Society for Judgment and Decision Making 1997 Annual Meeting

P2-30. Eric. R. STONE, & Carolyn J. RUSH (Wake Forest Univ.)

Risk communication: The effectiveness of graphical modes depends on the risk magnitude.

Previous research we have conducted comparing graphical presentational formats to numerical formats found that, for low-probability risk magnitudes, graphical formats induced greater professed risk-avoidant behavior than did numerical formats. In two studies, the present research found that this effect did not generalize to situations with higher risk magnitudes. These results suggest that framing effects found in the risk communication literature may be dependent on the risk level employed, and in particular may be most common with low-probability scenarios, perhaps due to people's unfamiliarity with such low probabilities.

P2-31. Stuart M. SENTER, & Douglas H. WEDELL (Dept. of Psychology, Univ. of South Carolina)

Information presentation constraints and judgment accuracy.

Participants judged the attractiveness of apartments under constraints which forced them to view information either by alternative or by dimension. Results from previous research on choice under presentation constraints has shown that dimensionwise constraints result in more accurate choices and less effort expenditure than alternativewise constraints. It was found that the dimensionwise constraint resulted in more accurate judgments within a choice set, while the alternativewise constraint resulted in more accurate judgments between sets. Results are discussed in relation to process tracing methodologies and the findings of previous research utilizing presentation constraints.

P2-32. Jonathan C. PETTIBONE, & Douglas H. WEDELL (Dept. of Psychology, Univ. of South Carolina)

Contextual sensitivity of ideal point preferences.

Contextual sensitivity of ideal point preferences has been established for psychophysical stimuli (Riskey, Parducci, & Beauchamp, 1979) such as sweetness of a beverage but not for social judgments such as attractiveness. In three experiments, we explore this issue using computer generated faces that varied in nose width and eye gap. In Experiment 1, participants made descriptive and attractiveness judgments of faces that varied on one feature. In Experiment 2, both features were manipulated together. In Experiment 3, participants chose which of three faces was most attractive. Results from all three experiments support the contextual sensitivity of ideals for attractiveness.

P2-33. Winston R. SIECK (Univ. of Michigan)

Range and frequency effects on probability judgment.

Two experiments examined the effects of the range and relative-frequency of objective probabilities (OP) in the stimulus set on subjective probabilites (SP). Experiment 1 stimulus sets had either relatively more or fewer mid-range than extreme OPs (i.e., hard vs. easy environments). Experiment 2 stimulus sets consisted of either a narrow or wide range of OPs corresponding to hard and easy environments. The hard-easy effect was essentially replicated. Results were consistent with range-frequency theory which suggests that the hard-easy effect reflects a tendency to assign the same number of presented OPs to each of the SP categories available.

P2-34. Bridget C. FLANNERY, Stacey A. NEFF, Jeremy D. JOKINEN, & Bruce W. CARLSON (Ohio Univ.)

Judgmental forecasting when changes occur in a time series.

In this study, we investigated how people respond to changes in the direction of a time series when making forecasts. We found that people produce forecasts that are too close to the most recent value of a time series, a result that is consistent with previous research and that has been described as an anchoring effect. We also found that this conservatism decreases as people gain experience with a time series. Finally, we found little evidence that the time to make a forecast is related to changes in a time series. The implications of these results are discussed.

P2-35. Ayse ONCULER (Dept. of Operations & Information Management, Wharton Sch., Univ. of Pennsylvania)

Modeling intertemportal choice under uncertainty.

The purpose of this experimental study is to examine if future uncertainty is treated differently than immediate uncertainty and future certainty. The findings suggest risk preferences depend on the time period over which the outcomes are evaluated. Specifically, risk aversion decreases with respect to future gains and increases with respect to future losses. Based on the experimental observations, an intertemporal choice model is constructed to study the change in the behavior due to future uncertainty.

P2-36. Dilip SOMAN (College of Business, Univ. of Colorado at Boulder)

Virtual progress: The importance of being on the move vs. getting there.

We examine service situations in which the start and end times are constant but the path characteristics are manipulated. Subjects indicated a preference for services with lower idle duration, in which idling occurs in the middle of the interval rather than at the beginning, and in which there is physical movement towards the goal for a large part of the interval. Thus, subjects choose alternatives in which they experience a sense of progress even though the actual goal might be reached at the same time. We refer to this as virtual progress. We show that virtual progress influences choice prior to the service experience but not the satisfaction when evaluated after the service. Further, virtual progress influences preferences in situations where the indivdual presonally experiences the passage of time due to salience.

P2-37. Scott HIGHHOUSE (Bowling Green State Univ.), Susan MOHAMMED (Pennsylvania State Univ.), & Jody R. HOFFMAN (Bowling Green State Univ.)

Temporal discounting of strategic issues: Bold forecasts for opportunities and threats.

Asymmetrical discounting of strategic issues was found such that students (N = 86) discounted distant threats more than distant opportunities. In addition, even though immediate threats were viewed just as likely to occur as immediate opportunities, distant threats were seen as less plausible than distant opportunities. Experiment 2 (N = 222) found that a manipulation of a threat's likelihood of occurring had no effect on the temporal discounting of the hypothetical threat. However, the perceived control of threats increased as temporal distance increased. We conclude that perceived control plays an important role in the reduced plausibility of distant threats.

P2-38. Michael J. ZICKAR, & Scott HIGHHOUSE (Bowling Green State Univ.)

Examining framing effects using item response theory.

Item response theory (IRT) models were estimated for four risky-choice problems, answered by students under either a gain or loss frame. IRT methodology allowed an in-depth examination of several issues that would be difficult to explore using traditional methodology. Results support the typical framing finding of risk-aversion for gains and risk-seeking for losses. However, results suggest that individual differences in preference-for-risk are more influential in predicting risky choice than framing condition. Also, these results suggested that the Asian Disease problem, most often used in framing research, has anomalous statistical properties when compared to other framing problems.

P2-39. Laurie ZIEGLER (Univ. of Texas at Dallas)

Risk preferences in strategic decision making: Influences of decision importance.

Research supporting prospect theory's predictions concerning the effects of gain/loss framing on risk preferences has focused on very important decisions. However, it has not systematically examined the effects of decision importance on risk preferences. The joint effects of gain/loss orientation and decision importance on managerial risk preferences in strategic decision making are examined. Experimental results indicate that decision importance moderates gain/loss framing effects. Subjects' choices were risk seeking for low importance decisions framed as gains. Consequences of this study for understanding how decision importance affects risk preferences are considered.

P2-40. Timothy R. JOHNSON (Dept. of Psychology, Univ. of Illinois at Urbana-Champaign)

On the (nonlinear) multivariate analysis and representation of subjective probability judgments.

One common form of data in judgment and decision making research is a n by m array of subjective probability judgments made by n judges regarding m events. This study demonstrates the application of multivariate analytical and representation methods to these arrays based on nonlinear principal components and biplot displays. In the application and interpretation of these methods, the distinction is made between half- and full-range judgments as well as between confidence and forecasting judgment data. The distinction is also made between absolute and ordinal level judgment data. The proposed methodology is extensively demonstrated with an empirical data set.

P2-41. Robert B. BRANSTROM (Dept. of Psychology, Univ. of California, Berkeley)

Choice response times and an interactive activation network model of multiattribute choice.

Multiattribute decision models typically assume that utilities derived from attribute values are based on fixed utility functions. Alternatively, an interactive activation network model assumes attribute nodes receive initial activations related to attribute values, but activations then change dynamically as the network "settles into" a solution. In a two alternative, two attribute choice task, this model, like utility based models, correctly predicts decisions for compensatory and dominated choices. However, the network model predicts a pattern of decision response times not predicted by conventional multiattribute models. Empirical results support the network model's predicted patten of response times.

P2-42. Kenneth RONA (Fuqua Sch. of Business, Duke Univ.)

Growing decision rules.

This paper presents a computer simulation that "grows" decision rules according to evolutionary principles. Decision rules are made up of elementary information processes with initial rules being randomly generated. A large number of rules are constructed, presented with a risky choice problem, and then evaluated based on their performance on some set of goals. Rules are then selected from the population based on their performance and combined with each other to create new decision rules to be represented in the next generation. Rules that are more fit will emerge over time, thereby identifying the most fit type of rules in a given decision environment.

P2-43. Marcus O'CONNOR (Univ. of New South Wales, Australia)

The asymmetry of judgemental confidence intervals in time series forecasting.

This study examines the prevalence and determinants of the symmetry of judgemental confidence intervals around the forecast in time series forecasting. Most prior research on judgemental confidence intervals has assumed that the intervals are symmetrically placed around the forecast. However, this study shows that people are extremely disposed towards estimating asymmetric confidence intervals and that many of these intervals are grossly asymmetric. Results indicate that the placement of the forecast in relation to the last actual value is a major determinant of the direction and size of the asymmetry.