Although missing information is often important, people are surprisingly insensitive to omissions (or unmentioned options, features, issues, or possibilities). Neglecting important omissions has serious consequences for decision making.
Frank R. Kardes and David M. Sanbonmatsu
When Sherlock Holmes asked Dr. Watson to consider the previous night’s “curious incident” involving a dog, Watson replied that nothing happened (in “The Silver Blaze”). “That was the curious incident,” observed Holmes. This clue enabled Holmes to deduce that the murderer must have been someone familiar to the victim’s dog because the dog did not bark when the murderer appeared. Most people would miss this important clue because most people, like Watson, pay little attention to non-occurrences (Ross 1977).
Nonevents are important in other situations as well. When forming beliefs about cause and effect, people typically focus on cases in which the cause and the effect co-occur.
Control group cases involving the absence of the cause tend to be neglected even though such cases are essential for establishing causality. In fact, including a control group in experimental design did not gain widespread popularity until the publication of A System of Logic by John Stuart Mill in 1887. Scientists failed to recognize the critical importance of a control group until relatively recently in the history of science because even scientists are remarkably insensitive to the absence of a property (such as the absence of a cause).
Another striking example of the difficulty people experience when attempting to think about non-incidents is evident in the history of zero. Numerical symbols first appeared in 3400 B.C. However, no symbol for zero appeared until many centuries later when mathematicians began to use zero as a placeholder to replace blank spaces to distinguish between numbers such as 1, 10, 100, 1000, etc. (Ifrah 1985). Zero served as a mere placeholder and was not used as a symbol for nothingness or the absence of quantity until about 800 A.D. It took early mathematicians thousands of years to develop the crucial concept of zero.
In everyday life, consumers typically make decisions based on thumbnail sketches of products described in advertisements and other biased promotional materials (Kardes 2002). Why are people willing to make decisions based on scraps of information provided by clearly partisan sources? People are accustomed to using whatever evidence is readily available to them—however scant—at the expense of other information that persuaders fail to mention (Sanbonmatsu, Kardes, Houghton, Ho, and Posavac in press). Omission neglect (insensitivity to unmentioned options, features, issues, or possibilities) is particularly problematic given the nature of the world. The amount of information used to describe various alternatives—such as various political candidates, job applicants, defendants, consumer goods, healthcare products, medical procedures, or possible decision outcomes—typically varies dramatically across situations. Reports, speeches, interviews, advertisements, and media coverage provide various levels of detail about different alternatives. Some alternatives are discussed at considerable length, while others are described only briefly. To some extent, nearly everything is described in terms of limited, incomplete, or fragmentary evidence.
Research on omission neglect has shown that people often fail to detect the absence of important missing information, and this leads people to form strong beliefs on the basis of weak evidence (Sanbonmatsu et al. 1991, 1992, 1997). Strong beliefs are beliefs that are overly extreme (those that are highly favorable or highly unfavorable when the available evidence is only moderately favorable or moderately unfavorable, respectively) and held with a high degree of confidence. In general, people should form more extreme beliefs when more rather than less information is available (the set-size effect; see Anderson 1981). However, when people are insensitive to omissions, people form extreme beliefs regardless of how little is known about a topic (Sanbonmatsu et al. 1991, 1992, 1997, in press).
For example, people should form more favorable evaluations of a camera when the camera performs well on eight attributes as opposed to only four attributes. When consumers are unknowledgeable or moderately knowledgeable about cameras, however, they form equally favorable attitudes toward the target camera regardless of how much or how little information is presented (Sanbonmatsu et al. 1992). Only the small subset of consumers who are highly knowledgeable about cameras form more favorable attitudes when the target camera is described by eight (versus four) favorable attributes.
Similar results are observed in inferential judgments (those that go beyond the information given; see Sanbonmatsu et al. 1991). Consumers received a brief description of a new ten-speed bicycle and were asked to judge its durability even though no information about durability was provided. When consumers inferred durability immediately after reading the description, they realized that no information about durability was available and they formed moderately favorable inferences about durability. However, when consumers inferred durability one week after reading the description, extremely favorable and confidently held inferences were formed. This result was observed even though memory tests revealed that after the one-week delay people forgot most of the information that had been presented. That is, consumers’ inferences were more extreme and were held with greater confidence when they remembered a little than when they remembered a lot (the remembering-less-and-inferring-more effect). In other words, people are often most confident when they are most wrong.
Inferences are also influenced by beliefs about the strength of the relationship between presented and missing information. When presented and missing information are highly related, people can make inferences about unmentioned information based on the presented information. For example, many consumers assume that price and quality are highly related (quality increases as price increases; you get what you pay for). Consequently, consumers infer that a high price signals high quality. However, consumers are unlikely to attempt to form an inference about a missing attribute if they fail to notice that information is missing. When a large amount of information is presented about one product and a small amount is presented for another, consumers are less sensitive to missing information when the product described by the larger amount of information is presented first rather than second (Kardes and Sanbonmatsu 1993). Consequently, people are less likely to form inferences and are more likely to prefer the two products equally when the product described by the larger amount of information is presented first.
The results of research published by Sanbonmatsu, Kardes, and their colleagues suggest that omission neglect occurs because missing information is not very salient or attention drawing. To the extent that this is true, omission neglect should be reduced when the salience of missing information is enhanced. This can be accomplished by manipulating variables of motivation and context that increase sensitivity to omissions and lead to more appropriate judgments. More moderate judgments are formed when people are sensitive to omissions due to an explicit warning that the given information is incomplete (Sanbonmatsu et al. 1992), due to very high levels of prior knowledge about the target object or issue (Sanbonmatsu et al. 1991, 1992), or due to comparison processes that make it painfully obvious that some objects are described by a large amount of information whereas others are described by a small amount (Sanbonmatsu et al. 1997, in press). Moderate judgments are more accurate than extreme judgments when information is limited (Griffin and Tversky 1992), are more readily updated as new information becomes available (Cialdini, Levy, Herman, and Evenbeck 1973), and are more justifiable to oneself and to others (Lerner and Tetlock 1999; Shafir, Simonson, and Tversky 1993).
Although beliefs are generally more reasonable when people are sensitive to omissions, such awareness is very difficult to foster. People frequently and typically neglect omissions. Research on the tendency to learn more quickly when a distinguishing feature or symbol is present versus absent has shown that people find it very difficult to learn that the absence of a feature is informative (Newman, Wolff, and Hearst 1980). Even when the presence or absence of a feature are equally informative, the relationship between the predictive feature and a desired event (e.g., food, water, positive feedback) is learned much more rapidly when the feature is present as opposed to absent. The feature-positive effect is so ubiquitous that it has been observed for humans, pigeons, rats, cats, and monkeys, and most young children and animals never learn that the absence of a feature is informative (Newman et al. 1980).
Research on the fault-tree effect also shows that it is extremely difficult to make people sensitive to omissions (Russo and Kolzow 1994). A fault tree is a list of possible reasons for system failure (e.g., a list of possible reasons why an automobile will not start or a machine malfunctions). Many managers believe that a fault tree is a useful troubleshooting device that helps busy employees to identify the cause of a problem more quickly. It is easier to consult a list than to think about all of the things that can go wrong from scratch. Fault trees are commonly used to troubleshoot complex systems such as devices used in airplanes and nuclear power plants. However, when using fault trees, people typically underestimate the likelihood that an unmentioned alternative could be the root cause of a problem. This result is observed regardless of how many or how few possibilities are included in the fault tree (analogous to the finding that extreme beliefs are formed regardless of how many or how few attributes are presented in a product description; see Sanbonmarsu et al. 1992).
Missing information is also neglected in the Ellsberg paradox (discovered by the famous economist who advised President Nixon): People prefer to bet on known probabilities rather than on unknown probabilities (Fox and Weber 2002). Most people are indifferent between red and black when betting on whether a red or black marble will be drawn from a jar containing half each red and black marbles. Most people are also indifferent between red and black when betting on whether a red or black marble will be drawn from a jar of red and black marbles with an unknown distribution. When given a choice between the two jars, however, most people prefer to bet on the jar with the 50/50 distribution rather than the jar with the unknown distribution. Just as it is the case that comparative contexts (e.g., judgment contexts involving descriptions of more than one product) increase sensitivity to missing attributes (e.g., Sanbonmatsu et al. 1997, in press), so too it is the case that comparative contexts (e.g., choice contexts involving more than one gamble) increase sensitivity to missing probabilities (Fox and Weber 2002).
People are insensitive to missing cases (as well as to missing attributes, features, possibilities, and probability distributions) and this makes it difficult to learn the relationship between two variables (e.g., X and Y, cloud seeding and rain, holistic medicine and good health). People often focus on cases involving the presence of both variables and ignore cases involving the absence of one or both variables (for reviews of covariation estimation, see Gilovich 1991, 1997; Nisbett and Ross 1980; Sanbonmatsu, Posavac, Kardes, and Mantel 1998). Statistically, all four cells of the 2 (X present or absent) by 2 (Y present or absent) contingency table are equally important (see table 1). However, most people focus on the X present/Y present cell exclusively. This can lead people to see relationships where none exist. For example, if a large number of people who take holistic medicines enjoy good health (X present/Y present), many people conclude that holistic medicines are beneficial. This conclusion is unwarranted, however, because a large number of people who take holistic medicines do not enjoy good health (X present/Y absent), a large number of people who do not take holistic medicines enjoy good health (X absent/Y present), and a large number of people who do not take holistic medicines do not enjoy good health (X absent/Y absent).
When attempting to assess the accuracy of their beliefs, people focus more heavily on the evidence that supports their beliefs than on the evidence that fails to support their beliefs (for reviews of confirmation bias, see Gilovich 1991, 1997; Nisbett and Ross 1980; Sanbonmatsu, Posavac, Kardes, and Mantel 1998). Although this bias is more pronounced when people want to protect their beliefs, it occurs even when people attempt to be objective. Supportive evidence about the occurrence of an expected outcome is attention-drawing and memorable. Unsupportive evidence about the nonoccurrence of an expected outcome is ignored or discounted as a fluke. Consequently, people’s beliefs tend to be remarkably resilient to evidence, and erroneous beliefs about psychology, business, law, and medicine persevere. For example, many people believe in ESP and subliminal persuasion despite the lack of scientific evidence for these phenomena. Investors continue to believe that they can beat the stock market even though the most sophisticated mathematical models (e.g., nonlinear regression, chaos theory) are unable to do so. Jurors believe that their verdicts are not influenced by inadmissible evidence, despite evidence to the contrary, and medical patients spend millions on useless holistic medicines, Laetrile clinics, psychic surgeons, and faith healers (Gilovich 1991).
People are insensitive to so many different types of omissions that developing effective debiasing procedures is daunting. However, research on this topic suggests that it might not be necessary to encourage people to think about specific omissions (Sanbonmatsu et al. 1997). Instead, merely increasing people’s awareness that something is missing, even if they do not know what, can improve judgment and decision making. After reading a large (versus small) amount of information about an irrelevant topic (i.e., soybeans), consumers form more moderate and appropriate evaluations of a briefly described product (i.e., an automobile or camera). Detecting unspecified omissions helps people to recognize that their judgments are based on limited or weak evidence.
One way to reduce the degree to which people overestimate the importance of the presented information is to encourage them to consider a wide range of attributes by asking them to evaluate two products described on different attribute dimensions. Another way is to ask people to rank the importance of each attribute in a lengthy list of attributes before asking them to read a brief product description.
To summarize, inappropriately extreme and confidently held judgments are formed when people overestimate the importance of the presented information and underestimate the importance of information that was not presented. It is surprising that people focus so readily and so heavily on the presented information, given that the presence or absence of information about an attribute has no influence on the objective importance of the attribute (e.g., miles per gallon is an important attribute even if no information about this attribute is provided for a specific brand). This effect is so strong that the presented information may actually interfere with the ability to think about unmentioned information, The more people focus on the presented information, the more difficult it may be to consider attributes that were not included in a product description.
Our minds may have evolved to process stimuli we encounter, not stimuli we do not encounter. The presence of a predator is a relatively rare event that requires immediate attention and action. By contrast, the absence of a predator is a commonplace event that does nor raise any immediate concerns. Because infrequently encountered stimuli are more informative, it is more efficient to focus on stimuli that are encountered rather than on stimuli that are not. Only if a large number of stimulus-outcome associations are learned does it become unnecessary to develop a system that monitors the absence of relationships among stimuli and outcomes (Newman et al. 1980). Consequently, associations among occurrences are learned more easily and rapidly than associations involving non-occurrences.
People are accustomed to making judgments and decisions based on whatever information they happen to encounter. Regardless of how much and what information is used, omission neglect is common because missing information is not salient, people overestimate the importance of readily available information, and presented information interferes with the ability to think about missing information. It is just as important to think critically about what we do not know as what we do know.
Anderson, N.H. 1981. Foundations of Information Integration Theory. San Diego, Calif.: Academic Press.
Cialdini, R.B., A. Levy, C.P. Herman, and S. Evenbeck. 1973. Attitudinal politics: The strategy of moderation. Journal of Personality and Social Psychology 25: 100-108.
Fox, C.R., and M. Weber. 2002. Ambiguity aversion, comparative ignorance, and decision context. Organizational Behavior and Human Decision Processes 88: 476-498.
Gilovich, T. 1991. How We Know What Isn’t So: The Fallibility of Human Reason in Everyday Life. New York: Free Press.
____, 1997. Some systematic biases of everyday judgment. SKEPTECAL INQUIRER 21(2), March/April: 31-35.
Griffin, D., and A. Tversky. 1992. The weighing of evidence and the determinants of confidence. Cognitive Psychology 24: 411-435.
Ifrah, G. 1985. From One to Zero: A Universal History of Numbers. New York: Viking.
Kardes, F. R. 2002. Consumer Behavior and Managerial Decision Making. Upper Saddle River, N.J.: Prentice Hall.
Kardes, F.R., and D.M. Sanbonmatsu, 1993. Direction of comparison, expected feature correlation, and the set-size effect in preference judgment. Journal of Consumer Psychology 2: 39-54.
Lerner, J.S., and P.E. Tetlock. 1999. Accounting for the effects of accountability. Psychological Bulletin 125: 255-275.
Mill, J.S. 1887. A System of Logic. New York: Harper & Brothers.
Newman, J., W.T. Wolff, and E. Hearst. 1980. The feature-positive effect in adult human subjects. Journal of Experimental Psychology: Human Learning and Memory 6: 630-650.
Nisbett, R.E., and L. Ross. 1980. Human Inference: Strategies and Shortcomings of Social Judgment. Englewood Cliffs, N.J.: Prentice-Hall.
Ross, L. 1977. The intuitive psychologist and his shortcomings: Distortions in the attribution process. Advances in Experimental Social Psychology 10: 174-214.
Russo, J.E., and K.J. Kolzow. 1994. Where is the fault in f~sult trees? Journal of Experimental Psychology: Human Perception and Performance 20: 17-32.
Sanbonmatsu, D.M., F.R. Kardes, and P.M. Herr, 1992. The role of prior knowledge and missing information in mulriatrribute evaluation. Organizational Behavior and Human Decision Processes 51: 76-91.
Sanbonmarsu, D.M., F.R. Kardes, D.C. Houghton, E.A. Ho, and S.S. Posavac. In press. Overestimating the importance of the given information in multiattribute consumer judgment. Journal of Consumer Psychology.
Sanbonmatsu, D.M., F.R. Kardes, S.S. Posavac, and D.C Houghton. 1997. Contextual influences on judgment based on limited information. Organizational Behavior and Human Decision Processes 69: 251-264.
Sanbonmatsu, D.M., F.R. Kardes, and C. Sansone. 1991. Remembering less and inferring more: The effects of the timing of judgment on inferences about unknown attributes. Journal of Personality and Social Psychology 61: 546-554.
Sanbonmatsu, D.M., S.S. Posavac, F.R. Kardes, and S.P. Mantel. 1998. Selective hypothesis testing. Psychonomic Bulletin and Review 5:197-220.
Shafir, E., I. Simonson, and A. Tversky. 1993. Reason-based choice. Cognition 49: 11-36.
Frank R. Kardes is a professor of marketing at the University of Cincinnati and the author of Consumer Behavior and Managerial Decision Making. He is a Fellow of the American Psychological Association, the American Psychological Society, the Society for Consumer Psychology and the Society for Personality and Social Psychology
David M Sanbonmatsu, professor of psychology at the University of Utah, has published extensively on the topics of attitudes, stereotypes, and judgment and decision making. Much of the research summarized in this article was supported by NSF grants SBR 9308383 and SBR 930830 awarded to Frank R. Kardes and David M. Sanbonmatsu.
From Skeptical Inquirer, March/April, 2003, pp. 42-46. © 2003 by Skeptical Inquirer,