Chapter 10

Logic

Deduction, Induction, and Inference to the Best Explanation

I never blame myself when I’m not hitting. I just blame the bat, and if it keeps up, I change bats. After all, if I know it isn’t my fault that I’m not hitting, how can I get mad at myself?

—Yogi Berra

Alice tentatively approached the odd pair, whom she found standing silently and wax-like under a tree. They reminded her of an old nursery song that had been sung to her as a young child and she began humming it to herself in her mind. Then one of the curious pair broke the silence.

“I know what you’re thinking about,” said Tweedledum: “but it isn’t so, nohow.”

“Contrariwise,” continued Tweedledee, “if it was so, it might be; and if it were so, it would be; but as it isn’t, it ain’t. That’s logic.”

Logic is the study of sound reasoning, which is critical in forming beliefs that can be justified and used to make defensible decisions, whether you are in Alice’s Wonderland or the C-suite.

This chapter focuses on what is often called informal logic or critical thinking, which refers “to the study of reasoning and [logical] fallacies in the context of everyday life.”¹ For example, one kind of logical fallacy that I discuss in the next chapter on logic is called ad hominem, a situation in which a person criticizes the claim of another by disparaging the other person, as opposed to examining that person’s claim on its own merits. In my experience, I have heard marketers discount marketing ideas that come, say, from the accounting department because, after all, “what do accountants know about marketing?”

More specifically, I introduce three general categories of logical reasoning that take the following forms: deductive arguments, inductive arguments, and Inference to the Best Explanation (IBE). Then, I discuss how they apply to marketing decision making.

In logic, an argument is a collection of statements that attempts to prove a conclusion.² Marketing arguments generally contain one or more premises plus a conclusion or “call to action.” For example, you may conclude that your company needs to invest resources in product development based on the premise that consumer needs and wants change over time. Additionally, I extend IBE to what has come to be called, Inference to the Loveliest Explanation (ILE). The next chapter further extends this introductory discussion to forming logical arguments and avoiding logical fallacies.

Deductive Arguments

A valid deductive argument is one in which “it is not possible for the premises all to be true while the conclusion is false.”³ Consider the following deductive argument.

Premise A:	Effective decision making about advertising media expenditures requires a basic understanding of logic
Premise B:	Marketers want to make effective decisions about media expenditures
Conclusion:	Therefore, marketers require a basic understanding of logic

If you accept that this argument’s premises are true, the conclusion has to be true as well. As you might suspect, there are no practical examples of valid deductive inferences in marketing because of the probabilistic nature of predicting consumer and organizational behavior (refer to Chapter 8: Causality). Consequently, if you could make an argument using premises that deductively guaranteed a marketing outcome (such as a certain proportion of consumers would purchase your product) then there would be no product failures, and you would be a marketing genius. In other words, conclusions of arguments in marketing are rarely if ever guaranteed, but in a valid deductive argument, they are.

You may be surprised to learn that the great deducer, Sherlock Holmes, did not solve his crimes deductively at all, despite his many claims. For example, the following is from The Adventure of the Speckled Band (emphasis added):⁴

Dr. Watson: “You speak of danger. You have evidently seen more in these rooms than was visible to me.”

Holmes: “No, but I fancy that I may have deduced a little more. I imagine that you saw all that I did.”

Dr. Watson: “I saw nothing remarkable save the bell-rope, and what purpose that could answer I confess is more than I can imagine.”

Holmes: “You saw the ventilator, too?”

Dr. Watson: “Yes, but I do not think that it is such a very unusual thing to have a small opening between two rooms. It was so small a rat could hardly pass through.”

Holmes: “I knew that we should find a ventilator before ever we came to Stoke Moran.”

Dr. Watson: “My dear Holmes!”

Holmes: “Oh, yes, I did. You remember in her statement she said that her sister could smell Dr. Roylott’s cigar. Now, of course that suggested at once that there must be a communication between the two rooms. It could only be a small one, or it would have been remarked upon at the Coroner’s inquiry. I deduced a ventilator.”

Dr. Roylott or someone else for that matter could have been in the bedroom smoking a cigar prior to Miss Stoner’s sister (who, I regret to inform you, was subsequently murdered) entering it. The odor of the cigar did not guarantee that there was a small ventilator between the two rooms. It was inductive, not deductive, logic that led Holmes to conclude that there was a ventilator in the room!

One final bit of terminology you should know: deductive arguments are either valid or invalid. A valid deductive argument is one in which, if the premises are true, the conclusion has to be true. This is sometimes referred to as the “logic condition” of a deductive argument. If the premises are true, then the “truth condition” is said to be fulfilled. If the logic and truth conditions are met, the deductive argument is then said to be sound. If one or more premises are not true in a valid argument, the argument is said to be unsound.

Inductive Arguments

In inductive arguments, conclusions are not guaranteed by their premises, as is the case with valid deductive arguments (assuming the premises of the deductive argument are true). For example, you may offer the following inductive argument supporting the development of a new product: (By convention, the field of logic uses a double line to separate premises from a conclusion in an inductive argument, while using a single line in a deductive argument.)

Premise A:	Consumers have the following needs: X1, X2, and X3
Premise B:	Consumers do not believe that any product on the market currently meets these needs
Premise C:	Consumers are willing to pay up to Y-dollars for such a product
Premise D:	We can make this product and meet consumer needs X1, X2, and X3
Premise E:	Our per-product cost to manufacture, distribute, and market this product will be 20% of Y-dollars
Conclusion:	Therefore, we will make money by manufacturing and marketing this product

Marketers know only too well that the premises of such arguments do not guarantee the argument’s conclusion—even if the premises are true! For example, a competitor may also introduce a similar product that consumers prefer over yours; your company may inadvertently manufacture a faulty product that consumers reject; or, by the time you introduce the product, consumer preferences may have changed.

As I described in the discussion of deductive arguments, you should know the following nomenclature issues regarding inductive arguments. Inductive arguments are not “valid” because their conclusions are not guaranteed by true premises, as is the case with a valid deductive argument. In contrast, inductive arguments are said to range from “weak” to “strong” depending on the plausibility of the argument’s conclusion—the greater the plausibility (assuming true premises), the stronger the argument. If an inductive argument is strong and the premises are true, the argument is said to be cogent. So ideally we all want to make strong, cogent marketing arguments, based on inductive logic.

Before we move on, I want to redefine the “truth condition,” which I discussed with respect to both deductive and inductive arguments, to make it more practical in your everyday job as a marketer. The important point here is that to make good arguments, the premises should be plausible to your audience—absolute, infallible truth is not required.

Truth and plausibility are both desirable features of our premises. But for real-world argumentative purposes, where one is offering an audience reasons to accept a conclusion, what matters is whether they judge your premises to be plausible. If they’re judged implausible, then your argument will not have provided that audience “good reasons” to accept that conclusion.⁵

In the next sections, I provide generally accepted definitions of two forms of induction: enumerative induction and IBE, discuss their relevance to marketing, and describe three inductive reasoning tools first proposed by the nineteenth-century British philosopher, John Stewart Mill (1806–1873).

Enumerative Induction

This form of induction is “any inference from the past behavior of things to their future behavior,”⁶ based solely on statistical data. Here is a marketing example:

Premise:	Sales in each of the past years has increased 3%
Conclusion:	Therefore, sales this year will increase 3%

Enumerative induction is sometimes referred to as “More of the Same” induction. Consider the story of Blockbuster as recounted in a US News and World Report article:

“This video-rental chain survived the transition from VHS to DVD just fine—but then failed to adapt to the next big change. Blockbuster remained flat-footed when Netflix started sending videos through the mail, cable and phone companies started offering video-on-demand, and Redbox started renting videos for a buck a night through vending machines. Now that video streams through computers and phones, Blockbuster’s conventional retail outlets seem hopelessly outdated. The firm is closing hundreds of stores, working off debt, and copying some of its competitors’ moves, with a fighting chance to catch up. But it’s now chasing its industry instead of leading it.”⁷

Blockbuster’s “More of the Same” strategy to distribute primarily through its stores and only recently offer a mail option drove it to Chapter 11. A similar fate is befalling Best Buy, which stuck too long to its traditional “big box” strategy and now finds it fortunes floundering “in the face of growing competition online, [to which its founder Richard Schultz]…was blamed in part for the company’s slow-footed response.”⁸

That decisions driven by enumerative induction can lead to ruinous results is supported by a University of Pennsylvania study suggesting that the success or failure of an organization almost always revolves around its ability to recognize when “More of the Same” should be abandoned. The authors state that many companies fail to recognize when they should change course and, consequently, “the average lifespan of a major corporation isn’t very long. If current trends hold, only one-quarter of today’s S&P 500 companies will be part of the index by 2020, and the other three-quarters probably don’t even exist yet.”⁹ Therefore, companies that rely too much on past success to guide their marketing future are fodder for forensic journalists and venture capitalists. We read about these companies all the time that vainly cling to worn-out business models or practices—American Airlines, Kodak, Sears, K-Mart, and AOL, just to name a few. When you think about it, overreliance on “More of the Same” inductive logic to support ineffective business models should get companies into trouble because the goal of companies is to make sure the future is not like the past.

Inference to the Best Explanation (IBE)

I am sure that you have been in many meetings in which executives have debated the relative merits of alternative hypotheses explaining a given marketing phenomenon. Why did sales decline last quarter? Why is Sales Region A more profitable than Sales Region B? For example, I conducted a study for a national retailer that wanted to understand why sales among members of its loyalty club were not growing. Given the information available, management inferred the following alternative hypotheses (denoted “H”) explaining this “marketing phenomenon:”

• H1: Club members were shopping more frequently at competitor stores.

• H2: No significant changes had been made to the club since its inception (e.g., the club’s “look and feel” and member benefits had virtually remained unchanged).

• H3: General demand for the stores’ products had declined.

Prior to conducting research to explore this issue in more detail, the client considered H2, if it were true, to be the “best explanation” of the situation, reflecting a type of inference called IBE. This is a type of explanatory inference in which one is presented with (a) some phenomenon to be explained—in our example, depressed loyalty club sales—and (b) competing explanations of this phenomenon—the three hypotheses presented above. One accepts as the best inference, tentatively at least, the hypothesis that, if true, best explains the phenomenon. In this case, management’s IBE was H2. Note that one is not compelled to characterize any inference as the “best”—we are free to reject all competing hypotheses in a given situation if we do not feel that any one is strong enough to warrant the adjective “best.”

In comparing and contrasting IBE to enumerative induction, you can see that both are forms of inductive reasoning. This means that their conclusions—“sales will increase 3% next year” or that loyalty club sales will continue to be anemic—are not guaranteed to be true even if the premises supporting them are true. It may be true that sales have historically increased by 3% but that does not mean that they will increase by that same amount next year. It may be true that no changes have been made to the loyalty program and it therefore has lost its luster in consumers’ minds, but those facts may not account for the decline in loyalty club sales.

From IBE to ILE

What makes one inference stronger than another? This is a complicated question that Chapter 12, Theory Development, discusses in more detail. For now, I want to draw your attention to an extension of IBE that goes by the acronym ILE, popularized by Cambridge University’s Peter Lipton (1954–2007).¹⁰ The term, lovely, comes from the field of epistemology, and refers to the extent an “explanation … would, if correct, provide the greatest degree of understanding”¹¹ (italic added). In this context, loveliness connotes understanding.

Let’s look at these terms—explanation, lovely, and understanding—more closely. An explanation is a statement that gives us a reason for believing in something. Thus, an answer to a “why-question” is an explanation. The three hypotheses discussed previously are explanations for why loyalty member sales had not grown. To describe how explanation contrasts to the other terms, I give two examples, one from physics and another from marketing.

If a cue ball strikes an eight-ball, we could simply explain the movement of the eight-ball by saying “it was hit by the cue ball.” Alternatively, we could explain the movement of the eight-ball by saying that it was caused by the transfer of kinetic energy from the cue ball to the eight-ball, and employ in our account Newton’s three laws of motion:

• First Law of Motion: An object remains at rest unless acted upon by another force.

• Second Law of Motion: Acceleration is produced when a force acts on an object.

• Third Law of Motion: For every action there is an equal and opposite reaction.

The alternative explanation gives us a greater “understanding” of the event, even though our first explanation was true. True explanations, therefore, vary in their degree of “loveliness,” from being “less lovely” to “lovelier.” Understanding, therefore, is an attribute of an explanation—the lovelier a true explanation is, the greater understanding we have of what is being explained.

Among a group of plausible explanations of a given marketing phenomenon, marketers should use ILE as their guide, as opposed to IBE. Why? Because “loveliness” is a characteristic of an explanation that helps marketers understand how the marketing environment interacts with consumer beliefs, feelings, and intentions to influence consumer purchasing behavior. The more we understand what is causing consumer purchasing behavior, the better able we are to change it. See Figure 10.1A and 10.1B. Both diagrams reflect a hypothetical characterization of factors influencing company loyalty.

In this example, we assume that both models are “true,” in the sense that product and service quality are the primary drivers of product loyalty; however, Figure 10.1B. gives us a greater understanding of the aspects of product quality (i.e., perceived durability and reliability) and service quality (i.e., perceived reactive and proactive service) that influence product loyalty. The explanation modeled in Figure 10.1B, therefore, is “lovelier” than that modeled in Figure 10.1A. (Note: the example above builds on Chapter 7: Attributes Versus Constructs, and demonstrates the significance of identifying and understanding those constructs that most affect consumer purchasing behavior).

Developing lovelier explanations of marketing phenomenon will help you better understand why stuff happens—“What things, processes, and mechanisms bring about the world we experience?” With this understanding, we are sometimes able, in varying degrees, to predict and control our world, or to influence markets and consumers.

To answer the why-questions, you need to use investigative methods to explore and identify potential cause-and-effect relationships explaining consumer purchasing behavior. British Philosopher John Stewart Mill is well known for his “methods of induction.” Three of these—Method of Agreement, Method of Difference, and Method of Concomitant Variations—are often employed when conducting marketing research.

Inductive Reasoning Tools

Method of Agreement¹²

You go to a party where food is served and six people get sick shortly after eating. If the only food that they all ate was the spinach dip, the spinach dip most likely caused the food poisoning. Mill describes his Method of Agreement as follows:

If two or more instances of the phenomenon under investigation [six people got sick] have only one circumstance in common [they all ate the spinach dip], the circumstance in which alone all the instances agree [eating the spinach dip], is the cause (or effect) of the given phenomenon.¹³

Assume we observe the following two sets of conditions:

• A B C D occur with 1 2 3 4

• A E F G occur with 1 5 6 7

• Therefore, A is the cause of 1

We infer that A is related to 1 because they are the only events in common. For example, if two cars—condition A—only get flat tires when they drive down Main Street—condition 1—we would infer that there is something peculiar about Main Street that is causing these car owner’s misfortune (e.g., a roofing company is headquartered on Main Street, and their trucks carelessly spill roofing nails on the street). Clearly, Mills Method of Agreement reflects an ideal condition in which the only feature shared by the cars is driving down Main Street. The cars also may share other features, such as their color, but our reasoning tells us that color is not a cause of their getting flat tires when driving down Main Street.

I did a study for the same retailer discussed previously to better understand the top 20% of its loyalty club members, as measured by individual member sales over a six-month period (condition A). “Why are these members spending so much money with us?” We attempted to discover what these very profitable customers had in common. Obviously, they shared many characteristics such as gender, age range, and approximate household income. Our task was to use our reasoning and knowledge of consumer motivation to infer what relevant factors they shared that might also serve to cause them to spend relatively more money at the company’s stores compared with other club members. Based on a series of qualitative and quantitative research studies, we discovered that one common trait they shared, which could be considered a good explanation of their behavior, was their disposition to be do-it-yourselfers (condition 1). Having a better understanding of the needs and wants of these do-it-yourselfers enabled the company to develop promotional and educational programs throughout the year to better meet this customer segment’s needs.

Method of Difference

Two people go to dinner. One has lasagna and the other has steak; otherwise they ate and drank the same items. If the one eating the lasagna gets sick, it’s likely that the lasagna caused the sickness. Mill describes his Method of Difference as follows:

If an instance in which the phenomenon under investigation occurs [a person gets sick at a restaurant on a given day], and an instance in which it does not occur [another person does not get sick at the same restaurant on the same day], have every circumstance save one in common [one person had lasagna and the other steak], that one occurring only in the former; the circumstance in which alone the two instances differ [eating the lasagna], is the effect, or cause, or a necessary part of the cause, of the phenomenon.¹⁴

Assume we observe the following two sets of conditions:

• A B C D occur with 1 2 3 4

• B C D occur with 2 3 4

We infer that A is related to 1 because when A is absent 1 is absent as well. For example: A common question all companies have is, “Why are some customers more loyal than others?” Here you might employ Mill’s Method of Difference, by interviewing two groups of customers—loyal and nonloyal customers. The following is a brief example based on a recent Wall Street Journal (WSJ) article about Sears’ growing misfortunes.¹⁵

We expect both groups to differ on many characteristics; however, by employing our reasoning skills, we can begin to uncover those differences that are causal in nature and help us explain these two different levels of store loyalty. In the Sears’ example, A and 1 fulfill Mill’s Method of Difference criteria. Compared to the loyal Sears customer, the less loyal ones perceive Sears as delivering poor service (A) and not being up-to-date (1). Problems in these two areas apparently have been growing at Sears ever since Sears Holdings was created in 2005. In attempting to answer the why question, the WSJ article quotes Customer Growth Partners, a New Canaan, CT-based marketing research and consulting firm: “With these ‘dead man walking’ stores, the objective of the parent company is not to maximize [store] productivity but milk it for what little it has left before it can sell the property.” Sears’ growing problem is that the ratio of less to more loyal customers is rapidly increasing.

Method of Concomitant Variation

Keeping with our epicurean examples: If you find that your party guests are more talkative and joyful the more liquor you serve at a party, and the only factor that changes among your parties is the amount of liquor served, then changes in the amount of liquor served is causing your guests to be more giddy. This method of inference focuses solely on the correlation between two variables. Mills describes this method as follows:

Whatever phenomenon varies in any manner whenever another phenomenon varies in some particular manner, is either a cause or an effect of that phenomenon, or is connected with it through some fact of causation.¹⁶

Mill’s Method of Concomitant Variation, therefore, is defined as follows:

• A B C 1 2 3

• Both A and 1 change in magnitude concurrently, with no corresponding changes among the other variables.

Concomitant variation—or “correlation”—among variables makes the task of uncovering causation difficult, as discussed in Chapter 8: Causality, simply because correlation does not imply causation.

“Why Do I Need to Know All This?”

If you want to apply scientific reasoning to marketing, you need to know the differences between the various forms of logic arguments, their strengths, weaknesses, and how you can sharpen your logical reasoning skills. So here’s why you should know this.

• Deductive arguments do not apply: This form of logic is the only one in which the premises of your argument, if true, guarantee your conclusion, and deductive arguments simply do not work in explaining human behavior, or in Sherlock Holmes’ case, when trying to solve a crime. Therefore, you rely on inductive logic arguments in which your premises, even if they are true, do not guarantee your conclusion. Because of this, we all should subscribe to the dictates of epistemic virtue. Epistemic virtue means that a person who “is conscientious in the way in which she forms her beliefs will be more likely to form true beliefs than someone who … [does so] unconscientiously.”²⁰ Forming true beliefs makes for better marketing decisions.

• Enumerative induction is useless in helping to understand consumer behavior. Try to avoid making marketing decisions based on “More of the Same” inferences. This form of logic may be all you need to conclude that the Sun will rise tomorrow. But marketing decisions need to be based on an understanding of cause-and-effect relationships—to the extent possible and as humanly imperfect as our understanding of these relationships may be. Moreover, marketers often need to develop “constructs” in an effort to understand consumer behavior (see Chapter 7: Constructs), and in this regard, enumerative induction is of little value.

• Recognize IBE’s strengths and weaknesses. IBE appeals to explanation, and that is its strength. In Chapter 12, we build on this idea by describing the role theory plays in explaining consumer purchasing behavior. Nonetheless, IBE can still mislead if one’s IBE is the “best of a bad lot.” Therefore, strive to develop the “loveliest” explanations—those that attempt to understand cause-and-effect relationships in describing consumer purchasing behavior. There is no guarantee that ILE will avoid the “best of the bad lot” scenario, but “lovelier” explanations should be more truth-conducive than other forms of inductive inference that possess little if any explanatory power (e.g., enumerative induction).

• Test your inductive arguments. Most likely, any proposition you make to management is an inductive argument. Test your inductive arguments as follows (note: the next chapter addresses this topic in greater detail):

◦ Assume your premises are plausible. How logically sound, then, is your conclusion or recommendation based on these premises? The stronger the logical connection, the better.

◦ How plausible are your premises? Make them as plausible as you can. For example, use premises that have been validated empirically or have the support of an industry expert.

• Sharpen your thinking skills. Most of us cannot hope to possess the reasoning power of an Einstein or a Steve Jobs, but that does not mean we cannot improve our reasoning skills by thinking more clearly and effectively. As Alfred Mander discusses in his book, Logic for the Millions:

Thinking is skilled work. It is not true that we are naturally endowed with the ability to think clearly and logically—without learning how, or without practicing. People with untrained minds should no more expect to think clearly and logically than people who have never learned and never practiced can expect to find themselves good carpenters, golfers, bridge players, or pianists.²¹

In this light, a great Internet resource for improving all of your logical reasoning skills is The Critical Thinking Academy (http://www.criticalthinkeracademy.com/), created by Dr. Kevin deLaplante, associate professor of philosophy at Iowa State University.