Professional Psychology: Research and Practice 1996, Vol. 27, No.

5, 521-526

Copyright 1996 by the American Psychological Association, Inc. 0735-7028/96/J3.00

A Decision-Making Technique for Choosing a Psychology Internship

Alan E. Stewart
University of Memphis

Elizabeth A. Stewart University of Georgia

A new method, the paired-comparison ranking technique, is proposed for assisting with internship selection decisions. This technique is discussed in terms of its ability to minimize the effects of decision frames (cf. D. Kahneman & A. Tversky, 1984) as internship sites are selected. The technique is also compared with M. C. Jacob's (1987) internship decision grid. The paired-comparison ranking technique requires a greater number of choices to be made about internship sites and may require internship applicants to think more deeply about their alternatives. Specific instructions and a detailed example are provided to illustrate how to complete and interpret the paired-comparison ranking technique.

Despite the breadth of published studies on internship selection, a specific focus on decisions and decision-making processes involved in choosing an internship site has developed only recently (Gloria & Robinson, 1994; Rodolfa, Haynes, & Kaplan, 1995; Stedman, Neff, Donahoe, Kopel, & Hays, 1995). Applicants are faced with a myriad of decisions in the interview and selection phases of the internship process (Stewart & Stewart, 1996). During and after interviews, applicants may decide how much interest to show in a site, how much of their own interests or needs must be compromised to fit with internship faculty or training opportunities, and how to respond to inquiries from sites (Johnson, 1987). Applicants must invariably make these decisions (and others) with incomplete information and without the levels of confidence they may desire. In the internship selection phases, applicants may consider how to prioritize or rank sites along important dimensions. Applicants may also select a site that represents their number one choice and then must decide whether and how to inform a site that it is a top contender. Prioritizing sites and making a number one choice may assume increasing importance as competition for limited internship placements grows and as internship sites seek talented, committed, and well-matched trainees (Stewart & Stewart, 1996). Our primary goal in this article is to present a new technique to assist with selecting internship sites. In presenting our

method, we review how applicants' decisions in selecting an internship may be affected by the way in which such decisions are framed and the alternatives are perceived. We describe Jacob's (1987) grid technique for assisting with internship decisions and discuss the limitations of her method. Step-by-step instructions for completing our paired-comparison ranking technique are presented. We also provide a detailed example of how to complete and interpret the grid. Finally, we discuss the ability of both techniques to minimize the effects of perceptual frames that could bias internship decision making.

Internship and the Psychology of Choice

Internship applicants do not prioritize and select sites in isolation. They approach the internship process with professional goals about what kind of work they ultimately would like to do and how particular internships may help them achieve their goals (Megargee, 1992). In a similar manner, the personal and practical considerations applicants bring to the selection process can influence choices that must be made (Rodolfa et al., 1995; Stewart & Stewart, 1996). Wider influences, such as the current state of the profession and social, economic, and even political variables, can also affect the applicant's ability to obtain a desired internship experience. Both students and training directors have expressed concern in recent years that the available number of internship slots has not kept pace with demand (Gloria & Robinson, 1994; Murray, 1995; Reich, 1996). As a result, students may be applying to progressively more sites to increase their chances of obtaining a placement. With a greater number of sites being considered, applicants must process more information about programs such that they have less overall time and ability to examine individual sites and compare them with others. Under these conditions, the manner in which applicants conceive the internship choices may affect the decisions that are made. Two variables may affect applicants' decision-making processes as they begin to prioritize and rank prospective internship sites in late January and early February. First, the way site rankings are framed predominantly as either a gain or a loss situation could affect the applicant's preferences for particular sites. The work of Tversky and Kahneman (1981) and Kahne521

ALAN E. STEWART received his PhD in counseling psychology from the University of Georgia in 1994. He is a research postdoctoral fellow in clinical psychology at the University of Memphis. His interests include professional training and supervision, constructivism, Adlerian psychology, and mathematical psychology. ELIZABETH A. STEWART received her JD in 1986 and her MEd in 1992. She is completing requirements for her doctoral degree in counseling psychology at the University of Georgia. She is currently on her predoctoral internship at the Memphis Veterans Affairs Medical Center. Her interests include geriatrics, ethics, professional training, and psychological birth order. CORRESPONDENCE CONCERNING THIS ARTICLE should be addressed to Alan E. Stewart, Department of Psychology, University of Memphis, Memphis, Tennessee 38152. E-mail may be sent via the Internet to AESTEWART@CC.MEMPHIS.EDU.

522

STEWART AND STEWART

man and Tversky (1984) can be applied to internship decision making. Specifically, applicants who maintain a focus on the potentially negative outcomes of the internship selection process (e.g., obtaining no internship or being placed at a less than optimal site) may take more risks in ranking their sites than persons who have a more neutral or positive view about the consequences of their decisions. Construing internship decisions primarily in terms of negative possibilities or outcomes may result in applicants preferring sites that eliminate rather than reduce the perceived probability of these negative events, even when such sites may actually have a higher probability of rejecting the applicant or of not meeting his or her expectations. The second variable that could affect decision-making processes occurs as applicants compare and contrast sites both during and after site interviews. In examining the social cognitive aspects of choice, Houston, Sherman, and Baker (1989) and Houston and Sherman (1995) described how the decision between two alternatives may depend on the order in which the comparisons are performed. The first of two alternatives being presented affects what criteria are used in making the comparison. In this way, the first alternative performs a kind of standard setting function. The first alternative also could set the general evaluative and emotional tone of the decision process such that an unfavorable decision frame becomes operative. By translating Houston and Sherman's (1995) work to internship selection, one can see that if two sites share equally undesirable features, then unique and positive features of the first site the applicant visited or considered may stand out and make it the attractive alternative of the pair. Conversely, if two sites share desirable features, then preferences may be established according to the sites' unique negative attributes. In comparing sites, negative features of the first site may emphasize undesirable features of the second site and create an avoidanceavoidance conflict along with indecision. In addition, emphasis on negative features of sites may induce one to focus on negative outcomes such that decisions are framed in terms of not obtaining the desired internship. How can the effects of decision frames and the order in which alternatives are presented be reduced? Kahneman and Tversky (1984) as well as FischhofF, Slovic, and Lichtenstein (1980) suggested that the effects of variable preferences (stemming from the way decisions are framed) may be minimized by examining a choice situation from multiple perspectives. In practice, this means asking the same question in many different ways and resolving the disparities among answers. In addition, decisions themselves should be deconstructed such that the influences of important criteria are considered individually and interactively. Regarding internship site selection, this would involve not just choosing one site over another, but preferring a site with respect to a particular criterion.

dents rank their internship choices. In completing Jacob's grid, internship applicants evaluate candidate sites along criteria that are important to them. Applicants also may weight certain criteria so that sites fulfilling these more important requirements receive correspondingly more support. The overall ratings for each site are determined from tallies of ratings that are made across the individual variables. Although Jacob's (1987) decision-making grid has undoubtedly helped those who have used it, her method possesses several shortcomings that ultimately limit its contribution to the process. First, Jacob does not provide for or encourage direct, sideby-side comparisons of sites on selection criteria. In not emphasizing such paired comparisons, Jacob's method is inconsistent with the binary manner in which persons may examine similarities and differences between two people, things, objects, or events. Instead of thinking about or evaluating sites all at once, internship applicants are much more likely to compare any given internship to their ideal site, the first site they interviewed, the most recent site visited, and so forth. A second, related criticism of Jacob's (1987) method concerns the use of Likert rating scales to evaluate each site. In applying such rating scales independently to each site, users could find that many sites receive very similar ratings when the final tallies are calculated. Such a result would fail to capture how applicants perceive sites to differ. Although Jacob's method may produce site ratings that differ numerically, how do such numerical distinctions translate into psychologically meaningful differences? Overall, the criticisms of Jacob's decision grid pertain to what it may offer to applicants as an exercise in assisting with choices about training sites. Similar to physical exercise, the benefits obtained are a function of the effort invested. In not requiring applicants to compare sites in a pairwise manner and in allowing applicants to assign any rating to any site, Jacob's (1987) method may not require applicants to think deeply about either internship sites or the important criteria they will use to evaluate sites. In this regard, Jacob's method does not provide a rigorous decision-making workout.

The Paired-Comparison Ranking Technique

The technique we propose requires applicants to systematically evaluate all possible pairs of sites with respect to important criteria they select. Although this technique has not been extended previously to internship decision-making, the method of making pairwise evaluations of stimuli can be traced to the psychophysical methods (e.g., the method of just noticeable differences) developed primarily by Gustav Fechner (Boring, 1950; Fechner, 1860/ 1966). In addition, personnel and management psychologists use paired-comparison methods to reduce central tendency, leniency, and strictness biases in selecting employees and evaluating their job performance.

Jacob's Decision Grid

To date, only one article has proffered a method for helping students evaluate competing sites along relevant dimensions (Jacob, 1987), Along with a general discussion of how to time various preparatory activities for internship application and interviews, Jacob described and illustrated the use of a decisionmaking grid that, along with other information, may help stu-

Selecting Variables
The first step of our technique begins with the applicant's selection of relevant professional, personal, and practical criteria that will be used in comparing sites to one another. Although any number and kinds of criteria can be chosen, in gen-

TRAINING: CHOOSING A PSYCHOLOGY INTERNSHIP Table 1 Some Useful Criteria for Comparing Psychology Internships Professional variables Match with professional interests Ability to obtain specialized training Alternative rotations available Amount and quality of supervision Ability to work with desired clientele Reputation and prestige of site Amount of stipend or salary Works hours per week Routine opportunities to work on dissertation Ongoing opportunities to conduct research Job opportunities after internship Opportunities to obtain long-term or shortterm treatment experience Opportunities for test administration and interpretation Availability of live supervision Opinions and feedback from present and past interns Availability and amount of didactic experiences Ability to choose rotations Relative emphasis of training versus service Ability of internship to supply sufficient hours for licensure in particular states Extent to which internship specifies its goals or objectives Personal and practical variables Location Proximity to partner and family members Ability to travel to partner and family Ability to travel to or contact doctoral program Cost to travel to internship interview Potential cost to relocate to internship site Cost of living at the internship site Incidental costs of moving and settling Availability of medical and dental coverage Availability of other employee benefits Work opportunities for partner Simplicity and ease of preparing and submitting internship applications Subjective feelings in interacting with internship faculty during interviews Receptivity and friendliness of faculty and staff Amount of needed or desired contact with peers during internship Ease of using internship's physical facilities Level of personal safety at the site and in the internship's locality Availability of personal psychotherapy Availability of sick leave and holiday time from internship site Amount of travel required between sites in a consortium internship Ability to sell or rent house before relocating Note.The listing of professional variables was compiled primarily from Megargee (1992), Stedman et al. (1995), and Tedesco (1979). The listing of personal and practical variables was compiled from Rodolfa, Haynes, and Kaplan (1995) and Stewart and Stewart (1996).

523

eral they should be applicable to the sites under consideration and also should possess significance for the person completing the exercise. To facilitate identification of relevant criteria, we present a listing of professional, personal, and practical variables in Table 1. Because this listing does not exhaust the possible dimensions along which internships may be evaluated, users may want to modify the criteria listed or add their own. Determining the Importance of Selection Variables The second step involves prioritizing the selection criteria. After identifying the important variables, the applicant should follow the procedures outlined herein. Write the name of each

selection variable on a single piece of paper. Next, write the name of each selection variable on an index card, listing only one per card and numbering each card lightly in pencil. The relative importance of the selection criteria is determined for use after all sites are ranked in the comparison process. To make this determination, take the first variable card (i.e., the card labeled # 1) and lay it beside the second card ( #2 ). Decide which is the more important variable or criteria for the internship. For the variable that is more important, put a vertical hash mark next to that variable on the paper listing all of the selection criteria. Compare the first and third variables in the same manner. Put a hash mark next to the name of the criterion that is chosen in

524

STEWART AND STEWART

this comparison. Do this comparison for the first and fourth variables next. Continue this process until all variables have been compared to the first selection variable, being sure to record hash marks next to chosen variables in each comparison. Next, compare the second and third variables (Cards #2 and #3), then the second and fourth, second and fifth, and so on. When all comparisons for the second variable are complete, compare the third and fourth variables. With regard to the total number of these comparisons that must be performed, if n is the number of selection factors, then (n2 n) -H 2 decisions about the importance of selection variables will be made. For instance, if 6 variables were chosen from Table 1, then 15 comparisons would be required to determine their relative importance before beginning the site-ranking process. To determine the overall relative importance of selection variables, tally the hash marks beside the name of each variable. The criterion with the most hash marks is given the # 1 rating. This is the most significant criterion for evaluating sites. Tally the number of times each variable was chosen and rank the remaining variables. In each case, boldly mark on the front of each variable card the rating that it received.

Choosing Internship Sites for Comparison

The third step involves generating a list of sites that will be compared to each other. Although any sites may be compared with one another in the exercise, we have observed that the most benefit comes from comparing sites that applicants are actively considering. These sites, along with other important contenders for which there is some level of doubt or indecision, should be included. Sites that are being marginally considered or that have been all but eliminated could be omitted from the exercise. When a suitable pool of sites has been identified, write the names of the sites on index cards (one site per card).

Comparing Internship Sites on Selection Variables

The fourth step of the technique involves the actual pairwise comparison of sites on selection variables. Results of comparing internship sites will be recorded in a grid. The grid sheet must be prepared, however, before the actual comparisons of sites begins. This is done by writing the names of internship sites along the rows of the grid. The order in which sites are listed is unimportant. The names or abbreviations of the selection variables are written across the top in the grid's columns, beginning with the most important variable and progressing to those that are less important. The rightmost column should contain the selection variable that was ranked as the least important. An example of a completed grid is provided in Figure 1. Internship sites can now be compared in a pairwise manner for each of the selection variables. To begin the process, lay the first, most important selection variable card on the table. Next, lay the cards for the first and second sites on the table underneath the variable card. Compare the first and second sites in terms of the most important variable. Which of the two sites most clearly satisfies the requirements of this variable? Make a choice and allow no ties. Put a vertical hash mark on the grid under this variable's column in the row next to the site chosen. Again, these hash marks will be counted and entered into cells

of the grid corresponding to the number of times a site was chosen. Compare Sites #1 and #3 on the first selection variable, and record the choice by placing these cards under the first variable's card. Repeat the procedure with Sites # 1 and #4, Sites # 1 and #5, and so forth. When all sites have been compared to the first site, compare sites to the second site (i.e., compare Sites #2 and #3, #2 and #4, and so forth). Do these comparisons until all sites have been compared to each other on the first selection variable. Again, be sure to record the outcomes of each comparison by placing a hash mark in the appropriate row. Also notice that, because Sites # 1 and #2 were compared when working with the first site, there is no need to compare the sites again when working with second site. When all sites have been compared on the first variable, it will be possible (but not necessary at this point) to count the hash marks in each cell and enter a tally in the small box in each cell. For instance, in Figure 1, the first variable is emphasis of training over service. Looking at the first row (Site A) under this important feature, there are two hash marks. Therefore, a 2 is entered into this cell. Notice how the remaining cells for this variable are tallied. The sites can be compared in a pairwise manner on the second variable. Again, start by laying the card with this variable's name on the table. Place the cards with Sites #1 and #2 on the table under this variable. Compare the two sites in terms of which best meets the requirements of this variable. Again, record choices by placing a hash mark in the second column on the row corresponding to the chosen site. Next, compare Sites #1 and #3, #1 and #4, and so forth. When sites have been compared on the first variable, move on to the second variable. Continue to compare all sites on the remaining variables in the aforementioned manner. After comparisons are completed for each variable, the hash marks may be tallied and entered into the respective grid cells. Although the process tends to progress more rapidly as the technique is learned, do not become rushed in making the comparisons. It may also benefit applicants to either alternate how they place site's cards in front of them or to alternate which site's cards they look at first when making comparisons. Mixing the sites in this manner will help to minimize the possibility that the simple order of making the comparison will affect the choices made.

Completing and Interpreting the Grid

Before calculating the final ratings of each site, it is possible to include the effects of variables that were excluded during the pairwise rankings. For instance, selection variables such as American Psychological Association (APA) accreditation or American Board of Professional Psychology (ABPP) approved neuropsychology supervision can only be answered in a yes-no or a yes-no-provisional format. Several sites may either satisfy or fail to meet these (or similar) requirements. In the case where a pair of sites satisfy such a criterion, a clear preference cannot be determined in a pairwise comparison. For example, one site cannot be more APA-accredited than another. Rather than comparing sites in a pairwise manner on these kinds of variables, it is possible to add one or more points to the totals of all sites that satisfy the criteria. That is, for all sites that are fully

TRAINING: CHOOSING A PSYCHOLOGY INTERNSHIP

525

Selection variables
(your criteria)

Sites
(your sites)

Site A Site B Site C Site D Site E Site F Site Site Site Site G H 1 J
III/
2

Mr
Wf

** - nil 19
Lo_
m

Z7
SO

LCL

LL LO_

71

21
31 30

///

77T W

III! lll\ 9 /// I 8 vwr -

35 39
Totals

Figure 1. Sample decision grid completed using the paired-comparison ranking technique.

APA accredited, 2 points could be added to the total; a single point could be added for provisional accreditation, and so forth. Although such dichotomous or polychotomous variables can be included in the decision grid, we have observed that applicants benefit maximally from working with selection variables that vary over a continuum, such as the variables in Figure 1. In the fifth step, the overall ratings for each site can be determined by summing the cells in each row. In the sample grid in Figure 1, Site D receives the highest rating of 71, followed by Site C, which runs a distant second with a rating of 50. The site receiving the lowest ranking after pairwise comparisons is Site A. Notice there is no need to obtain column sums because there are an equal number of comparisons per variables, that is, (s2 s) H- 2 where s equals the number of sites being compared. Regarding a fuller interpretation of the grid results, more information is conveyed by the grid exercise than may be revealed in the total ratings for each site. For the first two or three selection variables rated, find the sites with the highest ratings. These are the sites that satisfy the two or three most important internship criteria that were developed. In Figure 1, Site D leads the tallies for the first three selection variables. Site I is a close second on all three dimensions. The ways both ratings and running totals of ratings change as sites are evaluated on subsequent criteria shows how different sites are important in different ways. For instance, although Site I is a very close second to Site D across the three most important variables, Site I ultimately ranks forth, 32 points behind Site D. As the sites are evaluated in terms of the last four variables

(work opportunities for spouse, match with interests, freedom to choose rotations, and location) the ratings for both Sites C and J surpass those of Site I. Site C, ranking third and two points ahead of Site J after ratings of the first three variables are calculated, is rated as the #2 choice when totals are completed. Site G is rated fifth in terms of cumulative totals after comparing sites on the work hours criterion. This is the highest ranking the site received after it started and finished with a seventh-place ranking. Although there is not an appreciable change in site rankings from the seventh (freedom to choose rotations) to the eighth (location) variables, the rankings for five sites change when the seventh variable is evaluated after the sixth (match with professional interests) variable. This result suggests that most of the chosen selection variables are sufficiently salient to affect preferences for some of the lower-ranking sites. Discussion The grid in Figure 1 required approximately 1.5 hours to complete. This interval included preparing the grid, sorting the selection variables to determine their importance, performing the actual pairwise comparisons for the different sites, and calculating the tallies. Given the amount of time applicants spend preparing applications and completing interviews, the time invested in completing the grid is quite small by comparison. Grids larger than the one completed here (e.g., with 10 sites and 10 variables) may become tiresome as sites are rated on

526

STEWART AND STEWART

less important selection variables. We have found that reducing both the number of sites and variables does not appear to sacrifice the informational value provided by the technique. Compared to Jacob's (1987) grid method, the paired-comparison ranking technique requires applicants to make decisions between sites rather than to simply assign a site rating for particular criteria. Consequently, our method is consistent with real-world decisions applicants must make in choosing between individual sites. A result of our pairwise comparison method is that applicants must make more decisions about sites and thereby will examine both the sites and their selection criteria in a more varied and thorough manner. In completing the pairwise comparisons for the grid in Figure 1, 360 decisions were required. For the same size grid, Jacob's (1987) method would involve performing only 80 ratings, or about 22% of the ratings involved in the paired-comparison ranking technique. Another distinguishing feature of the two techniques is that, with the paired-comparison ranking technique, it is not easy to tell how it's going to come out with regard to the final site rankings. That is, because trends may be difficult to discern as the procedure is being completed, it is difficult for applicants to bias the overall results. Although Jacob (1987) suggested that a new site should be rated without comparison to previously rated sites, it would seem that an especially good interview or other contact with a site could easily bias a preponderance of the ratings given to a particular site. In practice, this result could occur by applicants assigning ratings to the favored site to make it match or exceed other contenders (Tversky, Sattath, & Slovic, 1988). Thus, it would be possible to fudge on Jacob's (1987) method to produce results desired at any one time. The paired-comparison ranking technique is more robust to these kinds of biases in that sites are rated within selection variables rather than across them. In making numerous comparisons within a single variable, the pull of a particular site will be diminished. Rating sites within selection variables also reduces the effects of the aforementioned decision frames (Kahneman & Tversky, 1984; Tversky & Kahneman, 1981). By considering specific criteria at the time choices are made in the grid, the influence of extraneous variables, such as thinking of sites as potentially rejecting or accepting of the applicant, will be minimized. Jacob's (1987) method may also reduce the susceptibility to decision frame effects but predictably not as much as the paired-comparison method, because her method seems vulnerable to the effects of overall impressions applicants develop about sites. A practical question could arise about when to complete this technique or similar exercises, such as Jacob's (1987) grid. In our experience, applicants may derive the most benefit from completing the paired-comparison technique after visiting all seriously considered internship sites. As an alternative, the exercise could be performed as applicants attempt to narrow the pool of sites to visit. The technique could also be used both before and after site visits. In summary, the process and experience provided by the paired-comparison ranking technique may be more important than the products (i.e., its numerical ratings and site rankings). The purpose of the technique is to help applicants think deeply about sites from the multiple perspectives provided by important selection criteria. Although the final ratings may agree gen-

erally with what an applicant expected, the more productive outcome of the exercise may be one in which the applicant's expectations and the final results are discrepant to some degree. In these cases, the technique creates a stimulus for further exploration of both sites and selection variables. Consequently, we hope the usefulness of the technique is judged not only by the results it generates, but also by the experiences it provides in helping applicants view their internship decisions from multiple perspectives.

References
Boring, E. G. (1950). A history of experimental psychology (2nd ed.). New \brk: Appleton, Century, and Crofts. Fechner, G. T. (1966). Elements ofpsychophysics(H. E. Adler, Trans., D. H. Howes & E. G. Boring, Eds.). New York: Holt, Rinehart, and Winston. (Original work published 1860) Fischhoff, B., Slovic, P., & Lichtenstein, S. (1980). Knowing what you want: Measuring labile values. In T. S. Wallsten (Ed.), Cognitive processes in choice and decision behavior (pp. 117-141). Hillsdale, NJ: Erlbaum. Gloria, A. M., & Robinson, S. E. (1994). The internship application process: A survey of program training directors and intern candidates. The Counseling Psychologist, 22, 474-488. Houston, D. A., & Sherman S. J. (1995). Cancellation and focus: The role of shared and unique features in the choice process. Journal of Experimental Social Psychology, 31, 357-378. Houston, D. A., Sherman, S. J., & Baker, S. M. (1989). The influence of unique features and direction of comparison on preferences. Journal of Experimental Social Psychology, 25, 121-141. Jacob, M. C. (1987). Managing the internship application experience: Advice from an exhausted but content survivor. The Counseling Psychologist, 15, 146-155. Johnson, M. C. (1987). The predoctoral internship selection process: Tactical and ethical realities. Professional Psychology: Research and Practice, 77,291-293. Kahneman, D., & Tversky, A. (1984). Choices, values, and frames. American Psychologist, 39, 341-350. Megargee, E. J. (1992). A guide to obtaining a psychology internship (2nd ed.). Muncie, IN: Accelerated Development. Murray, B. (1995, July). Are internships getting harder to find? APA Monitor, p. 58. Reich, J. (1996, May). Fixing the internship imbalance APA Monitor, p. 49. Rodolfa, E., Haynes, S., & Kaplan, D. (1995). To apply or not to apply: That is the intern applicant's first question. Professional Psychology: Research and Practice, 26, 393-395. Stedman, J. M., Neff, J. A., Donahoe, C. P., Kopel, K., & Hays, J. R. (1995). Applicant characterization of the most desirable internship training program. Professional Psychology: Research and Practice, 26, 396-400. Stewart, A. E. & Stewart, E. A. (1996). Personal and practical considerations in selecting a psychology internship. Professional Psychology: Research and Practice, 27, 295-303. Tedesco, J. F. (1979). Factors involved in the selection of doctoral internship in clinical psychology. Professional Psychology, 10, 852-858. Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211, 453-458. Tversky, A., Sattath, S., & Slovic, P. (1988). Contingent weighting in judgment and choice. Psychological Review, 95, 371-184.

Received February 2, 1996 Revision received May 17, 1996 Accepted May 20, 1996