drug use lead to delinquency?      There are those people who will say absolutely. It seems right to believe that some sort of illegal drug use would lead one to become a delinquent. We hear this all the time, and it is so reasonable to believe it ---- or is it?      Before we can actually say drug use leads to delinquency we must consider the following:

1. How many drug users become delinquent?
2. How many drug users do not become delinquent?
3. How many delinquents do not use drugs?
4. How many non delinquents do not use drugs.?

     Until we know the answers to these four questions we cannot automatically say drug use leads to delinquency. But we do. We do agree without thinking because of this thing called heuristics.

1. Are couples under the age of 25 more likely to have bigger families?
2. Do small colleges produce better medical doctors?
3. Are physically fit people more likely to have fewer colds?
4. Do students who study 2 hours or more for an exam get better grades?

It may seem reasonable to assume these statements are true. But we cannot say for sure until we look at all four combinations. Representativeness Heuristics Insensitivity to Sample Size This is the classic -- Four out of five dentists recommend Colgate. If only five dentists were asked this is not representative of enough dentists to make that claim. Yet many people will believe it.      The point of this heuristic is that any small sampling is likely to be terribly skewed. When I tell you that my surgeon is the best, she gives me excellent advice and you should see her. My sample size is ONE. Most people I know swear by their medical doctors. They may even argue with you over a recommended course of medical care if it differs from their physician's advice because their physician is better. And yet the person disagreeing with you has only one sample – his/her own physician.      We cannot determine what is true if we do not know how many "samples" have been included in the final outcome. The standard for this heuristic is --" the more items in the sample -- the better the result". Misperceptions of Chance I will illustrate this one with a series of statements.

• After pulling the one armed bandit many times -- it is due to hit!
• I have had three bad secretaries in a row. I am due a good one!
• The star basketball player has missed three shots. He is due to make the next one!
• The five stocks I purchased have gone down. I am due to buy a winner!

     This heuristic is terribly common and terribly wrong. It deals with the probability of any one event happening. One must realize that any one event will happen true to its probability and it does not matter how many times one does it ---- the probability of it happening stays the same.      The probability of getting a heads or tails on a coin flip is 50-50. No matter how many times I flip a coin the probability stays the same. It does not matter if I flip the coin five times in a row and it comes up tails each time. The sixth flip still has a 50-50 probability of coming up heads or tails. Each individual flip is a 50-50 probability. Regression to the Mean      It is best to read the example of this heuristic in Bazerman to really get the meaning. The airplane landing example is excellent and best illustrates the heuristic. However, I will give you two sample examples. These are examples and not necessarily accurate representations of a real team performance or personal performance.

• The Seattle Seahawks average about two touchdowns per game. They play in a very important game and score five touchdowns. Do we expect them to score six touchdowns in their next game?   Most sports fans understand that the likelihood of scoring six in the game after scoring five touchdowns is "pie in the sky". Fans may say “Gee wouldn't it be nice if the team scored six TDs in the next game" --- but they also realize that the probability of doing this is extremely minuscule.
• You tend to get slightly above average performance evaluations each year at your place of work. This year your boss assesses you at "outstanding". She then says to you  -- " This was a great year. I expect you to do even better next year." Is this a reasonable expectation? If you understand the regression to the mean heuristic you realize that her expectation is "pie in the sky" too.

The point of this heuristic is to gain an understanding of just what it means to have an average for performance. We each have some average performance for what we do that is derived from all of our performances. It is unlikely that every performance will be exactly the same. Some will be a bit better than the average and some will be a bit worse than the average --- and some will be on point -- average. Our average may go up slightly if we begin to perform a bit more frequently better than average. Or it may go down. But basically our performance will revolve around our average. Therefore, when we do well on one performance it is NOT REASONABLE  to assume the following performance will be even better. The following performance may be better -- but it is not likely. And sooner or later we will perform poorly -- in keeping with our average. Anchoring This heuristic illustrates how giving a person some idea of what an answer might be to a question will have a definite influence on what the person will provide as an actual answer to that question. For example:

• In a study of real estate appraisers -- they were asked if they could provide a completely objective assessment of the worth of a piece of property. They all said -- of course -- they could do that -- no problem. They were divided into two groups. One group was told the previous assessment was $140,000. The other group was told the previous assessment was $160,000. Each group was then turned loose to assess the property. Without  exception the collective assessment of the group who had $140,000 as the anchor, was lower than the assessment of the group given $160,000 as the previous assessment. They all thought they were being completely objective -- however, it is clear that the "anchor" had an influence on the outcome.
• In the educational system  where teachers are given the expected performance levels of groups of children -- amazingly those groups do tend to perform at that level. This is true even when a teacher of a low performing group is told the group is a high performance group. In other words, when the teacher thinks the students are high performing they do perform better than the expected performance level would be for the group they are actually in. (This should be a bit scary for you to contemplate)
• Certainly some of you have been asked what salary you expect when applying for a job. Whatever you say becomes an anchor for the employer. If you give too low a figure you may end up getting less than they were willing to pay. On the other hand --- if you give too high a figure, they may decide they cannot meet your expectations and not offer you the job. Very tricky. You need to give a salary figure (anchor) that stretches the employer's estimate of the worth of the job, but not so much that the employer feels meeting your expectation is impossible. If you can do this you may get a salary offer a bit higher than the employer was wanting to pay --- but willing to go the extra few bucks to get you.
• In a study where the answers to the question were completely irrelevant, people were asked to guess the number of countries in Africa. Before they gave their answer they had to spin a roulette wheel. The sampling of the results showed that those who had an initial spin of 10 versus those who had an initial spin of 65 were definitely influenced by the spin number. The 10 spin group tended to guess --  as the median point, 25, and the 65 group tended to guess as the median point, 45. They all knew that the spin number had absolutely nothing to do with the actual number of countries in Africa ---- and yet the spin number (the anchor) clearly had an impact on the number of countries the subjects guessed.

Overconfidence Bias People tend to be overconfident in their judgments. Individuals are imperfect at gauging their own level of accuracy when asked to estimate their ability to make accurate estimates of situation outcomes. A few tendencies can be identified. as follows:

• People tend to fail to account for the difficulty of questions -- in fact, we tend to be more overconfident about our answers when we respond to questions of moderate to extreme difficulty than to easy questions.
• We tend to overestimate our performance when it is at its poorest.
• Interestingly, people tend to demonstrate no overconfidence (to some degree some under confidence) when asked questions with which they have considerable familiarity.
• Groups are much better at reducing this tendency to overestimate ability when issues are uncertain and the degree of knowledge about the subject is minimal.
• This is true in groups where there is there is a lot of diversity and individuals have a wide degree of  judgments and considerable disagreement.

This tendency to be overconfident when we really know little about the matter can be reduced a bit if people are given feedback regarding their overconfidence. The concern is that we may make terrible decisions in situations where we know little about the issue. Consider situations where we disagree with a decision the leadership has made. We criticize it and may even believe the decision was totally self serving. But the facts upon which the decision was based may not be available to us. The leadership was aware of them and this information was critical in making the decision. We have an opinion regarding what the decision should have been  --- to a great degree because we do not have all of the information. If we did have all of the information (sometimes this information cannot be shared for confidential reasons) we would make the same decision as the leadership. Or, consider what happens when your physician tells you that you need an operation. It is fairly risky. You agree you need it and you ask the surgeon what success rate she has with this particular operation. Since we tend to imperfectly estimate our own performance, and we also tend to over inflate what we think of our own ability --- the surgeon may tell you she has a great success rate but in fact it is average. If we are willing to accept that this bias exists we may be better able to sift through issues and make rational decisions. NO matter --- beware your own willingness to offer opinions on issue about which you have little knowledge. The Confirmation Trap This heuristic serves to point out our tendency to notice any and all information that supports what we already believe or what we want to believe, and to ignore as irrelevant, any information that conflicts with what we want to believe. We tend to be emotionally attached to our beliefs. We make judgments -- many times resulting from the previously illustrated heuristics -- and we are immediately wedded to these judgment, beliefs. The attachment  to these beliefs seems to become a living part of our very being. We become the belief and the belief is us. Any input to the contrary threatens not just the belief, but our very existence. (Perhaps I am overstating this, but there is an element of truth to it.) Once we have adopted a position, we are very reluctant to "see" input to the contrary. It becomes very easy to simply dismiss any dissonant information as insignificant or just plain wrong. It is this heuristic that makes it very easy for groupthink to develop. "My way is the only way and it should be yours too." "If you disagree with me you do not belong (because you are stupid, obstructionist, difficult, etc.)." " If you agree with me you are intelligent".      Too many leaders become so full of themselves that they seriously threaten anyone who makes statements that contradict. But the really bad thing about this heuristic is -- suppose you want to hire a new person. You find one you like. He/she may be the worst person for the job ---- but if you fall victim to this heuristic you may only  listen to the feedback that says the person is great. You hire the person and it is a disaster. Or you want to change the direction of the company. You let people know about this and some of them say it will be a big mistake. You dismiss them as disgruntled limited vision employees. You only listen to those who say it is a great idea. ----- You get the point.!!! The wise person deliberately finds information that is contrary to the suggested course of action. This is the only way to really attempt a reasonable decision.  Hindsight Bias 

Alex Agase, former Northwestern University football coach, once quipped if you want to give him advice, do it on Saturday between 1 and 4 o'clock, during the 25 seconds between plays. Not on Monday. He knew the right thing to do on Monday.

Despite Agase's lament, Monday morning quarterbacking is still a favorite American pastime. Clearly the interception could have been avoided by running the ball. Since the outcome is so clear, Monday morning quarterbacks question why the coach couldn't anticipate it. We tend to expect that others should know by foresight what we have learned by hindsight. The problem is that this bias is not confined to football. It is quite pervasive and has the potential to adversely impact a wide variety of human performance assessments.

Being Smart After the Fact

Proclamations about human error are most always made "after the fact," rarely before. As noted by Reason (1990), the most significant psychological difference between individuals who were involved in events leading up to a disaster and those who are called upon to investigate after it occurred is knowledge of the outcome. Investigators have the luxury in hindsight of knowing how things are going to turn out; front-line operators and their supervisors do not. While most people would not expect much credit for picking a horse after it has won the race, many investigators are unaware of the influence of outcome knowledge on their perceptions and re-constructions of the incident. Given the advantage of a known outcome, what would have been a bewildering array of non-convergent events becomes assimilated into a coherent, causal framework for making sense out of what happened. In fact, it may be difficult to imagine it happening any other way. "Why couldn't they see it?" is the question that is often asked. Such hindsight results in expectations by investigators that participants should have anticipated the incident by foresight; it also blinds them to what actually would have been known had the roles been reversed. If investigations of human error are to be fair and impartial, appropriate actions and decisions need to be determined before the mishap; not from the comfortable vantage point of hindsight.

Reducing Hindsight Bias

Unfortunately, reducing hindsight bias is not that easy to accomplish. It is not as simple as issuing a friendly warning to those people who are likely to commit the offense. Although they may understand the rationale of how reconstructive memory works, the next time they serve in an evaluative capacity where the outcome is known, the same bias of allowing the outcome to carry too much weight in judging the process is likely to occur. One possibility comes from research by Slovic and Fischhoff (1977) who found that hindsight bias could be reduced when people are asked to consider how other possible outcomes might have occurred. By considering how past events may have turned out differently, we become less anchored to the outcome which otherwise dominates our reconstruction of the episode and are more open to other possibilities. Rather than looking back down the path after it has already been chartered with the outcome known, we instead are placing ourselves in the path to better appreciate the uncertainty, confusion, competing demands, and options that were open to the participants and could have led to a different outcome. When forced to consider alternative outcomes, self-serving statements that "I knew it all along" might start to diminish.

A related technique for those who suffer under all-knowing investigators or managers who are habitually smart after-the-fact is to withhold announcement of the outcome until they have given their own estimates or predictions regarding impending outcomes. Get them to write down their own views, what assumptions they are making, what trade-offs and contingencies they are weighing, what options they are considering ? once again, putting them in the decision path ? before the outcome announcement is made. This forces investigators to attend to the process, which is messy and riddled with unknowns, rather than having the luxury of a clean and known outcome simply handed over. Otherwise, the outcome appears inevitable and doubts may be cast in the direction of the front-line operators as to why they could not detect the obvious.

References

Fischhoff, B. and Beyth, R. (1975). "I knew it would happen": Remembered probabilities of once-future things. Organizational Behavior and Human      Performance, 13, 1-16.

Reason, J. (1990). Human error. Cambridge: Cambridge University Press.  

Curse of Knowledge The classic example of this heuristic involves giving someone directions to some place. You know exactly how to get there and you believe your directions are perfectly clear. The person does not get there and has to call you for clarification. You are befuddled as to how she/he could have gotten so confused. After all --- your directions were right on.      As related to this course --- the art and science of communication are at the very core of effective leadership. One must be able to communicate clearly, verbally and in writing. The above example illustrates the verbal. An example of written "curse of knowledge" is evident when one writes a paper. Far too often the writer knows exactly what is meant -- but the reader is confused. This is a mystery to the writer because the message seems to be crystal clear. I suggest, before submitting any written item, from E-mail to formal research paper, it would be wise to do one or all of the following:

1. Spell check it
2. Reread it VERY carefully
3. Depending upon the complexity of the document, have someone else read it before it goes to the intended recipient
4. Depending upon the complexity of the document wait a day and re-read it
5. Read it out loud
6. Read it out loud to someone else.

Comments All of these heuristics help us make decisions --- both good and bad. Heuristics are shortcuts that seem so "right". There is no easy fix to avoiding them.

• First we must admit that they exist.
• Then we must accept that we are each vulnerable to them.
• They compromise our ability to make rational logical good decisions.
• And yet, at times they are valuable.
• They save time and hence money.
• We must be constantly vigilant in order to use them where appropriate and to discard them when they will lead to errors.
• This is not easy.
• There is no such thing as a free lunch!