4 Biases in Decision Making
Week 4:
TL;DR: give helpful feedback to friends
readings (links) & lectures ～ assignments due ～ live session agenda
TL;DR: give helpful feedback to friends
readings (links) & lectures ～ assignments due ～ live session agenda
This week is heavily influenced by Kahneman’s Thinking Fast and Slow. We recommend starting with Kahneman, then watching the lectures, then looking at the rest of the readings, which better consider the consequences for business of biased decision making. While this is another reading heavy week, student have found this to be one of their favorite units. While you study, you’ll also have an opportunity to practice your collaboration skills by using GitHub.com to give and receive feedback to a small circle of your peers.
Figure 12: Week 4 “reading” time estimated at 250 words per minute
Readings
- ↓ Davenport (2009)
- ↓ Hammond, Keeney, and Raiffa (1998)
- ↓ Kahneman (2013b)
- ↓ Kahneman (2013a)
- ↓ Stauffer (2002)
Lectures
- ↓ 4.1 Introduction to Biases
- ↓ 4.3.0.1 Biases Can Be Fun
- ↓ 4.3.1 Deadly Serious Stuff I
- ↓ 4.3.2 Deadly Serious Stuff II
- ↓ 4.4 Common Biases Overview
- ↓ 4.4.1.1 Biases Example
- ↓ 4.5 Makes Choices Based on Interests
- ↓ 4.6 Bias Interview
- ↓ 4.7 Difference in Kind vs. Difference in Degree
- ↓ 4.8 Week 4 Wrap-Up
4.1 Introduction to Biases ↑
| Players | Slug | Excerpt |
|---|---|---|
| ι δ | including ourselves | these biases are present for everybody we know, including ourselves |
| δ ο | cognitively challenged | human beings are just really bad at processing information. We are cognitively challenged |
| υ χ | consistent mistakes | we make lots and lots of mistakes, and we make them consistently |
4.3.0.1 Biases Can Be Fun ↑
| Players | Slug | Excerpt |
|---|---|---|
| β ε | marshmallow deal | WOMAN: Do you want to have a seat? Let’s get comfortable. The deal is you can eat marshmallow now or you can wait until I come back and I’ll give you two marshmallows. If you decide to change your mind that’s fine okay. LITTLE GIRL: I’m waiting. |
| σ λ | mom or dad | Do you think your mom or dad would eat the first marshmallow or would they wait, like you did? LITTLE BOY: Well my dad would eat the first marshmallow and my mom would patiently wait. |
4.3.0.2 ↑
| Players | Slug | Excerpt |
|---|---|---|
| θ κ | visual illusion | I will tell you a little bit about irrational behavior, and I want to start by giving you some examples of visual illusion as a metaphor for rationality. |
| ι φ | haven’t learned anything | to the extent you believe I didn’t shrink the lines, which I didn’t, I’ve proven to you that your eyes were deceiving you. Now, the interesting thing about this is when I take the lines away, it’s as if you haven’t learned anything in the last minute. (Laughter) “You can’t look at this and say,”Now I see reality as it is." |
| ο φ | consistent foolishness | Our intuition is really fooling us in a repeatable, predictable, consistent way. and there is almost nothing we can do about it, aside from taking a ruler and starting to measure it. |
| λ υ | financial decision-making | what are the chances we won’t make even more mistakes in something we’re not as good at, for example, financial decision-making. (Laughter) Something we don’t have an evolutionary reason to do, we don’t have a specialized part of the brain for, and we don’t do that many hours of the day. |
| θ ρ | cognitive illusion | in visual illusions, we can easily demonstrate the mistakes; in cognitive illusion it’s much, much harder to demonstrate the mistakes to people. |
| λ η | organ donation | Why do some countries give a lot and some countries give a little? When you ask people this question, they usually think that it has to be about culture. How much do you care about people? |
| μ β | begging | You know the expression, “Begging only gets you so far.” It’s 28 percent in organ donation. (Laughter) |
| λ γ | opt-in | “Check the box below if you want to participate in the organ donor program.” And what happens? People don’t check, and they don’t join. |
| β ο | opt-out | “Check the box below if you don’t want to participate …” Interestingly enough, when people get this, they again don’t check, but now they join. |
| ξ ο | decision residence | many of these decisions are not residing within us. They are residing in the person who is designing that form. When you walk into the DMV, the person who designed the form will have a huge influence on what you’ll end up doing. |
| ρ υ | because we care | It’s not because we don’t care. It’s the opposite. It’s because we care. It’s difficult and it’s complex. And it’s so complex that we don’t know what to do. And because we have no idea what to do, we just pick whatever it was that was chosen for us. |
| ω θ | hip replacement | no physician would ever say, “Piroxicam, ibuprofen, hip replacement. Let’s go for hip replacement.” But the moment you set this as the default, it has a huge power over whatever people end up doing. |
| ζ ξ | both for 125 | This was an ad in The Economist a few years ago that gave us three choices: an online subscription for 59 dollars, a print subscription for 125 dollars, or you could get both for 125. (Laughter) |
| ρ δ | preferences | we actually don’t know our preferences that well. And because we don’t know our preferences that well, we’re susceptible to all of these influences from the external forces: the defaults, the particular options that are presented to us, and so on. |
| φ ξ | ugly Jerry | And the question was, will ugly Jerry and ugly Tom help their respective, more attractive brothers? The answer was absolutely yes. When ugly Jerry was around, Jerry was popular. When ugly Tom was around, Tom was popular. |
| ω ι | behavioral economics | when we think about economics, we have this beautiful view of human nature. “What a piece of work is a man! How noble in reason!” We have this view of ourselves, of others. “The behavioral economics perspective is slightly less”generous" to people |
| α μ | Superman or Homer Simpson | The silver lining is, I think, kind of the reason that behavioral economics is interesting and exciting. Are we Superman, or are we Homer Simpson? |
| ρ λ | mental world | for some reason, when it comes to the mental world, when we design things like healthcare and retirement and stock markets, we somehow forget the idea that we are limited. |
| α ρ | better world | if we understood our cognitive limitations in the same way we understand our physical limitations, even though they don’t stare us in the face the same way, we could design a better world |
4.3.0.3 ↑
| Players | Slug | Excerpt |
|---|---|---|
| α φ | we want overconfidence | we want overconfidence we support it we sustain it |
| γ ο | loss aversion | people put a lot more weight on negative events than on positive events on losses than on gains |
| υ α | opposite directions | overconfidence and loss aversion seem to be acting in opposite directions |
| ι κ | bad for you good for us | for society it is probably very good that we have a lot of optimistic entrepreneurs and you know who think they will succeed although most of them fail |
| χ σ | optimism | we know that being an optimist is useful under some conditions it is not always useful in making decisions |
4.3.1 Deadly Serious Stuff I ↑
| Players | Slug | Excerpt |
|---|---|---|
| ε ζ | test drive | why do people buy a car after test driving only two or at most three cars? That’s a really big investment, there’s a lot of time, energy, emotion, and money wrapped up in that |
| γ φ | story time | if you want to make that behavior seem rational, you tell a story where the costs of test driving another three cars exceed the expected benefits of getting a better car. But that’s just a story we tell ourselves. |
| ζ λ | just so stories | these kinds of rationality debates are referred to as kind of “just so stories.” The world works as if people are rational, and they can be very misleading. |
| τ σ | change behavior | So how are we going to make people drive more than two cars? |
| ω λ | data beats bias | the interesting question I think is not rationality or irrationality. It’s really the effort to find out what people actually do, to try to identify the repeated common biases in decision making and just work with them for what they are and then find ways to use data to address them for the good of the decision maker without worrying so much about whether it’s rational or irrational. |
| δ φ | deadly serious | imagine you have two identical cancer patients who walk onto your oncology ward. And they’re going to have to make a decision about whether they are going to agree to a new experimental treatment |
| ζ υ | right choice? | so on the one hand, 90% chance of surviving and on the other hand, 10% of dying. One of these people is going to say yes to the treatment, and one of them is going to say no. So who’s made the right choice? |
| φ ψ | right data | the right data presented in the right way at exactly the right moment in a decision process just might make a big difference in a decision like that and actually might save people’s lives |
4.3.2 Deadly Serious Stuff II ↑
| Players | Slug | Excerpt |
|---|---|---|
| δ ο | internal yes man | Instead of doing what we should do, which is seeking out evidence that would falsify a hypothesis that we hold, we tend to seek out and pay greater attention to the evidence that supports our hypothesis, that reinforces our assumptions, that make stronger our prior beliefs. It’s like the internal yes man that just wants to believe. |
| γ φ | beautiful apple | I think Apple is a great company. And I think its stock is way underpriced. So when I walk into an Apple store, what am I going to see? I’m going to see all the beautiful devices, and I’m going to see all the people playing with those beautiful devices. And I’m going to completely ignore the fact that nobody actually seems to be buying anything. |
| λ μ | I’m not lying | It’s not intentional. The biased person is not consciously trying to mislead himself or mislead others. If you’re doing that, it’s called lying. It’s not confirmation bias. |
| ρ δ | cold and hot | Cold elements are the cognitive information processing mistakes that people make. Hot elements are the emotional motivated mistakes. |
| ο η | people like me | this stuff can be strongly embedded in the self-image of the decision-maker. So I think of myself as a friendly person, who other people tend to like. Or I wish I was. |
| γ δ | Porsche or BMW | the decision is really close. They’re both great cars. They’re different in some ways. I go back and forth, I can’t decide, I can’t decide, I can’t decide. And then I decided to buy the Z4. OK, now ask me a week later, and here’s what I’ll tell you. It was a no-brainer. The Z4 is much better on almost every dimension |
| τ σ | split and bolster | I split the alternatives and make them seem more different than they really are. And then I bolster my own choice, making it seem much more dominant than it really was. |
| γ ρ | stress reduction | why do people do this? Some psychologists think it’s just an effort to reduce cognitive dissonance. You don’t want to go back after you make the decision and think about the dimensions that would have favored the other decision. You don’t want to look back. So it reduces stress and reduces the discomfort that would come with that. |
| ο λ | ambivalent leadership | it’s an effort to reduce the appearance of ambivalence. And a lot of people think ambivalence can be really detrimental to leadership. |
| υ μ | pro-con list | Imagine how the American public would have reacted if George W. Bush had stood up on TV and given on the one hand a list of reasons to invade Iraq, and on the other hand, a list of reasons not to invade Iraq, and then told the country that on balance, he thought it was a better decision to be made to invade, but just barely, and he still had a lot of doubts. |
4.4 Common Biases Overview ↑
| Players | Slug | Excerpt |
|---|---|---|
| η ψ | self-serving bias | SSB is a special case of the fundamental attribution error in which individuals who are confident of themselves tend to attribute their successes to personal skill and their failures to bad luck or someone else’s mistake. Low-confidence individuals often make the opposite attribution. |
| ξ μ | ssb hot cold | SSB has cold and hot components: Hot (emotional): enhancing self-esteem Cold (cognitive): enhancing the individual’s sense of how much control he or she has |
| τ δ | woman man old young | SSB is slightly more prevalent in men than in woman and slightly less prevalent in older than younger people. |
| ω γ | selection bias | in most corporate decision-making environments where selection bias may easily overwhelm these differences |
| γ ρ | ambiguity effect | there is a common bias to prefer options that have more precise probabilities over those for which the probability of a good outcome remains unknown |
| π η | bird in the hand | “A bird in the hand is worth two in the bush.” In formal terms, this may not actually be true, particularly if the probability of catching a bird in the bush exceeds 50 percent. |
| υ γ | provide probability | A rational decision maker would be willing to invest a certain amount of effort in searching for information or models that can provide a probability estimate to the ambiguous option, so that it could then be weighed on a level playing field against the more well understood option. |
| φ τ | endowment effect | your willingness to pay for something you don’t currently have should precisely match the price you’d be willing to accept to sell the same thing if you currently had it. |
| τ α | loss aversion | loss aversion (the phenomenon where people experience the negative impact of a loss as greater than the positive impact of an equivalent gain). |
| ω θ | value potential | Some potentially value-creating deals will not get done if the endowment effect is big enough to matter. |
| μ λ | DS vs EF | Can data scientists intervene to help decision makers become more fully aware of this bias, its impact on markets, and potential ways to compensate for it? |
| ρ ψ | hindsight bias | the mind’s inclination to interpret events that happened in the past as more determined and/or predictable than the same mind would have interpreted them before they happened |
| θ ε | learning mistakes | How does a decision maker learn from others’ mistakes and failures (usually the most efficient way to learn) if they’re certain they would never have made the same mistake? |
| υ κ | managing contingency | Managing hindsight bias, at the simplest level, would be helped by maintaining in the decision makers’ mind an appropriate understanding of the contingency of outcomes, even after that contingency has been removed by recorded history. |
| τ χ | mystery of history | It’s a problem that mystery novel and movie script writers encounter just as much as experimental researchers. |
4.4.1.1 Biases Example ↑
| Players | Slug | Excerpt |
|---|---|---|
| ρ θ | fundamental attribution | very simply, it’s the notion of attributing something to the kind of person you are rather than to the situation that you’re in |
| ε ψ | rich or poor | they’re greedy or they’re lazy. And it does affect, say, policies that we would make around that socioeconomic status, either taxes or different unemployment benefits |
| τ α | what to collect | do you collect data about the person or do you collect data about the labor market? |
| ρ ε | where to look | if you come in with that bias, you don’t know that you should be looking somewhere else for data. |
| μ σ | last game effect | it doesn’t really matter how we’ve been doing for the past several games, but the last game really affects us. |
| ι ν | Eli Manning | Being a long-suffering New York Giants fan, because I love Eli Manning, if he has a bad game, it can’t be because of anything wrong with him. It has to be because of the situation he was in, maybe the offensive line or something else. |
4.4.1.2 ↑
| Players | Slug | Excerpt |
|---|---|---|
| β α | recent attribution | Can you think of a recent case where you’ve run into an attribution error like this that has actually made an important difference in a decision-making setting that you’ve been involved in? |
4.4.1.3 ↑
| Players | Slug | Excerpt |
|---|---|---|
| θ χ | program evaluation | what have been the outcomes of, say, the money that you’ve invested? And if it’s been going poorly, you think, I’m going to look for all these negative things versus maybe it had nothing to do with the program itself, and it had everything to do with the economic climate or some other cultural reason. |
| α τ | inside my head | I wouldn’t speak up very much and my coworkers would assume that I just I wasn’t involved or I wasn’t participating because I would speak up when I had something to say and not as I was thinking things through. |
| π ω | high thinking | Part of what makes us high thinking functioning human beings is we assume a lot of things around us all the time. And most of the time, that serves us really well. |
| κ φ | dangerous thinking | I think that’s the most dangerous part is when you don’t know that you have this bias, and you are potentially making very important decisions for a company or even in your own personal life |
| μ ε | DS vs FAE | just imagine for a moment that we were going to use some of the new data tools, or transparency tools, in a determined effort to just try to reduce the frequency with which people fall prey to this attribution error. Can you imagine, what would we do? |
| λ ι | unexpected data | if you were collecting data about a certain group of people or a product, take data that people wouldn’t actually think to look at. And then, explain why it might be relevant |
| ν ο | brand reputation | It’s actually really, really hard to change a brand reputation. And no amount of data appears to be able to get over that attribution bias. |
| π ω | anchoring bias | the first piece of data that’s presented to you in a new situation– a decision situation, a negotiation– will have an inordinate impact on the rest of the conversation or the rest of the negotiation. It anchors the discussion to that data point. |
| ζ φ | salary negotiation | the iconic experience is when someone says to you, here’s the starting point to negotiate your salary, they’ve anchored you on that point. |
| φ γ | news cycle | in the 24-hour news cycle, when a story first hits, the first story that you read might shape your opinions as the entire event unfolds even if something very different ends up happening |
| λ ζ | fake price | JC Penney used to have the fake prices where you’d have a really high price and then, they would have sales and coupons and that would pay your actual price. And then a little while ago, they decided we’re not going to do that. We’re just going to show this is the price you’re going to pay. |
4.4.1.4 ↑
| Players | Slug | Excerpt |
|---|---|---|
| φ ζ | stick shock | What strategy is being used with anchoring, in this case, to get me to do what they want me to do? |
4.4.1.5 ↑
| Players | Slug | Excerpt |
|---|---|---|
| π ρ | inappropriate price | I always give what I think is the appropriate answer, as opposed to– because you don’t want to be the idiot that walks into a car dealership and asks to pay $2,000 for some car. But really, that’s negotiating with myself which is, pretty bad thing to do. |
| ω ξ | pay nothing | it doesn’t function as a useful anchor, actually. It would actually– it turns the initiative for anchoring over to the other side of the negotiating table which is something you probably never want to do. So you want to take advantage of the opportunity to anchor it to something that you like. |
| π φ | DS vs AB | let’s imagine that we have every data tool that we want to bring to that table to make people’s decision making a little more rational and get it off the anchor. Any ideas about how we would do that? |
| ξ π | exploration | I just like to play with the data. Move things around. Take a look at different sort orders. Are there any trends that happen that I can look at that would push me away from what I’ve assumed I can do with it. So just exploring with it |
| ν π | recency bias | in a progression of experience or a train of data that’s coming down the track at you over time, people tend to pay way too much attention from– from a rational choice perspective– to the piece of data that’s most recent |
| ψ α | peak-end phenomenon | there’s been a great deal of research about this with regard to what’s called the peak end phenomenon |
| σ ο | forgetful voters | people vote on the basis of the last couple of months rather than say the four years, or two years, or six years of the experience |
| ρ β | recession repression | 2008 seems like a long time ago now. Everything seems to be going well. People seem to be getting more jobs. The economy seems to be doing much better. So we must be out of the thick of it. |
| ρ ξ | mind grabbing | the other side of the recency bias is our tendency to overly discount things that happened eight or nine years ago as if that were some ancient history that’s no longer relevant to us. And actually, we have all the data. It just doesn’t grab people’s minds in any meaningful way. |
| ζ λ | all data are equal | all data are equal. Some data are more equal than others |
4.4.1.6 ↑
| Players | Slug | Excerpt |
|---|---|---|
| ξ ε | DS vs RB | If you could design a data presentation or visualization that aimed to counter, or “even out,” recency bias in a particular setting, how would you do it? |
4.4.1.7 ↑
| Players | Slug | Excerpt |
|---|---|---|
| ν ρ | 100 years | But say, if you take it for over 100 years or over the last couple of years and show the differences, it helps people put it in context more because you don’t usually look at a graph and add your own context to it immediately. |
| θ π | little goodies | when you leave the dentist’s office, you always get a bag full of little goodies, like a free toothbrush– free toothpaste. |
4.5 Makes Choices Based on Interests ↑
| Players | Slug | Excerpt |
|---|---|---|
| μ ε | lizard brain | negativity bias probably helped lizards survive in the prehistoric era, but it’s stuck in our brains now. We don’t really need it. All we need to do is try to get better at making the decisions that are actually consistent with our interests. |
| φ α | real people | if you’re dealing with technically and scientifically minded people, we’re going to naturally sort of think, OK, rationality is good. Irrationality is bad. And that’s actually not going to help us when we’re dealing with real people. |
| σ η | different mistakes | Esther Dyson, who you probably know– she was the founder of Release 2.0. She’s actually the daughter of the physicist Freeman Dyson. She once said something really profound, which was simply, “Make different mistakes.” |
| υ μ | same damn mistakes | The biases tend to drive people and organizations to make the same damn mistakes over and over again, and that’s what we really want to try to counter. |
4.6 Bias Interview ↑
| Players | Slug | Excerpt |
|---|---|---|
| μ ω | Jeff Zych | from Optimizely who is working on the web application of A/B testing and biases in user testing. |
| γ η | design testing | I make things that designers design. And part of that is doing an actual A/B testing because everything that we change we want to test to make sure it actually has some sort of positive affect. If it doesn’t, then we’re probably going in the wrong direction and to change something. |
| γ δ | MIMS | So prior to Optimizely, I was at the School of Information, getting my master’s in Information Management and Systems. MIMS, as they call it. |
| β κ | olden days | in the olden days, you make a change to a website and you would just do it, and you won’t really have any data or insight into if it made a difference. |
| ο ι | A/B testing | what A/B testing lets you do is you have an original version of your page, and then you make a second version of it that has different texts, or different buttons, or a different layout, or something. And then basically you measure those two against each other. |
| ζ α | key metrics | usually you have certain metrics that you care about, such as clicking a button, or buying some sort of product. And you can see people that look at version A, are they more likely to buy this product, or are people in version B? And that gives you really powerful data to see if your changes are actually having a positive impact, either on your business or whatever your key metrics are. |
| δ ο | bad way | a lot of companies and people have no idea what they’re doing, and they all come to a product like Optimizely and they start randomly changing things. That is really a bad way to go. |
| ν χ | aesthetic rules | The aesthetics of one person might be really different from the aesthetics of another person. There’s really no systematic rules around that, are there? |
| λ φ | behavioral economics | behavioral economics is a study of why people make decisions, and especially emotional decisions or decisions that seem irrational. |
| γ ο | split, bolstered | this can lead you to theories of testing. So you can say, if I changed whatever the default option is, then people are more likely to go with that. And that can lead to some really powerful data and insights that you wouldn’t get just from changing things randomly or changing the visual side of things. |
| θ ρ | overwhelming toothpaste | in behavioral economics they have this idea of choice sets. And the basic summary of it is that people will get overwhelmed with too many choices, and they all kind of get paralyzed with choices, essentially. It’s like when you go to the grocery store and you see an aisle full of toothpaste and you’re like, I don’t know which one to choose. |
| η χ | constructive bias | You’re essentially trying to be constructively biased in a way, in the sense of helping people to make the decisions that they want to make, but in a way that benefits your client |
| ε ν | population segmentation | a big shortcoming with A/B testing these days is that there’s not enough way of targeting different populations. And there are rudimentary ways of doing it. But the biggest thing we see is people will run A/B tests and then after they get the data, they will start segmenting it. After the fact, essentially. |
| φ ψ | personal web | it’s kind of what we want to move towards. Is having people get more personalized experiences on the web, and be able run A/B tests that specifically target different types of audiences. |
| υ θ | exact same experience | when you go to amazon.com, if you ever shopped there before or even really looked at the site before, they all start having recommendations on their home page of things to buy that are similar to what you looked at or what you bought before. As opposed to just giving everyone the exact same experience, saying like, here’s a Kindle. |
| τ ο | increase buying | they have tested what to show different people at different times and how to increase people buying products from them, essentially. |
| θ κ | visitor point | it’s something that is typically it’s fairly easy to set up. But you get lots of data because lots of people are using your website, well hopefully. But basically every visitor is a data point that you can view. |
| ρ ι | qualitative counterpoint | the counterpoint to that is something like qualitative research where you’re actually going out, and talking to users, and seeing what they do on your website. And that’s the classic user research. They’re different types of data, but this is a very cheap and efficient way of getting lots of data around very specific questions. |
| π ξ | clutter | what really matters is really just focusing on your website, on your message, you your call to action. Those types of things. Because the biggest thing that we see is people have these really cluttered pages. People get distracted really easily. |
| ζ β | holy grail | one of the jokes we’ve made is that we’ve talked about having an automated test suggestor. That’s kind of some holy grail. |
| ω ξ | find the core | we just remove elements randomly from a page, and that’s probably better than any other tests they could run. Because that will help people hone in on a page that is focused down to the core on what you want it to do. |
| β χ | culturally ingrained | we do have designers, of course, doing a lot of visual design. And they are very data-driven. As a company, it’s kind of ingrained in our culture. And whenever they’re designing something, they’re really good about thinking, how can we test this? |
| ψ κ | frustrated designers | there has to be a balance there between a designer’s intuition, and their gut, and what they think is good, and what’s good for the business. Because what happens with companies like Google who have a tendency to over test things is designers can get frustrated with having to prove that what they’re doing is right |
| β ε | pure look | Some other things purely just look better and they don’t necessarily convert better |
| π υ | 41 shades of blue | I’m the designer. I shall choose the blue that I think is best, and testing 41 shades of it just took away all his agency and his creative freedom. |
| β α | theoretical average user | what you’re doing is you’re kind of making an average best site. So your theoretical average user is having the best experience. But that doesn’t make the best experience for every user that’s coming to your site. And so I think it’s getting more personalized. I think testing is going to lead towards that. |
| ψ υ | cross-device testing | interaction isn’t captured in A/B testing today because you can’t track across devices. But it’s something that’s really valuable data that companies would love to have. But the tools of today are a bit too rudimentary to do that type of testing. |
4.7 Difference in Kind vs. Difference in Degree ↑
| Players | Slug | Excerpt |
|---|---|---|
| ζ κ | difference in degree | So we go out and gather more data, and we identify that certain behaviors and attitudes are related and they vary on a continuum. |
| φ υ | difference in kind | we go out and gather more and more data until we gather so much that people’s behaviors and attitudes appear less related, and we can no longer really identify what causes the bias. |
| ρ ε | fan mail | Imagine you’re a professional baseball team, and you send emails to your fans and customers, those who have purchased tickets from you in the past. Those emails advertise tickets to upcoming games. You find, of course, that some people buy tickets and many don’t. You want to look at those who don’t. Is bias playing a part? |
| θ υ | different combinations | You just want to try and identify what are they biased for or against. You try all these different sorts of combinations to see if you can identify a difference in degree for their bias. Maybe, for example, they’re biased against late-season weekend baseball games. |
| ν δ | more data less knowing | As you gather more data, it seems there’s seemingly no reason for you to understand. They just don’t buy tickets. Maybe they just don’t like baseball. We don’t really know. And as I mentioned, perhaps it could be because they’re football fans, but you don’t know that from the data that you get. In this instance, what you have is a difference of kind. |
| ε ι | not in the data | we don’t really know how or why someone is biased. It’s really complex, and, of course, like many things, it changes over time. There’s more to it than what’s in the data. |
4.8 Week 4 Wrap-Up ↑
| Players | Slug | Excerpt |
|---|---|---|
| π δ | optimist or pessimist | a lot of different perspectives on just how good or bad human beings are at decision making. Baruch Fischhoff is an optimist. Quite frankly, I’m more on the pessimistic side. I think human beings are lousy at decision making |
| φ χ | nutshell bias | biases, the repeated, reproducible ways in which the human brain makes mistakes as it absorbs and interprets information. |
| φ π | suffering | people just process information in a biased way. That’s what we know for sure. Here’s what I think we don’t know. First, we really don’t know just how significant some of these biases are as a source of, say, human suffering and business under-performance. |
| ε ν | destroyed value | is confirmation bias, which we talked about a great deal, an important reason why firms commonly engage in M&A, Mergers and Acquisitions, activity that we know destroys value? |
| λ α | in principle | it seems as if some of these decision biases could be replicated in a meaningful sense and then subjected to enough data analysis that we would be able to demonstrate to people why they shouldn’t make those kinds of mistakes |
| ψ π | viz bias | new biases are going to emerge in the course of people learning how to use data, particularly in info visualization. There is a lot of anecdotal and actually some systematic evidence that people are very biased when it comes to the way in which they interpret pictures. |
| η σ | unconquered bias | No one yet, at least that I know, has ever fully conquered a decision bias simply by learning about it, you might see it more often, you might compensate for it in some instances, but you’ll never really overcome it just by knowing it’s there. |
Bibliography
Davenport, Th. 2009. “Make Better Decisions.” HARVARD BUSINESS REVIEW 87 (11): 117–+. http://search.ebscohost.com/login.aspx?direct=true&db=edswss&AN=000271040500023&site=eds-live.
Hammond, John S., Ralph L. Keeney, and Howard Raiffa. 1998. “The Hidden Traps in Decision Making.” Harvard Business Review 76 (5): 47–58. http://search.ebscohost.com/login.aspx?direct=true&db=bth&AN=1040942&site=eds-live.
Kahneman, Daniel. 2013b. “The Characters of the Story.” In Thinking, Fast and Slow, 1st edition, Chapter 1. New York: Farrar, Straus; Giroux. https://smile.amazon.com/Thinking-Fast-Slow-Daniel-Kahneman/dp/0374533555.
Kahneman, Daniel. 2013a. “Part II: Heuristics and Biases.” In Thinking, Fast and Slow, 1st edition, Chapters 10–18. New York: Farrar, Straus; Giroux. https://smile.amazon.com/Thinking-Fast-Slow-Daniel-Kahneman/dp/0374533555.
Stauffer, David. 2002. “How Good Data Leads to Bad Decisions.” Harvard Management Update 7 (12): 3. http://search.ebscohost.com/login.aspx?direct=true&db=f5h&AN=8587204&site=eds-live.