P1-1:
Tracking Individual Differences at Different Stages of Decision
Making
Irwin P. Levin (Univ. of Iowa), Mary E. Huneke (Univ. of Iowa),
J.D. Jasper (Univ. of Toronto), Tara Burmeister (Univ. of Iowa),
Vy Nguyen (Univ. of Iowa), Joel Temperley (Univ. of Iowa)
Using a multi-staged decision making task and a new computerized information monitoring system, we were able to generate measures of information processing at both the consideration set formation stage and the final choice stage. These measures were compared across two experimental conditions, a condition where decision makers were instructed to incude items for later consideration and a condition where decision makers were instructed to exclude items from later consideration. Based on measures of effort, breadth and depth of search, decision makers high on the need-for- cognition (NFC) scale processed information more thoroughly than did those low on NFC.
P1-2:
The Effects of Consumption Delay on Risky Choice
S. Ramaswami, A.V. Muthukrishnan (Hong Kong University of
Science and Technology)
We show that when people are faced with a choice situation in which they have to choose in the present but will only consume the outcome in the future, the tendency to seek riskier options increases. Three possible mechanisms may underlie this phenomenon: the delay could reduce the difference between the perceived value of the two outcomes, or between the perceived riskiness between the two options, or it might simply make people more amenable to taking risks. Our experiments which include real gambles for money, as well as hypothetical choices of consumer products show some evidence for the third mechanism.
P1-3:
I Know What Is Risky, but It Doesn't Dictate What I Choose
Monica D. Barnes, Sandra L. Schneider (University of South
Florida)
Two studies of 54 multi-outcome lotteries were conducted to examine the relationship between riskiness ratings and risky choice. Rated riskiness was a predictable function of the expected value, variability, and skew. The results for choice were not as straightforward. When lottery outcomes were either all positive or all negative, participants tended to be risk seeking. When lottery outcomes were mixed, results were less clear. We suggest that the choice data can be explained by changes in approach and avoidance motivations along with related changes in contrasts between the sure thing and aspects of the lottery.
P1-4:
Occasion Setting in Causal Induction: The Importance of
Temporal Relations among the Candidate Causes
Michael E. Young, Janelle L. Johnson, Edward A. Wasserman
(University of Iowa)
The temporal relations among candidate causes were manipulated in a causal induction task. One of the causal events, the conditional cause, was immediately followed by the effect only on trials in which another causal event, the occasion setter, occurred before the conditional cause. In two experiments, human observers used the occasion setter to modulate their effect expectancy to the conditional cause. When the occasion setter was combined with candidate causes having various training histories, the transfer of occasion setting was different from that observed in a control group in which the candidate causes were presented simultaneously.
P1-5:
Causes of the Pruning Bias: Tests of Three Models
Clare Harries, Nigel Harvey (University College London)
Support theory, a focal hypothesis perspective and the ecological approach make different predictions about how pruning bias in familiar and unfamiliar scenarios should be affected by presentation order of different causal categories and by response format. Our data from an unfamiliar scenario were most consistent with the focal hypothesis perspective: there was an effect of presentation order and frequency responses reduced the likelihood given to the (unpacked) explicit description. For our familiar scenario, results were more consistent with Support Theory: there was no effect of presentation order and frequency responses increased the likelihood given to the (packed) implicit disjunction.
P1-6:
Adaptable Strategies for Testing Causal Hypotheses
Valerie M. Chase (Max Planck Institute for Human Development)
Lay hypothesis testing and causal attribution are often compared to models of scientific inference (e.g., Popperian falsification). However, what constitutes an appropriate strategy for seeking and integrating information relevant to a causal hypothesis ("C causes E") shifts with reasoners' goals, time limitations, and such ecological variables as base rates. It was predicted that under limited search, participants will prefer to examine instances of the rarer event to maximize Bayesian information gain. In tasks similar to the selection task, many participants indeed chose whichever event was rarer, while others consistently chose C over E. The rationality of these strategies is discussed.
P1-7:
Guessing Who You Are by Your Handwriting: Prototypicality
Helps Accuracy but Not Confidence
Christine M. Caffray, Sandra L. Schneider, Michele R. Devaux
(University of South Florida)
Two studies tested the theory of Probabilistic Mental Models (PMM; Gigerenzer, Hoffrage, & Kleinbolting, 1991). In Study 1, participants categorized handwriting samples according to the ease of identifying the writer's gender, and described the diagnostic cues used. In Study 2, another group was asked to guess the gender of each sample and provide a confidence rating. A week later, they rated each sample on a set of the most commonly cited cues from Study 1. The results showed that cue ratings (i.e., prototypicality) could predict the accuracy of judgments but, in contrast to PMM theory, could not predict confidence ratings.
P1-8:
Developing Accuracy in Probability Judgments: the Distinction
Between Calibration and Substantive Expertise
Ryan B. Opel, Eric R. Stone (Wake Forest University)
This study examined two distinct types of judgment expertise: calibration expertise and substantive expertise. Calibration expertise refers to one's ability to provide well-calibrated probability judgments, while substantive expertise refers to one's substantive ability in discriminating among possible events or outcomes in a particular domain. Using calibration and resolution as measures of calibration and substantive expertise, respectively, we tested whether the two types of expertise develop through independent processes by providing training in art history. The results suggest that the two types of expertise are distinct: calibration training affected calibration but not resolution, and substantive training affected resolution but not calibration.
P1-9:
A Multi-dimensional Scaling Analysis of Hard Decisions
Elizabeth S. Veinott and J. Frank Yates (Univ. of Michigan)
Two studies are reported with the goal of (1) understanding why people perceive a given decision as being hard or easy, and (2) identifying underlying dimensions of hard decisions. In Study 1, 140 participants rated 24 representative decision situations in terms of decision difficulty. In Study 2, 20 participants completed a card- sorting task with decision descriptions under two different instruction conditions and completed a feature-listing task. Our results suggest that some dimensions of decision difficulty may be less noticeable to the decision-maker than other dimensions. The implications of this research to training and aiding decision-makers are discussed.
P1-10:
Interpersonal Judgments within the Work Setting: Gender and
Public or Private Acknowledgments for Success
Cheryl Becker, George Smeaton (University of Wisconsin-Stout)
This study examined the influence of response type (modest or confident), compliment type (work or non-work), context (public or private), and gender on interpersonal judgments. Interpersonal Judgment Scale items were rated by 345 female and 311 male college students to assess a female or male actor within a written workplace scenario. Analysis indicated main effects for response type, participant gender, and compliment type. A two-way interaction for actor gender and context was also obtained. Post hoc comparisons show the female actor is rated as more likable as a work partner when complimented in a private context rather than among coworkers.
P1-11:
A Study on Physical Appearance and Level of Attraction to the
Opposite Sex
Sherynn J. Perry (Arizona State University)
This experiment examined the effects of physical appearance on level of attraction between people of the opposite sex. Level of attraction was tested using a questionnaire that included a picture of an attractive or an unattractive stimulus person. The participants consisted of 57 Research Methods (PSY 290) students. After viewing the stimulus person, they answered questions based on their attraction to and willingness to participate in activities with that person. The results of the Independent Samples Nonequivalent Control Groups Design found significance between level of attraction and physical appearance. In this study, the implications of attractive external characteristics are discussed.
P1-12:
Creating Uncertainty: Do Experimental Tasks "Say" What
they're Supposed to Say
Dean Yoshizumi, Irwin P. Levin (University of Iowa)
Introducing outcome uncertainty into psychological experiments often requires the use of a "random" generating process (e.g., lotteries, spinners). Experimenters select these tasks because they are easily understood by participants. Invoking an intuitive understanding of "randomness", however, has potential drawbacks. Evidence from subjective randomization experiments demonstrates that human perceptions of randomness are biased. These biases were expected to influence preferences for win/loss lotteries varying on potential outcome distributions and probability. Outcome distribution and winning/losing had large effects. Evenly distributed outcomes were selected most often on win trials and least selected in the loss trials.
P1-13:
Failure to Prime Configural Information Processing
William P. Neace, Stephen E. Edgell (University of Louisville)
An attempt was made to prime a configural strategy in a decision making task using a categorization task. Participants were primed by solving a deterministic categorization task that required either a configural or a dimensional solution. It was found that regardless of the priming task they did not differ in either their use of configural information or of dimensional information in a probabilistic decision making environment with both relevant configural and dimensional information.
P1-14:
Bias in Relative Frequency-Based Subjective Probabilities
Winston R. Sieck, J. Frank Yates (University of Michigan)
People are typically overconfident in general knowledge probability judgment tasks. Overconfidence has recently been generalized to subjective probabilities in a relative frequency learning environment (Yates et al, 1998). Associative learning theories suggest that probability matching (e.g. predicting event A 70% of the time when A obtains on 70% of the trials) occurs in these situations, and we show that biased judgments would arise from such matching. An experiment confirmed that probability matching is a primary driver of overconfidence in these tasks.
P1-15:
Perceived Expertise, Confidence, and Accuracy in Multiple-
Alternative Judgments
Douglas D. Mann, (Ohio University College of Osteopathic
Medicine), Bruce W. Carlson (Ohio University)
This study examined the relationship between unmanipulated perceived expertise and item confidences on a multiple-alternative medical diagnostic task. Fourth-year medical students (n=96) rated perceived diagnostic expertise relative to peers on a 0-100 percentile rank order scale and then chose diagnoses from extended lists for 200 brief case descriptions, with confidence in each diagnosis indicated on a 0-100% scale. Perceived expertise was correlated with mean item confidence (r = .54) and diagnostic accuracy (r = .42). Mean item confidence averaged 0.12 (12%) below diagnostic accuracy. The results extend previous studies of the relationship between perceived expertise and item judgments.
P1-16:
Combining Forecasts: Weighting of and Confidence in
Different Sources of Information
Nigel Harvey, Clare Harries (University College London)
People expressed their confidence in four forecasters' judgments and combined them. When forecasts were unbiased or biased in an atypical way, weights given to forecasters were quite appropriate and were highly correlated with corresponding confidence judgements. When forecasts were biased in a typical way, weights given to forecasters were less appropriate and were significantly less well correlated with corresponding confidence levels. Greater use of feedback when forecasts appear counterintuitive may provide an account of these findings.
P1-17:
The Influence of Task Instructions on Judgemental
Extrapolation
Alexander, B. Edmundson, M. O'Connor (University of New
South Wales, Sydney, Australia)
This paper examines the effect of imposed task structure on the accuracy of judgemental extrapolation (forecasting) of time series data. It compares the accuracy of three different treatments that varied the extrapolation technique. Subjects were randomly assigned to one of three groups - a) "smoothers", b) "forecasters" or c) "smoothers and forecasters." In the first group,smoothers, subjects were required judgementally to fit a linear "smooth" line to the cue time series data and to extrapolate a line into the forecast horizon. In the forecasters group subjects were merely asked to produce a judgemental forecast from the cue data.
P1-18:
Partitive Formulation of Information in Probabilistic
Problems: Behind Heuristics and Frequency Format
Explanations
Laura Macchi (Istituto di Psicologia, Universit di Milano)
Neither heuristic nor frequentist factors underlie the occurrence or elimination of the base-rate fallacy. What is crucial (given that all of the other elements of difficulty are the same, Macchi and Mosconi, 1998) is the absence (for having the bias) or presence (for avoiding the bias) of a particular formulation (partitive one) of the most specific information (likelihood). This partitive approach consistes of defining the set of which the datum (expressed in percentage or frequency terms) represents a part, and then relatising numerical content.
P1-19:
Now You See It, Now You Don't: The Effects of a Phantom
Decoy
Jonathan C. Pettibone, Douglas H. Wedell (University of South
Carolina)
The phantom decoy (Highhouse & Johnson, 1996) dominates one of two other alternatives in a trinary choice set but is an unavailable option. In three experiments, we explored the effect of this attractive but unavailable decoy in judgment and choice. In Experiment 1, phantom decoy effects were demonstrated in choice, with the dominated alternative selected more often. In Experiment 2, phantom decoy effects were not found in judgment. In Experiment 3, phantom decoy effects were tested in both judgment and choice and results were similar to the previous experiments. Overall, results suggest a similarity/substitution process in choice that does not occur in judgment.
P1-20:
A Connectionist Model of Asymmetric Dominance
Robert B. Branstrom (University of California, Berkeley)
Decision theory requires that choice alternatives be independent. In particular, the preference relationship between two alternatives should not change based on the presence of a third alternative. Yet a common finding is that a third, dominated alternative increases the probability of choosing the dominating alternative. This asymmetric dominance effect is replicated in a connectionist model, suggesting that the finding may be due to underlying cognitive mechanisms. In addition, the model makes two predictions about the strength of the effect based on the location of the third alternative relative to the other two alternatives. Empirical data support all three predictions.
P1-21:
The Role of Environmental Predictability in the Calibration of
Subjective Probabilities
Gregory L. Brake (Bowling Green State University)
This experiment utilized a Lens Model approach to investigate effects of environmental predictability on the calibration of subjective probability judgments. Ninety-five participants judged the likelihood of outcomes of 150 baseball games. The results show that when predictability of the environment is high, people tend to be underconfident, but as predictability decreases, calibration improves, then shifts to overconfidence. In addition, the level of domain knowledge was assessed independently of task performance. Analyses indicate that the predictability of the environment is more closely related to the direction and degree of miscalibration than the level of domain knowledge of the judge.
P1-22:
Overconfidence in Probabilistic Judgment: A Study in Saudi
Subjects
Ali S Alhakami
This study examines the appropriateness of confidence in probabilistic judgment. One hundred Saudi students answered 51 general knowledge questions and indicated how confident they were about their answers. The results yielded a calibration curve showing that subjects were overconfident in their judgments. Analysis of easy and difficult subsets of questions showed that overconfidence was most extreme with difficult questions. The results showed a strong tendency of the Saudi subjects to use 100% assessments inappropriately. This is similar to the findings of previous research in other Asian subjects. The study concludes with a cross-cultural discussion of the present findings.
P1-23:
Adaptation of Decision Behavior under Feedback or Memory
Load
Stuart M. Senter, Douglas H. Wedell (University of South
Carolina)
Previous research (Senter & Wedell, in press) established that constraining participants to view information in dimensionwise or alternativewise sequences does not always result in the use of different choice strategies. The present set of studies examined the effects of memory load and accuracy feedback under these two presentation sequences. In E1, half of the participants received accuracy feedback and half did not. In E2, participants were required to hold seven digits in memory while making their choices on half of the trials. Results are discussed in relation to the adaptive decision maker hypothesis (Payne, Bettman, and Johnson, 1993).
P1-24:
Experimental Power
Clark G. Ohnesorge (Carleton College)
Experimental power is an issue that has received an increasing amount of interest in the past few years. Power to find an effect is typically calculated under two conditions. Either prior to conducting a study (for the purpose of determining the number of subjects required), or after failing to obtain a statistically significant result in a study. However, an application of Bayes theorem demonstrates that it is also important to be aware of experimental power after obtaining a statistically significant result. This is because the level of power present can be critically important for interpreting the results of your study.
P1-25:
The Cochran Criterion: Discriminating Power as an Index of
Expertise
David J. Weiss (California State University, Los Angeles)
Whereas Einhorn (1972) argued that an expert must be reliable, we add an additional criterion. An expert must be able to discriminate among the stimuli within the domain. The single-subject experimental design calls for the judge to evaluate each stimulus object several times. The F-ratio for stimuli, tested against within- cells error, incorporates both discrimination, in its numerator, and reliability, in its denominator. We propose to use the F-ratio as an index of expertise. The use of this statistic to evaluate experts is adapted from Cochran's (1943) suggestion that the F-ratio be used to assess the quality of a response instrument.
P1-26:
Mental Accounting and Riding Loser Stocks
Marlys Gascho Lipe (University of Oklahoma)
Investors have a tendency to hold onto stocks which have declined in value. This 'disposition effect' could be explained by the strong distaste that people have for closing mental accounts at a loss. Thus, if an investment is held in a narrow mental account the investor will avoid closing at a loss by continuing to hold the stock. Here, mitigation of the disposition effect is attempted through (1) broadening the mental account ("New Car Investment Account" rather than "IBM Stock Account") and (2) helping investors mentally 'transfer their assets' by emptying one mental account into another rather than closing it.
P1-27:
Extension of Decision Field Theory to Multiple Choice Tasks
Robert M. Roe (Indiana University)
Decision field theory (Busemeyer & Townsend, 1992) is a stochastic (diffusion) model of choice that provides a description of the evolution of preferences over time within a single choice trial. Under special conditions, it provides a dynamic representation of the classic Thurstonian choice model. The importance of including dynamics is to provide a mechanism for explaining how choice probabilities change as a function of deliberation time. Earlier applications of decision field theory have been restricted to the binary choice case. The purpose of this paper is to present an extension of the theory to applications with mutliple (more than two) choice data.
P1-28:
Is Teaching Competence Related to Self- and Other-Insight?
James Hogge (Peabody College of Vanderbilt University)
This poster will examine the relationships among self-insight (ability to accurately describe one's own judgment policy), other- insight (ability to accurately describe others' judgment policies), and supervisors' ratings of the teaching competence of education students. It is hypothesized that students with high other-insight will receive higher ratings than students with low other-insight and that students with high self-insight will receive higher ratings than students with low self-insight. It is further hypothesized that the relationship between other-insight and teaching performance ratings will be stronger than the relationship between self-insight and teaching performance ratings.
P1-29:
On the Value of Human Life: Two Forms of Psychophysical
Numbing
David Fetherstonhaugh (Stanford University)
Previous research (Fetherstonhaugh et al., 1997) identified a form of "psychophysical numbing" in which the perceived value of saving a fixed number of lives decreases as the number of lives at risk ("reference group size") increases. Several new studies identify a second form of psychophysical numbing ("non-linear valuation of life") and demonstrate how both forms simultaneously and independently contribute to insensitivity to the value of human life. A final study assessed beliefs about how society "should" value the loss of human life, revealing striking and systematic disharmony between the beliefs people claimed to hold and the judgments they actually produced.
P1-30:
Do Utilities Capture the Emotional Responses to Outcomes?
Peter McGraw, Barbara Mellers (The Ohio State University)
Mellers et al. (1997) provide evidence that, unlike utilities, emotions vary with beliefs about likelihoods of monetary outcomes and are not monotonically related to monetary outcomes. We test decision affect theory in two non-gambling situations, grades in a psychology course and pregnancy tests, using laboratory experiments and real-world surveys. Emotional reactions to grades depend not only on grades received, but also on comparisons with expectations and beliefs about expectations. Emotional responses to the outcomes of pregnancy tests are determined by desired states and beliefs as well as test results. Both experiments provide evidence that emotional experiences differ greatly from utilities.
P1-31:
Fear Appeals: What Happens When the Credit Collector
Calls?
Bud Gibson, Mark Fichman (University of Michigan)
When fear appeals are used to gain compliance, the target is asked to do something for which he or she has disincentive in order to avoid a worse consequence. Much emphasis has been placed on how the emotional and information content of messages concerning these consequences might be manipulated in order to raise the rate of compliance. We examined 192 contacts between credit collectors and debtors to understand the effectiveness of fear appeals in a functioning organization. Surprisingly, messages concerning the consequences of non-compliance have inconsistent or insignificant effects.
P1-32:
A Bit of What You Fancy: Food Choice in the Face of Health
Risks
David Hardman (London Guildhall University)
A "teleshopping" task measured the degree to which information search varied in the face of presented health risk information. Experiment 1 varied the nature of arguments (eat/don't eat) appended to basic risk scenarios, as well as the information source. Experiment 2 examined individual differences in relation to the effect of seriousness of the risk and its identifiability. The examination of individual differences was extended in a questionnaire survey of the general population. We looked at the factors underlying propensity for information seeking, how information seeking related to risk perception, and whether risk perception was a major factor in food choice.
P1-33:
Direct Assessment and Use of Tolerance for Fuzziness in Risky
Gambles
Hsiaoping Yeh (Chang-Jung College, Taiwan), Fran?ois Sainfort
(University of Wisconsin-Madison)
In this paper the concept of the decision maker tolerance for fuzziness is defined. An assessment procedure is developed and a new decision algorithm to choose among risky options is proposed. The feasibility and validity of the new method is tested through an experiment based on Lichtenstein and Slovic (1971)'s study of preference reversals in gambling decisions. The results show that the proposed method leads to fewer preference reversals than traditional utility theory with respect to actual choices. It is concluded that the concept of fuzziness attitude can play a significant role in decision making between fuzzy alternatives.
P1-34:
Are Probability Judgments Responsive to Task Base Rates or
Ecological Base Rates?
P. Juslin, P. Wennerholm (Uppsala University)
One robust finding in research on judgment and decision making is base-rate neglect; participants' failure to use prior probabilities appropriately when assessing posterior probabilities (e.g., Kahneman, & Tversky, 1973). In a series of experiments, we investigated whether novices and experts either made use of (a) task base rates (fictional "summary statistics") or (b) ecological base rates (objective frequencies of base rates in the ecology) when assessing their posterior judgments. We discuss various factors that may lead people to integrate task- or ecological base rates in their posterior judgments.
P1-35:
Explaining the Inverse Base Rate Effect in Category Learning:
The Elimination Model
Pia Wennerholm, Anders Winman, Peter Juslin (Uppsala
University)
As an explanation to the inverse base rate effect (D. M. Medin, & S. M. Edelson, 1988), we propose a rule-based model (the elimination model) that attempts to capture the intervention of high-level reasoning processes at the initial stages of category learning. A key-assumption of the model is that participants, in states of partial learning, rely on a meta-cognitive mechanism, the elimination principle, to decide among candidate categories. This principle, that eliminates the execution of rules with high confidence, is tested in two experiments. Across all conditions in both experiments, the inconsistent base rate patterns observed are supported by the predictions of the elimination model.
P1-36:
A Memory Processes Account of the Calibration of Probability
Judgments
Michael R. P. Dougherty (University of Oklahoma)
The present research examines the applicability of MINERVA-DM (MDM) to study the calibration of probability judgments. In particular, two MDM mechanisms (experience and encoding) are predicted to decrease overconfidence. In Experiment 1, participants in the high experience condition showed almost zero overconfidence, whereas participants in the low experience condition showed a high degree of overconfidence. Experiment 2 replicated the effect of experience, and further revealed that overconfidence was reduced under conditions that facilitated encoding. The results of both experiments support MDM's predictions.
P1-37:
"Nothing Endures but Change": Why Base-rate Neglect Might
Not Be Irrational after All
Adam S. Goodie (University of Georgia), Peter M. Todd (Max
Planck Institute for Human Development )
Several studies have demonstrated base-rate neglect under ecologically valid conditions of direct experience. We propose that these effects may not be irrational when base rates may change frequently. A computer simulation compared twelve decisionmaking strategies in environments where base rates changed without notice, and more or less drastically. When base rates changed more frequently than cue accuracy, base-rate neglect was indistinguishable in performance from Bayesian integration. Connectionist model-based strategies were superior to event- tallying strategies, and probability maximizing outperformed probability matching. These results confirm the ecological rationality of base-rate neglect in natural environments where base rates vary more than cue validities.
P1-40:
NSF Funding Opportunities
Hal Arkes, Jeryl Mumpower (National Science Foundation)
The program officers for the Decision, Risk, and Management Science Program at the National Science Foundation will be present to provide information and answer questions about funding opportunities.
Under what circumstances do previous incidents, outcomes or choices become cognitively attached to current decisions? Many studies have demonstrated that sunk costs, prior winnings or losses, earlier decisions or unrelated chance events can sometimes become attached to a focal decision, but little is known of the factors making such attachments more or less likely. We report two studies exploring these factors in the context of risky choice. Findings suggest that subtle issues of labeling, temporal proximity and causal connection exert powerful effects on the attachment process and thus on the focal decision.
P2-2:
Graphical Presentation of Information and Perceived Risk
Reduction Size
Benita Bull (Wake Forest University), Eric Stone (Wake Forest
University), Winston Sieck (University of Michigan)
Previous research has shown that people are willing to spend more money to reduce a risk when risk information is presented graphically rather than numerically. The present work extends this research by including a graphical condition (pie charts) that highlights the denominator of the risk information, as well as examining whether the effect is mediated by a perceived difference in the magnitude of the risk reduction. As predicted, participants perceived the reduction in risk to be smaller with the pie charts display than with other formats, and this translated (though non- significantly) into willingness to pay less for the risk reduction.
P2-3:
Domain and Probability/Expected Value in Risky Choice
Framing Effects
Francis A. Cleland, Sandra L. Schneider (University of South
Florida)
We hypothesized that the size of framing effects in "Asian disease" type scenarios would be affected by changes in the domain of the problem (e.g., lives versus jobs) and by changes in the probability and expected values of the options. As we anticipated, more risk seeking behavior in both frames was observed in situations involving potential loss of life versus other, less serious situations. We also found differential responses as a function of expected value/probability, but they did not conform to any current theory of risky choice. We explore both attentional and motivational processes as possible sources of variations in response.
P2-4:
Status Quo and Personal Responsibility in Affective Responses
to Good and Bad Outcomes
Don T. Ritter, Sandra L. Schneider (University of South Florida)
Instead of hypothetical scenarios, this investigation of regret and satisfaction examined participants' reactions to their own performance on one of two verbal tasks. Independent variables included whether participants (a) performed better or worse than they expected, (b) completed the originally assigned task or switched to the other task, and believed that they had control over (i.e., personal responsibility for) which task was completed. A change in status quo (i.e., switching tasks) rather than perceived responsibility led to exaggerated regret. However, this increased regret did not always coincide with differences in the overall affective reaction to the outcome.
P2-5:
Examining the Impact of Cohesiveness on Group Accuracy
Interventions
Rebecca A. Henry, Erica Israelson, Jill Kmet, Alyson Landa
(Purdue University)
This study examines the impact of group cohesiveness on the relative effectiveness of two group accuracy interventions. Both interventions have been successful in improving group judgment accuracy by providing minimal strategic guidance, specifically: a) sharing information, or b) explicitly trying to identify the most accurate member (Henry, 1995). What has not been investigated is whether the cohesiveness of a group influences how members respond to these distinct interventions. Results support the hypothesized interactive effect of cohesiveness and intervention strategy on group judgment accuracy. The cues individuals reported relying on also varied by condition, providing insight into the link between interaction cues and accuracy.
P2-6:
Modeling Group Judgment: The Impact of Discussion on
Member Opinion and Final Group Judgment
Oleksandr S. Chernyshenko, Andy Miner, Michael Baumann, Janet
Sniezek (University of Illinois at Urbana-Champaign)
This paper proposes the Differential Cue Weighting Model (DCW) to explain how groups favor information held in common by all members over that held uniquely by each. The DCW model extends previous research to consideration of initial cue weights, initial distribution of information, and changes in weighting of information during group discussion. In this research, the initial distribution of shared and unshared information was manipulated, and initial cue weights were estimated (by participants who did not engage in group discussion.) Results generally supported the model. Cue weights changed depending on whether information was common or unique, and discussed or not discussed.
P2-7:
What Causes the Failure to Uncover Hidden Profiles?
Lyn M. Van Swol (Univ of Illinois), Yuhong Wang (Univ of
Illinois), Lucia Savadori (University of Padua),
Janet A. Sniezek (Univ of Illinois)
Individual decisions were used to simulate effect of repetition of common information in group discussion on uncovering hidden profiles. Two factors were manipulated 1.) whether or not common information was repeated and 2.) whether participants made 2 decisions (first with partial information and second with full information) or made 1 decision with all information. Most participants who made 2 decisions without the repetition of common information uncovered the hidden profile in their second decision. However, when the "common" information was repeated, participants seldom uncovered the hidden profile, despite having access to all the information.
P2-8:
Suspicious Minds
Janet A. Sniezek (University of Illinois at Urbana-Champaign),
Chip Heath (Duke University), Lyn M. Van Swol (University of
Illinois at Urbana-Champaign), Michal Nochimowski (University
of Illinois at Urbana-Champaign)
When people suspect their advisors' motives, they gain immunity to persuasion attempts. Two studies examined how a decision- maker (DM) used advice from two advisors: the accurate advisor was instructed to be accurate, and the persuasive advisor was instructed to persuade the DM to accept the advisor's initial answers on the task. In study one, DMs accepted the advice of the persuasive advisor more and rated them as more knowledgeable and accurate. In study two, when DMs were warned that some advisors might be motivated by persuasion, their bias toward the persuasive advisor was eliminated.
P2-9:
Policy Persuasion: The Impact of Source Credibility and
Recipient Reactance on Judgment Policy Change When the
Source Is a Computer
Elise Axelrad Weaver, PR Costanzo (Duke University)
A study was conducted to ascertain the impact of the traditional persuasion variables of source credibility and recipient reactance on policy change in a multi-attribute judgment task. The study introduces a dependent measure of policy persuasion and compares the extent of policy persuasion when the policy advice is attributed to a human expert or to a non-human expert system. The roles of confidence and a set of individual difference variables (aspects of need for closure, reactance and impression management) in the policy persuasion process are explored in this context.
P2-10:
Comparing Group and Individual Decisions in a Condition of
Positive Mood: Grouping Induces Optimism
Michael D. Bussman, Verlin B. Hinsz (North Dakota State
University)
We examined the influence of positive mood on decisions made by groups and individuals for oneself or others. Participants were assessed on four hypothetical events involving risk. No significant mood effects were found, nor were there significant differences in decisions for oneself or others. We found that group discussion induced more risky shift compared to the shift for individuals. Moreover, groups rated positive events as being more likely than did individuals. Thus, groups were more optimistic about the likelihood of positive events and, as a result, made riskier decisions. We discuss these results with regard to context influences on decisions.
P2-11:
Repeated Ultimatum Bargaining with a Twist: The Effect of
Uncertainty and Verifiability of Pie Size and Outside Options
on Honesty and Acceptance Rates
Terry L. Boles (University of Iowa), Michelle Buck (McGill
University), Rachel Croson (University of Pennsylvania), David
Messick (Northwestern University), Keith Murnighan
(Northwestern University)
Dyads, who were anonymous to one another, played a series of 4 ultimatum bargaining games for cash using Group Systems decision software. We varied whether the pie size and outside option were certain or uncertain (to responder and proposer respectively), and whether these amounts were verified after 2 bargaining trials or not. Players were able to send messages to one another about pie size, outside options, or reasons for offers or rejections. We found the highest rejection rates in the round following a round where the pie size was uncertain, but later verified, and the responder learned the proposer had lied.
P2-12:
The Effect of Gender Stereotypes on Sunk-Cost Decisions
Elmer Anita Thames (John Carroll University)
The sunk-cost effect is the tendency to invest additional time, money, or effort in an endeavor once an irrevocable investment, or sunk cost, has been made. The present study was designed to test the influence of gender stereotypes on reinvestment decisions involving sunk costs. Participants imagined planning to attend a ballet or football game and mistakenly assuming that the $25 cost included transportation. However, to get to the event, they would have to pay an additional cost of $15 for transportation. Half imagined they already purchased the ticket while half did not.
P2-13:
Worldviews and Environmental Judgments: Dumping on the
Rich
David H. Ebenbach, Colleen F. Moore (University of Wisconsin-
Madison)
56 participants made multiple decisions siting either a chemical waste dump or an environmental improvement project (EIP) in one of two neighborhoods. In the descriptions of the neighborhoods, we manipulated their poverty relative to one another, the relative impact of the project, and their probable reactions to the siting. All manipulated variables influenced judgments, including a "Robin Hood" effect where participants preferred siting the dump in richer neighborhoods and the EIP in poorer neighborhoods. We will also report the relationship of these judgments to hierarchical, egalitarian, and other worldviews.
P2-14:
Judgment of Risk to Self vs. the Environment: Gender
Differences and Environmental Attitudes
Scott B. Caldwell, Colleen F. Moore, David H. Ebenbach
(University of Wisconsin - Madison)
The possible relationships between environmental attitudes and risk perceptions were investigated. Participants (89) rated 24 environmental hazards along dimensions of risk to Self and Environment. Participant risk ratings were then related to their scores on the Environmental Attitudes Scale (EAS). Results supported the hypothesis that risks to the natural environment are particularly salient for those who have strong pro-environmental beliefs. Furthermore, a gender effect was found. Men rated risks to Self and Environment as significantly lower and hazard controllability as significantly higher than women did. Regression analyses suggest environmental attitudes play a key role in these gender differences.
P2-15:
Women, Decision Content, and Other Dangerous Things
Ann-Rene Blais, Elke Weber (The Ohio State University)
One hundred thirty-one French-Canadian and ninety American undergraduate students completed questionnaires about decision making strategies and risk preferences in different decision situations that reflected five domains of life. Individual differences on Epstein's rational-experiential processing scales and on Schwartz' Survey of Values were found to interact with decision style and decision content. We also found surprising gender-by- style-by-content interactions, with women being more likely to use rational decision strategies than men, especially in some content domains. A similar three-way interaction was found for the two nationalities.
P2-16:
Information Search Strategies of Novice and Expert Financial
Forecasters
Fergus Bolger (Erasmus University), Gulnur Muradoglu (Bilkent
University, Tuyrkey)
Expert financial forecasters and management undergraduates were presented with several stocks and invited to invest in one of them. Before making a selection, potentially relevant information regarding each stock could be viewed one-at-a-time by clicking on boxes on the computer screen. The selections made, and the time spent considering each piece of information, were logged. After an investment had been made feedback about the performance of the chosen stock was given then a new set of stocks presented. This procedure was repeated over several trials. The effects of manipulations of information quality and cost on information search are discussed.
P2-17:
The Concept of Indecisiveness in the Accuracy-Informativeness Tradeoff
Terry Rowlands (University of New South Wales)
Decision makers frequently tradeoff the accuracy and informativeness of information about uncertain events, either as consumers or suppliers of information. A narrow estimate of uncertainty may prove more informative, but a wider interval is more likely to contain the actual value (thus be more accurate). This study provides an extension of the Yaviv and Foster (1995) additive model of the accuracy-informativeness tradeoff by utilising a vector (rather than scalar) model to reveal indecisiveness, which is the difficulty in choosing between two intervals.
P2-18:
Indexes of Probability-Judgment Accuracy as Measures of
Metacognition
Ju-Whei Lee (Chung Yuan University)
The general-knowledge questions have been frequently adopted to measure participants' accuracy of probability judgments. The present research shows that the calibration index and the bias measures typically examined in probability judgments can be borrowed to reflect memory monitoring effectiveness. Consequently, the almanac-question task can serve as a metacognitive judgment task.
P2-19:
Confidence in Clinical Judgment: The Role of Mere
Experience
Paul C. Price, Jenifer Tymn , Christine B. Edmondson (California
State University, Fresno)
Psychology undergraduates in clinical and abnormal psychology classes categorized nine fictional psychiatric patients in terms of their suicide risk without feedback and judged the probability that they correctly categorized each patient. They performed this task once at the beginning of the semester and again at the end. There was very little evidence that participants' confidence increased as a function of mere experience with the task either across patients or across sessions. This suggests that the increase in confidence as a function of experience observed in previous studies (e.g., Oskamp, 1965) depends primarily on the judge's receiving additional information about each patient.
P2-20:
Doctors' Judgements of the Likelihood of Early Mortality
Following Paediatric Cardiac Surgery
Tim Rakow, Nigel Harvey (University College London)
Total cavopulmonary connection (TCPC) is an operation for repair of univentricular hearts, with a reported in-hospital mortality of 12 to 15%. For the 'real life' clinical question: "What do you think is the probability of this patient dying after a TCPC?", this study aims to identify the most important influences on judgement, and compare the quality of professional and statistical prediction. The methods employed include interviews, pencil-and-paper tasks involving hypothetical patients, and an experimental study using data from past patients with known surgical outcomes.
P2-21:
Allocation of Medical Resources: The Influence of Frame of
Reference, Just World Beliefs and Responsibility for One's
Condition
H. David Smith, Elizabeth Burns (Middlebury College)
The framing effect was studied using a modified version of the "Asian Disease Problem". Subjects chose safe or risky treatment options and rated each program on several continuous dimensions. In addition to the manipulation of frame, subjects read scenarios describing cancer victims who had either contracted the disease through cigarette smoking (high responsibility for condition) or by unknowing exposure to carcinogens (low responsibility for condition). Subjects also completed the "just world scale." Frame had a marginally significant effect on choice. Analysis of continuous ratings yielded a more complicated picture. Also, the "just world scale" significantly predicted subjects' treatment choices.
P2-22:
Legal Decision Making The Fast And Frugal Way
Mandeep K Dhami (Max Planck Institute for Human
Development), Peter Ayton (City University)
Fast and frugal models are proposed as psychologically plausible descriptions of judgement processes (Gigerenzer & Goldstein, 1996). In the present study, 342 bail decisions made by 52 benches of magistrates in two English courts were observed. The benches' policy was modelled using "Take The Best", "Minimalist" and "Take The Last". The relative descriptive validities of these models were measured on a set of 60 holdout cases. "Take The Best" correctly predicted more decisions. Thus, legal decision makers appear to use limited search procedures and one reason decision making, and so are processing information in a non- additive, non-linear and non-compensatory fashion.
P2-23:
Attention Profiles for Speeded and Unspeeded Decision
Making
David A. Washburn, R. Thompson Putney (Georgia State
University), William Tirre (Armstrong Laboratory, Brooks Air
Force Base)
Attention is a multi-dimensional construct, and individuals differ in these component skills or factors. We tested 176 Air Force recruits on a battery of attention tasks and on a flight simulator that required concurrent-task performance. Participants were grouped into the best and worst performers on the flight simulator according to primary-task measures (maintaining pitch, roll, yaw) and speeded secondary-task judgments (detecting oil temperature and pressure warnings). Attention profiles were computed and compared across groups. Attention focusing is important both to flight-orientation and oil-response performance. However, these latter, speeded-decision tasks also benefitted from attention scanning and filtering skills.
P2-24:
Personality and Medical Decision Making
Susan Longley, Irwin Levin, Susan Shaffer, Kirsten Redalen
(University of Iowa)
The influence that indvidual difference variables have on risky judgment and decision making is increasingly being recognized. Superordinate Big 5 personality traits have been associated with differential responding in a numberr of realms. The Big 5 heirarchy has been recognized as providing a more parsimonious explanation for the relations between different personality types and health. In this study classic decision making stimuli were used to investigate the association between these broad personality traits and biases in health-related judgments. Findings pinpoint which personality traits accounted for significant variance on specific judgments in the domain of health.
P2-25:
Medical Treatment Decisions: Individual Differences in Risk
Tradeoffs
Andrea L. Washburne, Sandra L. Schneider, Teresa Broughton
(University of South Florida)
This presentation examines individual differences in the tradeoffs people make among treatment options involving varied short term life expectancy, quality of life, and long term life expectancy. Although previous studies have demonstrated conflicting results regarding preferences for risky medical treatment options, few have examined individual differences in the tendency to choose one type of risk over another. We will explore potential differences including previous direct or indirect experience with life-threatening medical conditions, as well as demographic characteristics such as age and gender. In addition, we will examine differences in the strength of preference for particular types of risk.
P2-26:
Adaptors and Innovators in a Novel Dynamic Problem Solving
Task
Julia Pounds (Federal Aviation Administration, Civil Aeromedical
Institute)
Human interaction with computers is a vital component of air traffic systems. However, the role of individual differences in cognitive style--one's preferred approach for problem solving--in air traffic control tasks has not received much attention. This study examined whether cognitive style would influence performance outcomes in dynamic tasks resembling air traffic situations. Performance across several learning stages was measured. Adaptors generally outperformed Innovators over all tasks, although Innovators demonstrated an initial, though not statistically significant, advantage.
P2-27:
Modeling Symmetric and Asymmetric Vague-Vague Cases of
Ellsberg's Paradox
Karen M. Kramer, David V. Budescu (University of Illinois at
Urbana-Champaign)
Ellsberg's paradox has been shown to be prevalent with symmetric vague-vague cases, where neither probability is specified precisely around a common midpoint (Kramer & Budescu, 1996). Most conclusions from studies on precise-vague cases were extended to vague-vague cases. We have delineated models which capture the relationship between subjects' choices and characteristics of symmetric gambles. We now explore the extent to which the paradox occurs in asymmetric vague-vague gambles, where probability ranges are not centered around the same midpoint. Range widths and midpoints jointly play a role in the occurrence of the paradox. Models which generalize to the new asymmetric cases are examined.
P2-28:
Non-Expert Decision Makers' Reactions to Point versus Range
Estimates of Probability
Paul Zarnoth, Janice Nadler, Adrian K. Rantilla, & James H.
Davis (University of Illinois at Champaign-Urbana)
A study explored the use of statistical information by non-expert individuals and groups making decisions about environmental hazards. Participants received information from opposing sides of an issue, including risk assessments that were presented as either point or range estimates. Point estimates were rated as more informative and more useful than range estimates. Sources who provided information in the form of point estimates were perceived as more confident. Nevertheless, the range estimates were more influential in determining personal risk judgments, and participants were more confident in their own range estimates than in their own point estimates.
P2-29:
Illusion of Control and College Students' Perceptions of Luck
in a Hypothetical "Powerball" Lottery Game
Helen Swanson, James Byrd , Cheryl Becker (University of
Wisconsin-Stout)
536 college students rated their luck after reading a scenario about a entry. A 4 x 3 x 2 x 2 between-subjects design included: medium by which winning number is learned (television; newspaper; clerk; internet); timing of information receipt (this evening, or next morning for newspaper; next evening; three evenings later); number source (self-selected or computer- generated); and gender. Lottery behaviors were covariates. Unexpected findings included: timing x gender interaction; number source and gender main effects (on-line only players); number source x timing interaction (scratch-off only players). Results extend illusion of control literature.
P2-30:
Development of a Risk Ranking Methodology
Kara M. Morgan, Michael L. DeKay, Paul S. Fischbeck, M.
Granger Morgan, Baruch Fischhoff (Carnegie Mellon University)
A ranking procedure is under development that strives to improve the quality of the output by increasing the lay participants' understanding of their own risk judgments. The method aims to illuminate judgment biases by using multiple judgment methods, and to diminish noise by using an iterative decision process. Clearly presented, easy to compare risk information and a multi- attribute judgment model are the center of the method. The objectives of a good method will be discussed, and results indicating how this method can produce a reliable and robust risk ranking will be presented.
P2-31:
Content and Discontent: How Cover Stories Influence Decision
Making
David A. Rettinger, Reid Hastie, Walter Kintsch (University of
Colorado)
How do the contents or subject matter of a decision problem affect the outcome of the decision? Although decision making research typically dismisses content as merely "cover story," this research indicates that it plays a fundamental role in the decision process. Experimental results demonstrated that the same decision structure in different domains engendered different mental representations, decision strategies and preferences. A recent model of comprehension was applied to simulate the differences in problem representation that we hypothesized were the causes of the differences in behavioral outcomes. We conclude that "cover stories" affect decisions by changing the mental processing underlying decision making.
P2-32:
Accidents as Probabilistic Penalties: The Effect on Learning
Rachel Barkan (Indiana University)
Barkan, Zohar and Erev (1998) suggested to model accident prevention tasks utilizing signal detection theory with an adjustment of probabilistic penalty for Miss error. They showed that a decrease in the probability for penalty resulted with an increase in risk-taking and impaired the learning process. The present work studies further the effect of probabilistic penalty on learning in accident prevention tasks. Three experiments show that different kinds of non-payoff feedback for the non-penalized error (i.e. Near-Accident), can modify the effect of the probabilistic penalty and restore learning towards safer behavior. We discuss the theoretical and practical implications for accident prevention.
P2-33:
Quick Estimation: Letting the Environment Do Some of the
Work
Ralph Hertwig (Max Planck Institute for Human Development)
How do we estimate real-world quantities, such as the number of murders committed in Chicago each year? We propose a simple heuristic for estimation, QuickEst, which exploits a common structure in real-world environments (the J-shaped distribution) to make quantitative estimates using few cues. Our computer simulations demonstrate that where knowledge is scarce - as is typically the case in naturalistic settings - the fast and frugal QuickEst performs as well or better than more complex methods such as multiple regression and estimation trees. QuickEst is an ecologically rational strategy whose success highlights how environmental structure can be used to simplify inference.
P2-34:
The Impact of Presentation Media on Forecasting and Decision
Making
Marcus O'Connor (University of New South Wales), Kai Lim,
William Remus (University of Hawaii)
This paper reports an experiment on the impact of multimedia on the feedback of information in forecasting and decision making tasks. The study was motivated by media richness and complementary cue theories that suggest multimedia would have impact on cognitive feedback. No evidence was found to support these theories. Support was found, however, for predictions based on the theory of Stone and Kadous (1997) which relates the impact of "affect" on performance. Specifically, multimedia was differentially most beneficial when negative affect cognitive feedback was provided. These results did not occur when positive affect feedback was provided.
P2-35:
Measuring Preferences and Positioning Really-New Products
Steve Hoeffler (Duke University)
The ultimate goal of this research is to improve preference measurement practices for really new products (RNPs). The first task is to understand what it means for a product to be "really new" for consumers. The main proposition put forth in Study 1 is that newness stems from the consumer's difficulty or uncertainty in estimating the future utility (benefits, drawbacks, and social implications) of a new product. In Study 1 information was used from the actual patents for products that are currently under development to successfully support the construct of uncertainty in predicting future utility as a formative indicator of newness.
P2-36:
The Effect of the Global Vs. Analytic Display of Prices of a
Bundle Offer on the Bundle Purchase Likelihood and its Cost-
level Estimate
Nicolao Bonini (Universit. di Cagliari, Italy), Rino Rumiati
(Generale Universit. di Padova, Italy)
We examine whether the global (only one all-inclusive price is presented) vs. analytic (several prices are presented) display of prices of a bundle offer affects the bundle purchase likelihood and its cost-level estimate. Results of four experiments show that the bundle purchase likelihood is lower in the analytic format (where an out-of-market price is displayed) than in the global format. Also, they show that the bundle cost level estimate is higher in the analytic format than in the global format. Findings are explained by a revised version of the transaction utility theory (Thaler, 1985).
P2-37:
Attribute Ambiguity and the Attraction Effect
Janet A. Schwartz, Gretchen B. Chapman (Rutgers University)
The purpose of this study was to determine whether the meaningfulness of stimuli attributes was related to the presence of the Attraction Effect (Huber, Payne, & Puto, 1982) when physical structure of the task is held constant. 102 participants each reviewed six consumer choice scenarios where the alternative's attributes had been matched on importance but varied on ambiguity. The results showed that the attraction effect was not more likely to occur with ambiguous attributes than non ambiguous attributes. Thus, the role of attribute ambiguity in the attraction effect is less clear cut than has been previously suggested (e.g., Ratneshwar, Shocker, & Stewart, 1987).
P2-38:
Contingency Planning for the Year 2000
Laurie Broedling (LB Organizational Consulting)
The Year 2000 (Y2K) computer glitch will have a significant worldwide impact. While many of these computer problems will be fixed before January 1, 2000, many will not. Those that are not fixed will have a significant impact on virtually everyone around the globe although the specific impacts will vary by individual, community, region and organization. This paper describes the need for increased awareness of the impending impact to society and the need for contingency planning. It analyzes causes for the current state of low awareness and lack of contingency planning. It suggests that this event offers a unique opportunity for scholars to study judgment and decision-making.
P2-39:
Modes of Adapting in Time-Pressured Decision Making
John Maule, Isabel Andrade (University of Leeds)
A field study is presented investigating whether the cognitive strategies for adapting to time pressure identified by laboratory research adequately describes the full range of strategies used by managers in their 'everyday' decision making. Thirty-four modes of coping with time pressure were derived from the literature and given to a sample of 100 managers. Factor analysis revealed five strategies, including some identified by laboratory studies (e.g. acceleration) and some that are not (e.g. involving others in the process). The findings are discussed in the context of contemporary theories of time pressure and used to identify priorities for future laboratory studies.
P2-40:
How Task Characteristics Affect Cognitive Mode
Phil Dunwoody, Eric Haarbauer, Robert P. Mahan, Christopher
Marino, Chu-Chun Tang (University of Georgia)
The present study had two main goals: (a) to construct a Task Continuum Index (TCI) and a Cognitive Continuum Index (CCI) and (b) to test whether task characteristics will influence a judge's cognitive activity as predicted by Hammond's cognitive continuum theory (CCT). Task characteristics had a significant affect on cognitive mode as operationally defined in CCT. However, CCT did not accurately predict the observed pattern of results of task structure on cognitive profiles. A possible explanation for this failure is that CCT, in its current form, does not adequately incorporate changes in task perception as a function of expertise.