Society for Judgment and Decision Making 1997 Annual Meeting

Society for Judgment and Decision Making

1997 Annual Meeting

Poster Presentations

Poster Session #1

P1-1. Lehman BENSON III, Markus GROTH, & Lee Roy BEACH (The University of Arizona)

In pursuit of a model of perceived time pressure.

The relationship between time constraints for task completion and reported perceptions of time pressure were examined. The first experiment found that rated perceived time pressure (PP) was a function of the ratio, Rr/Ra, of required time (Rr) and available time (Ra), and that two alternative measurement methods yielded highly similar ratings. The second experiment showed that the importance of the task can be represented as PP = (Rr/Ra) + i, where i represents task importance. Implications and suggestions for future research are discussed.

P1-2. David H. EBENBACH, Colleen F. MOORE, & Jessica GERSHAW (Dept. of Psychology, Univ. of Wisconsin-Madison)

Assuming the worst: Environmental decisions in the context of missing information.

Across 2 studies, participants made decisions in environmental dilemmas. In each study, complete or incomplete information was given about three factors (local effects, ecosystem risks, economic effects, etc.). Participants also rated the importance of factors, and their assumptions about incomplete information. Results indicated that participants' assumptions about missing information strongly affected the way they used available information, as well as their ultimate decisions. Our findings imply that people facing difficult environmental problems may base decisions not only on available facts, but also on important but missing information.

P1-3. Cheryl Brown TRAVIS, Danny S. MOORE (Univ. of Tennessee), & Bruce E. TONN (Oak Ridge National Laboratory)

Building a tool for environmental decision making.

The Environmental Problem Inventory is a tool devised by researchers at the NSF National Center for Environmental Decision Making Research to assist in the first steps of environmental decision making. Dimensions of the inventory include ecological impact, health effects, economic considerations, identification of stakeholders, and regulatory issues as well as other factors. Results of a factor analysis of the Environmental Problem Inventory encompassing over 500 environmental problems at the local, state, and regional level indicated that information from both the natural and social science realms is necessary for issue diagnosis and problem characterization.

P1-4. Osvaldo F. MORERA (Survey Research Laboratory, Univ. of Illinois at Chicago)

A psychometric assessment of the divide and conquer principle in multiattribute decision making.

The principle of "divide and conquer" (DAC) suggests that decomposing multiattribute decisions enhances decision quality relative to holistic decision making (Ravinder & Kleinmuntz, 1991; Fischer 1977; Ravinder, 1992). Other evidence suggests that this generalization is dependent upon the choice of the criterion variable (Cornelius & Lyness, 1980; Lyness & Cornelius, 1982). A new comprehensive experimental framework designed to test the validity of the DAC principle was developed. Results from a recent study using the SMARTS procedure (Edwards & Barron, 1994) indicate that the generality of the DAC principle is indeed dependent upon the choice of the criterion variable.

P1-5. Holly ARROW (Dept. of Psychology, Univ. of Oregon), & Jonathan A. RHOADES (Dept. of Psychology, New York Univ.)

Face-to-face and computer-mediated decision-making in established groups.

This study examines the role of membership change and communication medium in group decision-making. Thirty established groups completed a hidden profile task either face to face or via computer. Each group had a "guest" member visiting from another established group. First, members individually reviewed materials on 20 job applicants, took notes, and decided which candidates fit their assigned criterion best. Next, the group discussed the candidates and made final recommendations. Results indicate that guests showed performance deficits similar to those observed for gender tokens (Lord & Saenz, 1985). Information processing strategies differed across communication media, but performance quality did not.

P1-6. Xiao-Ping CHEN (Dept. of Management of Organizations, Hong Kong Univ. of Science and Technology)

The effects of past group performance and the provision of public goods: Perceived criticality, group identity, and conformity.

This study investigated the effects of other members' behavior in the past on new members' cooperative choice in a step-level public goods dilemma. A newly developed experimental paradigm enabled us to test three major hypotheses in explaining high cooperation rates in social dilemmas: the group identity hypothesis, the perceived criticality hypothesis, and the conformity hypothesis. 168 business students participated in a laboratory experiment. The results suggest that (1) perceived criticality is more effective than group identity or perceived group norm in inducing cooperation; (2) group identity is necessary but not sufficient in evoking cooperation; and (3) group identity moderates the effect of perceived group norm on members' choices: Members conform to group cooperative norms when they highly identify with their group. The implications of these findings are discussed.

P1-7. Daniel GIGONE (Fuqua Sch. of Business, Duke Univ.)

Group discussion and small group decision making: Effects of task and subjective meaning of information.

The study explores issues related to the assumption that discussion content affects group decisions: What determines discussion content and what is the relationship between discussion content and decisions? The study also explored task differences. Groups judged fictional candidates. Members rated the value and weight of facts before and after each group decision. Task differences were few. Groups discussed facts with high weight ratings and facts about which members disagreed. After discussion, members agreed about facts which were discussed. Discussion did not predict group judgments, controlling for members' judgments. The findings are related to a model of group discussion and decisions.

P1-8. Joann L. KRAUSS (Dept. of Management & Systems, Washington State Univ.)

Improving individual decision-making during organizational crisis: A preliminary investigation.

Organizational crises are unexpected events that require an urgent decision. The characteristics of a crisis, urgency, ambiguity, uncertainty, and complexity, increases individual stress and hampers an individual's ability to make a decision. This poster session will examine two alternative methods of improving individual decision-making during crisis. It is hypothesized that individuals who receive training in either method will perform better on a crisis decision task than individuals who receive no training. It is further hypothesized that individuals who receive training in creative analysis will perform better on a crisis decision task than individuals who receive training in crisis management planning.

P1-9. R. Scott TINDALE (Loyola Univ. Chicago), Joseph FILKINS (DePaul Univ.), & Susan SHEFFEY (Jewish Vocational Services)

Perceptions of processes in decision-making groups.

Certain models of group decision making imply that some aspects of group process (e.g., majority effects) stem from shared norms about how groups operate. Two studies were conducted to assess whether potential group members had accurate perceptions of group processes. Results indicated that participants underestimated the power of faction size for juries and underestimated the power of correct alternatives for problem-solving groups.

P1-10. George CVETKOVICH, & Ryan O. MURPHY (Dept. of Psychology, Western Washington Univ.)

The asymmetry principal of trust: Disconfirming evidence.

Trust occurs when an individual allows another person or organization to make decisions for him/her. Slovic (1993) indicates that trust is asymetrical -- it is easier to loose it than to gain it. "Trust-increasing" and "trust decreasing" events used by Slovic were paired together. The asymmetry principal leads to the expectation, not confirmed in this study, that two-event descriptions, since they contain a trust decreasing event, should produce lower trust judgments than do single trust-increasing descriptions. The present evidence suggests that a trust impression, once initiated, tends to affect the meanings given to subsequently learned information.

P1-11. Maurice SCHWEITZER (Dept. of Management, Sch. of Business Administration, Univ. of Miami)

Omission, friendship, and fraud: Lies about material facts in negotiations.

Deception poses a particularly important problem for negotiations. Negotiators typically possess asymmetric information and can gain an advantage by misleading others. While some types of lies have been condoned, such as lies about reservation prices, both ethicists and legal scholars have classified lies about materially relevant facts as unacceptable. Results from this work, however, reveal that lies about material facts are pervasive in negotiations. Negotiators are more likely to lie by omission than commission, are more likely to lie to strangers than friends, and are particularly likely to lie to strangers who do not ask probing questions.

P1-12. David J. WEISS (California State Univ., Los Angeles), & Anthony D. ONG (Univ. of Southern California)

The truth is in there.

In estimating the frequencies of behaviors not carried out in public view, researchers accept the accuracy of respondents' reports. We explored two factors expected to influence the decision to reveal, anonymity vs. confidentiality and framing the question so that a behavior is reputedly commonplace or rare. A key feature was the inclusion in our survey of a socially disapproved behavior (cheating) for which we had validational information. The privacy variable had an enormous impact; of those who had cheated, 25% acknowledged having done so under confidentiality, but 74% admitted the behavior under anonymity. Question framing had no effect.

P1-13. Lyn M. VAN SWOL (Univ. of Illinois at Urbana-Champaign)

The effects of task type, interaction type, and expertise on acceptance of advice.

Study examines effects of task type (intellective[math]/judgmental[movie review]), advisor expertise (high/low), and interaction (face-to-face/through writing) on a decision-maker's (DM) acceptance of an advisor's advice. Advisor's confidence was highly correlated with DM's acceptance of advice, and DM's matched advisor's advice significantly more when the advisor had high expertise, especially for the intellective task. When interacting face-to-face, participants had higher group identity, and DM's had more trust and confidence in advisors and thought advisors were more honest and competent. DM's had a significantly harder time understanding advice when it was given through writing by a low-expertise advisor.

P1-14. Russell S. COOPER, & Susan LEVENE (Dept. of Psychology and Family Studies, United States International Univ.)

Cultural effects on confidence and information search in decision-making.

Past research demonstrated that consensus (or lack thereof) among information sources affects the information search process (Cooper and Sniezek, 1995) such that less information is sought and confidence is higher. The current study considers the effect of the cultural variable of individualism/collectivism (Hofstede, 1980). The prediction is one of an interaction between consensus and individualism and collectivism. Specifically, the normative influence of consensus will have greater effect on people who adhere to collectivistic values. The study was a quasi-experimental design with consensus manipulated and individualism/collectivism assessed for each participant. Results and implications are discussed.

P1-15. Seth HAMMER (Dept. of Accounting, Coll. of Business and Economics, Towson Univ.)

Ambiguity and safe harbors: Determining whether to claim a home office deduction.

An experiment was conducted which found that under high ambiguity tax reporting conditions, tax professionals perceive that clients would be decreasingly likely to claim a home office deduction as dollar levels of the expense increased, even where the expenditures are believed to have met the Internal Revenue Service's "realistic possibility of success" standard, which precludes the imposition of penalties against either taxpayers or preparers. The study provides findings consistent with Howard Raiffa's argument (1961) that under conditions of high ambiguity, decision making is based largely on individual temperament.

P1-16. Matthew H. OLSON, & Emily STARK (Dept. of Psychology, Hamline Univ.)

Personality factors and decision making.

80 undergraduate students completed the NEO-PI personality assessment, and scores on traits of neuroticism, extraversion, openness, agreeableness and conscientiousness were recorded. Two weeks later, all participants indicated their answers to decision scenarios involving risk aversion, risk seeking, likelihood estimation, susceptibility to the conjunction fallacy, and belief in "good luck." Results indicated that personality factors, particularly extraversion-introversion, may help to account for some but not all variability commonly observed in judgment and decision data.

P1-17. Michael SIEPMANN, Jonathan BARON, Karen STEINBERG, & John SABINI (Univ. of Pennsylvania)

Individual differences in attitudes toward one's own wishful thinking.

Subjects judged the probabilities of eight desirable and eight undesirable statements about their future, and then indicated the probabilities they would have assigned if they thought the way they would ideally like to think. They also provided reasons for discrepancies between initial and ideal probabilities, and indicated how strongly they wanted each statement to be true or false. Some subjects consistently felt they would ideally think more optimistically; some more pessimistically. We explored correlations with psychological well-being, and distinctions among subjects' ideals of accuracy, effective goal achievement, and psychological well-being.

P1-18. Karen STEINBERG (Univ. of Pennsylvania)

Regret-proneness and decision style: Evidence for individual differences.

These studies examined whether people's degree of regret over their decisions and difficulty making them varies dispositionally, not just situationally. Subjects completed the Decision-Style Questionnaire (DSQ)--a measure that operationalizes regret-proneness and decision style through questions about hypothetical decision scenarios--and answered DSQ-based questions about actual decisions to take a particular course and to attend a particular college. Regret-proneness and decision style were consistent over a 1-month period and across the assessed domains, suggesting that they may constitute a stable individual difference.

P1-19. David A. WASHBURN (Center of Excellence for Res. on Training, Morris Brown Coll., and Dept. of Psychology, Georgia State Univ.), Harold H. GREENE (Center of Excellence for Res. on Training, Morris Brown Coll.), & R. Thompson PUTNEY (Center of Excellence for Res. on Training, Morris Brown Coll., and Dept. of Psychology, Georgia State Univ.)

Individual differences in attention and shoot/don't-shoot judgment skills.

Students were tested with a battery of 18 tasks to determine individual differences across basic dimensions of attention, and on a series of shoot/don't-shoot scenarios in a firearms training simulator. The sensitivity of threat detection (d') was reliably predicted by measures of scanning (RSVP and visual search), decision time, and dual-task disruption. Response bias was significantly predicted by workload measures and the continuous performance task. Each regression model accounted for over half of the respective variance. Thus, accurate shoot/don't-shoot judgments were reliably associated with rapid scanning of visual attention, fast mental speed, and sustained attention.

P1-20. Rebecca J. WHITE, Nancy E. BRIGGS, & Ching-Fan SHEU (DePaul Univ.)

Gender effect in judging self-performance.

This study investigates how individuals evaluate their performance as well as that of others in a group. Participants individually worked on three crossword puzzles in a room with seven other people. The groups varied in gender composition. The puzzles varied in film categories. This task is chosen in contrast to previous research on self-evaluation of academic performance. Participants estimated how many items they completed, as well as the average number the group completed. Preliminary results show participants over-estimated their performance, males more so than females. We found no effect for gender on group composition or task content.

P1-21. Gwen GRAMS, Robert TRACY, Ching-Fan SHEU, & Fred HEILIZER (DePaul Univ.)

Self deception in perception of personal appearance.

This study examines three hypotheses: (1) that people create favorable self deceptions or illusions given motive and opportunity; (2) that this tendency is associated with healthy psychological functioning; and (3) that deception will decrease when the level of perceptive distortion required by the task exceeds a noticeable difference. To test these suppositions, each experimental participant was asked to rate a series of photographs that included an original, unaltered photo of the participant with six other photos that were altered to be either more or less attractive than the original. Results showed participants disproportionately rated both the altered and the unaltered photos as "less attractive" than themselves; and these differences were associated with higher self esteem and depression scores. Further, participants tended to give unaltered ("real self") photos higher attractiveness ratings (relative to the altered photos), than did independent observers. Yet, many participants attenuated use of this practice when the altered images were highly distorted.

P1-22. Janet A. SCHWARTZ (Dept. of Psychology, Rutgers Univ.), & William P. NEEDHAM (Dept. of Psychology, Purchase Coll., SUNY)

Hypothetico-deductive reasoning and formal operations: A reinvestigation of Wason's THOG task.

Recent research has suggested that performance on concrete versions of Wason's abstract classic THOG task might be mediated by cognitive development, as well as contextual presentation. The purpose of this study was to reinvestigate Wason's question with the classic THOG task and ask whether hypothetico-deductive reasoning ability is related to formal operations. Seventy-two undergraduates completed the classic THOG task, the Executioner problem (Needham & Amado, 1995), and six formal reasoning tasks. A significant correlation between formal operations score and performance on the THOG tasks was found, indicating that performance on the THOG task is mediated by both contextual presentation and cognitive development.

P1-23. Marcelle A. SIEGEL (Graduate Group in Science and Mathematics Education, Univ. of California at Berkeley)

Teaching high school students with Convince Me software: Decision as scientific theory building.

High school students in an issue-oriented science class (SEPUP) learned to make decisions using Convince Me (CM) software. CM possesses a connectionist model of explanatory coherence theory. Students entered "hypotheses" and "evidence" and linked these with positive or negative weights. CM assisted students in constructing a scientific argument while providing them with simulation-based feedback about the coherence of their decision. CM students significantly distinguished between hypotheses and evidence. They improved the structure of their arguments after receiving feedback from CM. Student work was also compared with a SEPUP control class which did not use CM.

P1-24. Julie GOLDBERG (Dept. of Psychology, Univ. of California, Berkeley), & Baruch FISCHHOFF (Dept. of Social and Decision Sciences, Carnegie Mellon Univ.)

Perceived and experienced risks and benefits of potentially addictive activities.

Young people face an array of potentially addictive activities. Their decisions regarding these activities may depend on their judgments of the probabilities and magnitudes of possible positive and negative consequences. This study examined whether misperceptions about the experience of addiction influence the decision to use drugs. Survey results indicate that underestimation of both the risk of addiction and the pleasure of using drugs is related to problematic drug-use. Similarly, beliefs about the consequences of engaging in potentially addictive activities were related to judgments about actual drug-use. These findings provide insight into the potential positive and negative effects of informational interventions.

P1-25. Christine M. CAFFRAY, & Sandra L. SCHNEIDER (Dept. of Psychology, Univ. of South Florida)

Enhancing positive and reducing negative affective states as motivators in adolescent risky behaviors.

Adolescents decide to participate in many risky behaviors that may have life-altering consequences. Adolescents' outcome expectations about the desired states that may be achieved or avoided by engaging in risky behaviors represent important precursors to these decisions. We found that a group of adolescents who chronically engaged in risky behaviors held stronger beliefs that the reduction of several negative affective states (e.g., depression, boredom) and the enhancement of positive affective states (e.g., good time) motivated their participation in risky behaviors. In contrast, the group who were less inclined to participate in risky behaviors were more focused on the negative consequences of those behaviors.

P1-26. Janis E. JACOBS (Pennsylvania State Univ.)

Developmental changes in the use of base rates and heuristics.

Earlier studies with children have shown that the use of the representativeness heuristic increases across middle childhood (e.g., Davidson, 1995; Jacobs & Potenza, 1991). The current study extended that work by manipulating the base rate ratios to determine the conditions under which children would use quantitative versus heuristic information. In a sample of 200 subjects of ages 4, 6, and 8, we found that judgments of the youngest children were not significantly affected by changes in ratio or instruction, but that those of the older children were significantly affected by both, so that they gave the most accurate judgments when they looked at disparate ratios and heard the base rate instructions.

P1-27. Ambrocio Mojardin HERALDEZ, Charles J. BRAINERD, & Valerie F. REYNA (Univ. of Arizona)

Children's spontaneous false memories.

Six-, nine-, and twelve-year-olds listened to a series of sentences and received immediate, one-week, and one-month delayed recognition tests. Test items included targets and three types of distractors with different degrees of semantic relationship to targets. Prior memory tests both preserved true memories (hits) and created false memories (false alarms) on later tests. False memories were more persistent over time than true memories. All effects increased with age. These results are interpreted in terms of fuzzy-trace theory's analysis of judgement processes in children eyewitness testimony.

P1-28. Lilian M. STEIN, & Valerie F. REYNA (Univ. of Arizona)

False memories and judgments in a juror-type situation: Fuzzy-trace theory analysis.

Theorists have made contradictory predictions about effects of repetition on memory for narratives. These contradictions are explained by fuzzy-trace theory. To test this explanation, a juror-type situation was created where participants were presented with either one, two or eight witnesses' versions of the same event. Two events were presented, counterbalancing order. As predicted, contradictory effects of repetition were obtained on recognition tests for decisions about verbatim statements versus implications made either consistently or inconsistently by different witnesses about the same event. Applications of fuzzy-trace theory to false memories and judgment processes will be discussed.

P1-29. Ronald L. WOODARD, & Valerie F. REYNA (Informatics and Decision Making Lab., Univ. of Arizona)

Memory-reasoning independence in covariation judgment: A fuzzy-trace theory analysis.

This study investigates the relation between memory and reasoning using a covariation estimation task of alpha-numeric stimuli, followed by a memory probe for the input information. Stimulus pairs are of targets and distractors that are consistent, ambiguous, or inconsistent with regard to the overall relation of the stimuli. Memory performance was such that the proportion of affirmative responses was greater for consistent pairs than for inconsistent pairs across targets and distractors (reconstruction from gist), and greater for targets than for distractors across consistent and inconsistent pairs (verbatim memory for the input information), supporting predictions of Fuzzy Trace Theory.

P1-30. Michael R. P. DOUGHERTY (Univ. of Oklahoma), Rickey P. THOMAS (Kansas State Univ.), Charles F. GETTYS, & Eve E. OGDEN (Univ. of Oklahoma)

The conjunction error as a memory retrieval phenomenon.

This research examined MINERVA-DM's account of the conjunction error. Participants were trained on the frequency of various traits in a population of fictitious animals. The probability of each trait characteristic was determined a priori. Participants rated the probability of each trait individually, e.g., P(A), P(B), and the conjunctions, e.g., p(A&B). Experiment 1 found that people commit the conjunction error when judgments are based on memory. Participants in experiment 2 made either frequency estimates (frequency format) or probability estimates (probability format). The frequency format decreased, but did not eliminate, the number of errors made, even when participants reported using their memory. Results of both experiments support MINERVA-DM.

P1-31. H. David SMITH (Middlebury Coll.), Mark F. STASSON (Virginia Commonwealth Univ.), & William G. HAWKES (Sch. of Medicine, Univ. of Maryland)

Diagnosticity and the dilution effect: Is more diagnostic information less prone to dilution?

The addition of nondiagnostic to diagnostic information yields less extreme judgments - a phenomenon known as the "dilution effect." The influence of highly vs. moderately diagnostic information as a possible moderator of this effect on judgments of student grade point averages (GPA) was investigated. Judgments of fictitious student profiles were significantly affected by the type of diagnostic information presented, and less extreme judgments of GPA were noted when nondiagnostic information was included. Findings were consistent with the dilution effect and generalized across both types of diagnostic information.

P1-32. Noel E. WILKIN (Univ. of Mississippi), & Glenn J. BROWNE (Texas Tech Univ.)

The influence of argument-based evidence on degree of belief.

Extending the theory that beliefs are constructed using arguments and judgments, this research evaluates the influence of argument type on subjects' degrees of belief. Subjects were presented with argument and/or information-based assertions with the intent of manipulating degrees of belief. Assertion type, order, and direction (for or against) were manipulated. Degrees of belief were evaluated using likelihood and support measures. It is hypothesized that degree of belief is (1) influenced more by causal arguments than authoritative arguments; (2) influenced more by argument-based evidence than simple information-based evidence; and (3) not affected by argument order and direction.

P1-33. Daniel G. GOLDSTEIN, & Gerd GIGERENZER (Max Planck Institute for Psychological Research, Munich, Germany)

Recognition: How to exploit your own lack of knowledge.

The recognition principle is a fundamental heuristic for inference. This heuristic advises considering only recognized alternatives when choosing among several. From this principle is deduced a counter-intuitive state of affairs wherein certain incomplete knowledge states allow one to make better inferences than more complete knowledge states. The conditions necessary for this less-is-more effect are stated. Through experiment and computer simulation, a less-is-more effect in a real-world environment is demonstrated.

P1-34. Laura MARTIGNON, Gerd GIGERENZER (Max Planck Institute for Psychological Research, Munich), & Kathryn LASKEY (Dept. of Systems Engineering, George Mason Univ.)

Evaluating fast and frugal choice heuristics.

Fast and frugal heuristics are evaluated against subtle and mighty Laplacean Demons. The question concerning the real nature of Laplacean Demons is debated (the metaphor used by Gigerenzer and Goldstein is a slightly modified version of Laplace’s omniscient creature: she does not have all information but operates optimally on the available information). Multiple Regression is only one of the candidates Demon may use. What characterizes Demon is the flexibility to use one or the other strategy according to each environment. Good candidates in Demon's toolkit are well specified Bayesian Networks and Classification Trees. But if an analysis of computational costs is performed by Demon, she may well end up choosing a simple and more frugal algorithm like Take The Best, whose accuracy does not fall too far behind that of mightier algorithms.

P1-35. Michael SCHMITT (Institute for Theoretical Computer Science, Technische Universitaet Graz, Austria), & Laura MARTIGNON (Max Planck Institute for Psychological Research, Munich)

Improving the performance of satisficing cognitive algorithms.

We investigate a family of cognitive algorithms that has been proposed recently by Gigerenzer and Goldstein (1996) to model a kind of human behavior - known as one-reason decision making - in the task of comparing two objects as to which scores higher on a given criterion based on binary cue information. How should the cues be ranked in order to achieve the largest number of correct decisions? We provide a theoretical framework for studying this question by analyzing the approximation capabilities of satisficing cognitive algorithms. We introduce an algorithm that has not been considered before and show that it can be used to improve the performance of any cue-based algorithm in many cases. We also exhibit a relation between the comparison task and a class of problems that is studied in the area of machine learning.

P1-36. Ryan O. MURPHY (Dept. of Psychology, Western Washington Univ.)

Stability of calibration biases over time.

Recent research in judgment and decision making (Yates, 1989) indicates that there are significant cross-national differences in levels of miscalibration biases (suboptimal resolution and overconfidence). However, investigations of miscalibration biases are based on an implicit assumption that has yet to be empirically validated. The researchers assumed that individuals have a relatively stable level of judgment biases on the same task over time. Ongoing research indicates that this previous assumption is valid. The distribution of miscalibration biases are also reported.

P1-37. Todd R. DAVIES (Koc Univ., Istanbul)

Effects of available time on confidence following choice.

Previous experiments indicated that subjects' generic confidence is reduced by the "half-range method" of probability assessment, which forces the subject to make a binary choice between propositions prior to expressing a probability. Further experiments by Davies (1995) indicated that this effect can be eliminated (and possibly reversed) when subjects reflect sufficiently on the binary choice and gain insight through this process. New studies in Turkey and the U.S. show that manipulating the time available for assessment is sufficient to induce reduction in confidence following choice. With enough time, the effect of choice on confidence depends systematically on propositional content.

P1-38. Jack SOLL (INSEAD, France), Joshua KLAYMAN (Graduate Sch. of Business, Univ. of Chicago), Claudia GONZALEZ-VALLEJO (Ohio Univ.), & Sema BARLAS (Direct Marketing Technology)

An unbiased test of the hard/easy effect.

The hard/easy effect is a well-known yet disputed result in probability judgment tasks: People are overconfident for hard items and underconfident for easy ones. Several authors have illustrated how the traditional method of dividing questions by difficulty leads to artifactual results. Random sampling methods are also problematic. We employ a "split-sample" technique that provides an unbiased test. The hard/easy effect disappears when comparing domains, but reappears when comparing people. Participants who are less accurate or more overconfident on one sample of questions tend to be more overconfident on another sample. We discuss several possible explanations.

P1-39. Gregory L. BRAKE, Michael E. DOHERTY (Dept. of Psychology, Bowling Green State Univ.), & Gernot D. KLEITER (Univ. of Salzburg)

A lens model approach to calibration.

The calibration task used in the present studies was designed with Brunswik's strictures concerning representative design in mind, and conceptualized within a Lens Model framework as well as a calibration framework. Twenty subjects who were knowledgeable about baseball predicted winners of rich but incomplete descriptions of 150 randomly sampled baseball games, making half-scale probability judgments that the predicted teams would win. In a replication, 20 additional subjects were run, ten making half-scale judgments and ten full-scale judgments. In both experiments, substantial underconfidence was found in the great majority of subjects. The relationship between calibration accuracy measures and the Lens Model indices is explored.

P1-40. Paul C. PRICE (Dept. of Psychology, California State Univ., Fresno)

Wishful thinking about sporting event outcomes is reduced by a relative-frequency elicitation question.

College sports fans judged the likelihood that one team would beat another in various football and basketball games. Some answered a probability question: "What is the probability that Team A will beat Team B?" Others answered a relative-frequency question: "Out of 100 games like this one in all important respects, how many times would Team A beat Team B?" The positive correlation between the judged likelihood that Team A would win and the stated desire that Team A would win (wishful thinking) was reduced or eliminated in the relative-frequency condition.

P1-41. Alan SANFEY, & Reid HASTIE (Univ. of Colorado at Boulder)

Judgment of events: Are we influenced by frequency or probability?

Many everyday judgments and decisions are based on an evaluation of previously acquired information. This study examined the influence of both frequency and probability of event occurrence on judgments of event likelihood. Participants saw a series of election poll results for various candidates. Each candidate's frequency of winning was varied independently of the their probability of winning, enabling a determination of whether win frequency or win probability influenced subjects' predictions of the winner of the election. Experiment 1 used a new context to replicate Estes' (1976) finding. Experiment 2 addressed some methodological limits of experiment 1.

P1-42. Peter MCGRAW, & Barbara MELLERS (The Ohio State University)

Anticipation of value and the endowment effect.

Good decision making requires the ability to make accurate predictions of value. This study investigates how well people predict the values they later assign to objects. First we ask people to predict the value they would attach to a coffee mug if they were given one. Then half receive a mug, and half do not receive a mug. For those given a mug, we assess the minimum selling price they would be willing to accept for the mug (WTA). Those without a mug are given a choice between a mug and a range of cash amounts. The minimum price they would accept in lieu of a mug is defined as their maximum buying price or their willingness to purchase a mug (WTP). Actual WTA value is greater than anticipated value, and actual WTP value is less than anticipated. Discrepancies were in a self-serving direction despite both groups’ awareness of their initial valuation.

P1-43. Pia WENNERHOLM, & Peter JUSLIN (Dept. of Psychology, Uppsala Univ., Sweden)

Base-rate inverse and base-rate neglect in categorization: A test of the elimination hypothesis.

A new explanation of the base-rate inverse effect (Medin & Edelson, 1988) and base-rate neglect (Gluck & Bower, 1988) in the categorization literature is tested. This explanation rests on two assumptions: (1) At the early stages of learning, a high-level reasoning process referred to as elimination operates, leading to base-rate inverse, (2) After further training, with more trials than in previous experiments, participants change into exemplar-based processing and learn to appreciate base rates. Two experiments, which partly replicate the studies by Medin and Edelson (1988) and Gluck and Bower (1988) reveal a pattern consistent with the elimination hypothesis.

P1-44. Ralph HERTWIG (Max Planck Institute for Human Development, Berlin, Germany), Peter SEDLMEIER (Univ. of Paderborn, Germany), & Gerd GIGERENZER (Max Planck Institute for Human Development, Berlin, Germany)

Judgments of letter frequencies: Are they systematically biased due to availability?

How do humans estimate whether a particular letter is more frequent in the first versus in a later position in written words? We tested four hypotheses, two of them precise versions of the "availability heuristic," a third that assumes that frequency processing occurs on the level of the phonological classes of vowels and consonants, and a fourth--the regressed-frequencies hypothesis--that assumes an (imperfect) monitoring of individual letters. The results are closest to the predictions of the regressed-frequencies hypothesis. They are inconsistent with Tversky and Kahneman's (1973) conclusion that judgments of letter frequencies are systematically biased due to availability.