Acrwebsite.org

Making Probability Judgments of Future Product Failures: Packing versus Unpacking the
Dipayan (Dip) Biswas, Bentley College, USA L. Robin Keller, University of California, Irvine, USA Bidisha Burman, Appalachian State University, USA EXTENDED ABSTRACT
product (e.g., negative experience in the form of product failures).
While making product use or purchase decisions, consumers Koriat, Fiedler, and Bjork (2006) found that “in making conditional often make explicit or implicit judgments regarding potential predictions, people are subject to a prediction inflation bias, over- product failures in the future. In a related vein, through advertise- estimating the likelihood of occurrence of the stated outcome given ments and other actions, managers and regulators often have the the stated envisioned condition” (p. 430). This implies that if people flexibility to influence what types of information are presented to have had prior negative experiences with a product class, they are consumers. For instance, in their magazine advertisements and more likely to have unfavorable predictions regarding future prod- package inserts, anti-depressant drug PROZAC® unpacks the uct failures. However, more interestingly, people are likely to have adverse reactions related to “digestive system” in the form of a greater level of prediction inflation bias, when they can build “a “nausea, diarrhea, anorexia, dry mouth, and dyspepsia.” causal scenario leading from the condition to the outcome” (Koriat, Prior research has found that people often do not follow the extensional logic of probability theory while making probability We are proposing that a general negative experience with the judgments and are frequently influenced by the information format product in the past, would lead to a stronger causal mapping in the (Kahneman et al.1982). Specifically, in the context of the PROZAC® “packed” data condition, since a general experience can be more example outlined above, would consumers have differential judg- directly mapped with a general (than specific) problem in the future.
ments for “packed” (“adverse reaction related to the digestive In contrast, when the questions are unpacked, consumers are less system”) versus “unpacked” (“adverse reaction in the form of likely to form a direct causal mapping between the past experience nausea, diarrhea, anorexia, dry mouth, and dyspepsia”) informa- and making the conditional prediction regarding the future, for tion? Prior research (e.g., Fiedler and Armbruster 1994; Rottenstreich more specific problems. Hence, we hypothesized that a general and Tversky 1997; Tversky and Koehler 1994) would suggest that negative product experience in the past would enhance probability people would have a higher probability judgment for potential judgments regarding future product failures to a greater degree adverse reactions for the unpacked than for the equivalent packed when the problem is packed (than unpacked).
condition. This is because people tend to assign probabilities to In contrast, prior general positive experiences with a product descriptions of events, called hypotheses, rather than to actual would lead to lower probability judgments regarding future product events (Tversky and Koehler 1994). As a result, consumers’ prob- failures, to a greater extent for unpacked (than packed) conditions.
ability judgments for the focal hypothesis relative to the alternative This is because, since reliable performances by a product is the hypothesis are influenced by how events are described. In this norm (Meyvis and Janiszewski 2002), there will be a greater degree context, Support theory “predicts that the judged probability of an of “seizing” and “freezing” for the focal hypothesis under prior event increases by unpacking the focal hypothesis and decreases by positive experience conditions (Kruglanski and Freund 1983).
unpacking the alternative hypothesis” (Rottenstreich and Tversky In Study 1, the results of a 2 (packed vs. unpacked) X 3 1997, p 406). Support theory also predicts that the judged probabil- (negative prior experience vs. positive prior experience vs. no prior ity assigned to an unpacked description of an event is greater than experience) between-subjects experiment supported our hypoth- the packed description (Tversky and Koehler 1994). Hence, when the description of an event is unpacked into its individual compo- The two critical questions asked in Study 2 were: (1) Do nents, people are likely to assign an overall higher probability participants have higher probability judgments of future product judgment for the event’s occurrence. This phenomenon has been failures for “packed” versus “unpacked” formats when they them- termed as “subadditivity” in probability judgments. Many other selves generate the unpacking variables? (2) Do participants have research streams (e.g., Fiedler and Armbruster 1994; Fischhoff, increased probability judgments for unpacked versus packed con- Slovic and Lichtenstein 1978; Teigen 1974, 1983) have found ditions, in an increasing linear fashion? That is, if 4 levels of unpacking lead to greater probability judgments than a packed We extend the basic premise of information packing/unpack- scenario, will a 12-level unpacking lead to an even greater probabil- ing, and examined scenarios where participants themselves gener- ity judgment? We hypothesized that when the generation of un- ated the unpacking variables, unlike as in prior studies where packing variables is easy, consumers would have a higher probabil- participants were given explicit unpacking questions. We also ity judgment of future product failure for the mentally “unpacked” examined how prior general negative/positive experiences with the than “packed” conditions; when the generation of unpacking vari- product can influence the evaluation outcomes for packed versus ables is difficult, consumers would have similar probability judg- ments of future product failure for the “unpacked” and the “packed” When a consumer is asked to make probability judgments conditions. Moreover, these hypothesized relations will become about future product performances, in light of past experiences, in stronger for high (vs. low) need for cognitive closure participants.
essence it translates into making a conditional prediction. Condi- The results of 3 (packed condition vs. unpacked with 4 variables tional predictions conform to the format of asking participants the generated vs. unpacked with 12 variables generated) X 2 (Need for probability of an outcome y, given condition x has occurred (Koriat, Cognitive Closure: High vs. Low) between-subjects experiment Fiedler, and Bjork 2006). In the context of the present research, y can be conceptualized as the probability judgment regarding future Combined, the two experiments showed that a general prior product failure, while x can represent past experiences with the negative experience had a stronger impact on participant judg- Advances in Consumer Research
Volume 35, 2008
858 / Making Probability Judgments of Future Product Failures: Packing versus Unpacking the Problem
ments, for future product failures, in the packed (than the unpacked)condition, with the pattern of results getting reversed for priorpositive experience with the product. Also, when asked to generatethe unpacking variables, participants had higher probability judg-ments of future product failures only when they were able togenerate the unpacking variables with relative ease.
References
Fiedler, Klaus and Thomas Armbruster (1994), “Two Halfs May
be More Than One Whole: Category Split Effects onFrequency Illusions,” Journal of Personality and SocialPsychology, 66 (4), 633-45.
Fischhoff, Baruch, Paul Slovic, and Sarah Lichtenstein (1978), “Fault trees: Sensitivity of estimated failure probabilities toproblem representation,” Journal of Experimental Psychol-ogy: Human Perception Performance, 4, 330-344.
Kahneman, Daniel, Paul Slovic, and Amos Tversky (1982), Judgment under uncertainty: Heuristics and biases,Cambridge, England: Cambridge University Press.
Koriat, Asher, Klaus Fiedler, and Robert A. Bjork (2006), “Inflation of Conditional Predictions,” Journal of Experimen-tal Psychology: General, 135 (3), 429-447.
Kruglanski Arie W. and T. Freund (1983), “The freezing and unfreezing of lay inferences: Effects on impressionalprimacy, ethnic stereotyping, and numerical anchoring,”Journal of Experimental Social Psychology, 19, 448-468.
Meyvis, Tom and Chris Janiszewski (2002), “Consumers’ Beliefs about Product Benefits: The Effect of ObviouslyIrrelevant Product Information” Journal of ConsumerResearch March, 28(4), 618-635.
Rottenstreich, Yuval and Amos Tversky (1997), “Unpacking, Repacking, and Anchoring: Advances in Support Theory,”Psychological Review, 104, 406-15.
Teigen, Karl H. (1974), “Overestimation of Subjective Probabili- ties,” Scandinavian Journal of Psychology, 15, 56-62.
Tversky, Amos and Derek J. Koehler (1994), “Support Theory: A Nonextensional Representation of Subjective Probability,”Psychological Review, 101, 547-67.

Source: http://www.acrwebsite.org/volumes/v35/naacr_vol35_452.pdf