American Journal of Medical Genetics 33: 130-135 (1989) Effect of Framing on the Perception of -Genetic Recurrence Risks Shoshana Shiloh and Michal Sagi Institute for Child Development, Sheba Medical Center, Tel-Hashomer, and School of Education, Tel-Aviv University (S.S.) and Department of Human Genetics, Hadassah Medical Center, Ein-Kerem, Jerusalem (M.S.), Israel Individuals asked to evaluate genetic recurrence risks were found to be influenced by the way the risks were framed. Presenting a single risk figure resulted in overweighting of low probabilities and underweighting of high probabilities, as compared to presenting a list of sequential risks. Differences were also found between meanings attached to verbal expressions of risks when translating from verbal to numerical expressions, and vice versa. The implications of these findings for genetic counseling are discussed. vational factors have been implicated in the process of recurrence risk perceptions, including binary transformations, scenario presentations, prior expectations, and reproductive motivations [Lippman-Hand and Fraser, 197913; Shiloh and Saxe, 19881. The present study was undertaken to determine how risk perceptions are influenced by the way they are presented to individuals. This is a particularly important factor, since it is completely controlled by the genetic counselor. There is ample evidence that even subtle changes in the way risks are expressed can have a major impact on perceptions and decisions. Numerous examples of these “framing effects” have been demonstrated by Tversky and Kahneman  and Slovic et al. . For exKEY WORDS: risk perception, framing efample, McNeil et al.  found that presenting treatfects, genetic counseling ment alternatives for lung cancer in terms of survival rates versus death rates caused a dramatic shift in preferences. INTRODUCTION Manner of presentation was also suggested to influence Providing information on recurrence risks is a major how risk information is interpreted in genetic counseling component of genetic counseling. This information is [Pearn, 19791. Kessler and Levine  recently comconveyed in order to provide a basis on which counselees pared percentage and odds presentations and found that can make informed decisions about very important per- some individuals use a “number strategy” and therefore sonal matters such as marriage and childbearing. Several interpret the percentages as representing higher risk, studies have shown that risk information is interpreted while others use a “person strategy” and evaluate odds by counselees in a personal manner and that their deci- as indicating a higher risk. sions are related to their subjective perceptions of recurWe chose to study two other modes of presentation of rence risks rather than the “objective” risk [Lippman- risk in genetic counseling: single versus sequential, and Hand and Fraser, 1979a; Wertz and Sorenson, 1984; numerical versus verbal. A recurrence risk can be preShiloh and Saxe, 19881. These findings are consistent sented either as a single figure or in comparison to a with general rules of decision-making under risk, as for- variety of genetic risks, such as risk in the general popmulated by Kahneman and Tversky [ 19791 in their “pros- ulation or in other genetic conditions. Many counselors pect theory.” This model describes how prospects-beprovide some comparative data, but there are no definifore being evaluated-are psychologically represented tive guidelines regarding its importance and influence. through the application of several editing operations. Thus, we have set out here to compare people’s reactions The primary importance of subjective perceptions of to single versus sequential presentations of recurrence recurrence risks makes it essential to understand the risks. factors that influence them. Some cognitive and motiRecurrence risks can also be presented verbally or numerically. Counselors usually avoid verbal expressions of risk, such as “high,” “doubtful,” “low,” and “unlikely,” in an attempt to avoid judgmental and directive counseling [Hsia, 19791. Verbal expressions (which are also Received for publication September 12, 1988; revision received referred to as “qualitative expressions”) are not comDecember 16, 1988. Address reprint requests to Shoshana Shiloh, Ph.D., Institute pletely missing in genetic counseling; however, counselors do use them when referring to extreme risk figures, for Child Development, Sheba Medical Center, Tel-Hashomer 52621, Israel. or when asked directly for a verbal evaluation by the 0 1989 Alan R. Liss, Inc. Framing Effects in Risk Perceptions counselee, or when trying to correct a seemingly biased perception by the counselee [Carter et al., 1971; Emery, 1977; Lippman-Hand and Fraser, 1979al. Verbal expressions are almost always used by the counselees in their discussion of recurrence risks with their counselors. Therefore, we considered it informative to compare meanings attached to risk numbers versus words, and to assess whether translation of an expression from one mode to the other changes its meaning. MATERIALS AND METHODS Subjects Undergraduate and graduate students in Psychology, Educational Counseling, and Medicine at Tel-Aviv University and The Hebrew University of Jerusalem participated voluntarily in the study. Design and Procedure A two-group experimental design was employed in each study. Subjects were randomly assigned to an experimental group. No subject took part in more than one study. 131 Some such verbal expressions of genetic risks are presented here. You are requested to write down for each expression a percentage or percentages that seem to you equivalent to that verbal expression of genetic risk. You may use any percentage in more than one category.” Nine expressions were presented very low, quite low, less than moderate, moderate, above moderate, quite high, high, very high. The median was recorded for answers given as ranges of percentages: thus, 0-5% was coded as 2.5%. 2. The instructions for the numerical-to-verbal experiment were: “In genetic counseling a family is informed about their risk of having a child with a genetic disorder or disease. The risk is presented in percentages. These percentages can be interpreted differently (in words such as high or low) by different individuals. Some risk categories are presented here (in percentages). You are requested to write down, for each category of genetic risks, a verbal expression (e.g., very high, very low) which seems to you equivalent to these percentages. You may use any expression for more than one category.” Eight categories were presented, covering the full range of probabilities: 0-3%, 4-796, 8-12%, 13-20%, 21-30%, 31-50%, 51-70%, 71-100%. Study 1: Single Versus Sequential Presentation of Risk Factors A short questionnaire was designed, asking subjects to The median percentage of each category was recorded evaluate specific recurrence risks, given in percentages, for data processing. Verbal expressions given by respondcn an 11-point Likert scale (1 = very low, 11 = very ents were coded according to the same categories defined high). Two versions of the questionnaire were prepared above in the verbal-numerical experiment. When a verbal and administered to two different experimental groups, expression other than the nine listed ones was used, it as follows: was evaluated by two independent judges (social psy1. Thirty-seven subjects received the “sequential” chologists), and assigned to one of the categories. Agreequestionnaire consisting of nine objective risk figures ment between judges reached 94%. Cases of disagreement presented in a sequential manner: I%,5%, lo%, 15%, were resolved by discussion prior to substantial analysis 25%, 35%, 50%, 70%, 90%. of data. Expressions such as “almost certain” or “tremen2. Nine “single” questionnaires were prepared, each dous” were assigned to the “very high” category; “very dealing with one recurrence risk and administered to unlikely” was coded as “very low.” These procedures a group of 19-21 subjects (total 180 subjects). permitted comparison between numerical values attached to specific verbal expressions for both experiments. Subjects were instructed as follows: RESULTS In genetic counseling a family is informed about their risk of having a child with a genetic disorder or disease. Table I summarizes the subjective interpretations of The risk is usually expressed in percentages. A list of recurrence risks presented in the “single” versus the such genetic risks is presented below. (In the “single” study: one family was given a risk of . . .%). You are TABLE I. Interpretations of Recurrence Risks in “Single” requested to evaluate how this (these) risk(s) seem high and “Sequential” Modes of Presentation or low to you, by putting an x on the line(s). Study 2: Verbal Versus Numerical Presentation of Risk Factors Two questionnaires were designed, one to assess the translation from verbal to numerical probabilities and the other to assess the translation from numerical to verbal expressions. Each questionnaire was given to a group of 24 students. 1. The instructions for the verbal-to-numerical experiment were: “In genetic counseling a family is informed about their risk of having a child with a genetic disorder or disease. The risk is presented in percentages. These percentages can be interpreted differently (in words such as high or low) by different individuals. Subjective interpretations Recurrence risk (%) 1 5 10 15 25 35 50 70 90 * P < .05. ** P < .O1 Single presentation mean & SD (n = 19-21) Sequential presentation mean +- S D (n = 37) 5.09 & 2.51 6.74 -C 2.85 7.58 t 1.95 8.00 f 1.89 8.05 f 1.55 8.84 & 1.39 9.21 & 1.27 9.68 t 1.06 9.95 +- 1.32 3.19 t 1.98 4.78 f 2.02 5.97 k 1.93 7.00 t 1.73 7.69 t 3.09 8.73 f 1.43 9.40 t 1.04 10.32 t 0.75 10.92 t 0.28 t 2.99** 2.67** 2.91** 1.93* 0.57 0.28 0.56 2.35* 3.25** Shiloh and Sagi 132 “sequential” mode. There were significant differences between the two modes of presentation a t the ends of the scale. Single presentations of lower recurrence risks (up to 15%) were assigned higher values of risk by the subjects, while single presentations of higher risks (70% and over) were assigned lower values of risk by the subjects. The subjective interpretation of risk in both modes of presentation increased in direct relation to the objective risk. This was determined by calculating the Pearson correlation coefficients between objective risks and subjective interpretations: the results were 0.86 for the single mode and 0.93 for the sequential mode. A logarithmic relationship between objective and subjective risks was observed in both modes of presentation. Increases in subjective interpretations of risks were much steeper in the lower range of risks than in the higher range. In the single presentation mode, nearly the same increase in subjective scale points was obtained between 1%and 10% and between 10% and 90%. In the sequential presentation mode, the increase obtained in subjective interpret.ations between 1%and 15% was equivalent to that obtained between 15% and 90%. However, as illustrated in Figure 1, the regression line in the sequential group is much steeper than the regression line in the single group. This explains the increasing differences in interpretations toward the extreme ends of the recurrence risks and the crossing over of interpretations at about 40% risk. Table I1 and Figure 2 present data comparing the translations of numerical-to-verbal and verbal-to-numerical expressions of risk. Subjects assigned a higher numerical risk when translating from words to numbers than from numbers to TABLE 11. Recurrence Risks Attached to Verbal Expressions: Translation of Numerical to Verbal and Verbal to Numerical Risk Expressions Numerical risk (%) Verbal to numerical Verbal exmession Mean f SD Very- high 68.25 t 26.63 62.13 f Hieh 28.12 Quite high 55.29 t 29.21 Above moderate 46.70 k 24.85 Moderate 33.85 f 16.67 Less than moderat,e 25.23 k 14.70 Quite low 14.39 +8.81 Low 7.73 +5.34 Very low 3.85 +3.25 * P < .05. ** P < .O1 - Numerical to verbal n Mean f SD n t 24 52.58 t 25.52 80 2.61* 24 23.90 t 19.48 38 6.33** 24 26.93 f 16.89 7 2.43* 24 21.50 2 9.62 8 2.77** 23 10.96 f 6.09 14 4.92** 22 13.25 t 3.25 2 1.13 22 4.92 2 2.89 6 2.56” 24 4.41 f 2.61 21 2.59* 23 2.12 k 1.44 13 1.81 100- Y u) ._ 80- 11 VJ = .-0 10 U m very low 9’ U ?! 8. L Q) 7. ” P 0 .-Q) 0 3 v) ‘Y 3 t 1 5 1015 quite Less mo e more Low * than r a k - than moderate moderate *. hi& very .-- high * I. Fig. 2 . Verbal and numerical expressions of genetic recurrence risks. -, Numerical-to-verbal (frequency of expression use-in parentheses); - - - - - - -, verbal-to-numerical ( n = 24). * P < .05; **P < .01. U U LOW 25 35 50 Recurrence 70 Risks * 90 (%) Fig. 1. Subjective interpretations of recurrence risks: Sequential presentations ( n = 37) versus single presentations ( n = 19-21). -, Sequential presentation; -----, single presentation. words. This held for all nine of the verbal expressions, and was statistically significant for seven of them. “High” risk produced the most dramatic results: translating this verbal expression into numbers resulted in a mean risk value of 62.13%. On the other hand, when translating from numbers to words, the expression “high” was assigned to much lower risk values: mean 23.90%. The data also show that subjects tended to use relatively few verbal categories in response to numerical stimuli: they responded with a mean of 4.88 verbal categories for eight numerical categories. And, most of their responses were extreme; for example, 24 subjects used the expression “very high” 80 times, “moderate” 14 times, Framing Effects i n Risk Perceptions and “low” 21 times. The spontaneous use of numerical categories in response to verbal expressions was much more varied mean 8.46 numerical in response to nine verbal categories. DISCUSSION Our results support the hypothesis that the way in which genetic recurrence risks are presented (i.e., single risk figures versus a sequence of risks, verbal versus numerical expressions) determines the meanings attached to these risks. Presenting recurrence risks as single figures resulted in overweighting of low probabilities (less than 25%) and underweighting of high probabilities (over 70%), indicating an insensitivity to increasing values of recurrence risks. This effect was diminished when a recurrence risk was presented with several other risks. Insensitivity to statistical information has been found in other cognitive-psychological studies as well [Lyon and Slovic, 1976; Kahneman and Tversky, 19731. Fischhoff et al.  argued that this insensitivity to statistical information is due to the between-subject designs employed in these studies, exposing participants to only one value of the statistical information. Such a procedure tends to ensure low salience of the statistical information, thereby reducing the likelihood of it being used. It follows that respondents’ tendency to use statistical information could be increased by providing this information in a within-subjects design, which exposes them to several different values of the statistical variable. A series of experiments designed to test these ideas confirmed them [Fischhoff e t al., 19791, and our findings also support this cognitive rule. These processes are influenced by the way information is framed and can explain why the adverseness of a small chance of a severe loss is amplified and the value of long shots is enhanced [Kahneman and Tversky, 19841. This framing effect can also explain some findings obtained in other studies on genetic counseling. Wertz et al.  found a “flat” curve for the modal category of interpretation of risks between 10% and 50% for counselees, and between 25% and 50% for counselors. This “flatness” can be due to insensitivity to the statistical information, caused by presentation of single risk figures without a comparison to other recurrence risks. This notion requires further study within the context of genetic counseling, and its effect should be tested on counselors as well as on counselees. We suggest exploring the possibility that some gaps in interpretation between counselors and counselees arise because counselees, unlike counselors, are exposed to single risk figures. The logarithmic nature of the relationship obtained in both modes of presentation between objective risks and subjective interpretations is equivalent to a phenomenon discovered in perceptual judgments and known as Weber’s Law [Fechner, 18601. This law states that a change in stimulus intensity, in order to be detected, must be a constant percentage of the initial value. Therefore, slight changes are sufficient for detection at lower stimulus intensities, and larger changes in stimulus intensity are required for detection as the baseline stimulus 133 becomes higher. We observed the same trend, with the stimulus being the objective risk and perceptions being the subjective interpretations. We suggest that interpretations of genetic risks are subject to these same general rules of perception, which may explain some of the findings in studies on genetic counseling, irrespective of the personal meanings attached to the stimulus. Risk perceptions were affected by the use of verbal as opposed to numerical expressions. Words such as “high” and “low” to describe probabilities were found to have different meanings when spontaneously ascribed to numerical probabilities than when read directly and translated into percentages. Subjects attached lower percentages to verbal expressions of probabilities used by themselves, and higher numerical values to verbal expressions presented to them (except for the “very low” category). For example, when subjects used their own words, the term “high risk” had an average risk value of 23.90%; when subjects ascribed a numerical value to the given term “high risk”, the average value was 62.13%. “Low risk” presented verbally earned a mean numeric value of 7.73%; when asked to ascribe words to numbers, “low risk” scored 4.41%. This finding might stem from the different cognitive processes applied in active versus passive use of language [Clark and Clark, 19771, and/or the different perspectives of actor-observer behavior [Jones and Nisbett, 19711. We suggest that given verbal expressions are interpreted by their “lexicographic” (dictionary) meaning-“When you say high risk you mean something over 60%.” On the other hand, active use of verbal expressions denotes its personal meaning-“25% is ‘high risk’ for me.” Another explanation may reside in the nature of the experimental situation. Subjects requested to use numerical expressions, which is not their natural language, try to be more objective and precise and therefore less personal. Subjects requested to use verbal expressions tend to express more of their personal feelings. Cognitive studies performed on the verbal-numerical probabilities distinction have produced mixed results. Most individuals use verbal terms when spontaneously expressing their opinions about probabilistic events, but most decision-makers prefer to receive numerical probabilities [Erev and Cohen, in press]. Considerable interpersonal variance has been found in the meanings ascribed to words used to define probability [Lichtenstein and Newman, 19671. High consistency was found in interpretations of these expressions by specific individuals, leading to the conclusion that numerical rather than verbal expressions should be used for communication [Bryant and Norman, 19801. The effectiveness of the two modes of presentation for decision-making purposes remains unclear. Zimmer  suggested that people are less likely to make judgmental biases using verbal than numerical expressions. Budescu et al. [ 19881 found that numerical expressions were better, and Erev and Cohen [in press] found no difference in the efficiency of the two modes of presentations. The verbal-numerical distinction was found in our study to be highly relevant for genetic counseling. Further research is required to elucidate its effect on the communication and decision-making in genetic counseling. 134 Shiloh and Sagi Another interesting finding was that in translating from numbers to words, fewer and more extreme categories were used. This is consistent with the findings of Cohen et al.  that: “in the domain of losses, subjects’ choices are not made on the basis of the precise probabilities of relevant events but on the basis of vague categories of, say ‘belief which they assign to these events” (p. 220). These verbal categories appeared to be extremely coarse for some individuals. This phenomenon can also explain the tendency of genetic counselees to transform recurrence rates into binary perceptions (whether it will or will not happen). Described by Lippman-Hand and Fraser [ 1979b], this can be viewed as an expression of the basic cognitive process of translation of risks from numerical to verbal language. It can be assumed that a personal meaning is added in these translations, which tends to yield more extreme interpretations, and to reduce the number of categories used. In our study, this tendency was found as a cognitive process, without the emotional burden and the personal consequences involved in the real genetic counseling situation. The addition of an emotional factor might have an amplifying effect (more extreme perceptions and less categories) on this basic cognitive process, leading to the described binary perceptions. The compatibility between our experimental results obtained in a nongenetic population and observations obtained in studies conducted in real genetic counseling situations supports the generalizability of our results to genetic counseling. However, a replication of our findings in genetic counselees’ populations is recommended, although ethical considerations may complicate a design of such a field study. Practical Implications The implications of our findings for genetic counseling are straightforward. Presenting a recurrence risk of 5% by itself would lead a counselee to perceive it as higher than if it were presented in comparison with other recurrence risks. Single or compared presentations would not make a difference in the 25-70% range. And, in the 70% and higher range, a comparison to other risks would result in higher risk perceptions. The double meaning found for verbal probabilistic expressions is also important for communication in genetic counseling. Clarifying the differences between sending and receiving a verbal message is essential for ensuring reliable communication between counselor and counselee. Antley [ 19791 argued that neutral information-giving in genetic counseling is not possible, partly because the counselor chooses the portion of genetic information to be provided, and that choice reflects his/ her own values. We suggest, in addition, that neutrality of information-giving is also impossible because of the framing effects involved in any mode of presenting risk information. This property of genetic information-giving has been described bv Kessler 11979. 19801 and was referred to as metacommunication-the deeper, connotative level of communication, which includes the syn‘Ontext in which it takes place’ tactic, physical, and Formulation effects can Occur unconsciously without anyone being aware of the impact of the frame on the ultimate decision. They can also be used deliberately to manipulate the relative attractiveness of options in various settings and in genetic counseling as well [Thaler, 1980; Kessler, 19811. We think that the genetic counselor, as a professional, has an obligation to be aware of the magnitude of framing effects of different presentation modes of risk information. A list of framing effects of genetic information, such as that demonstrated in the present study and in the Kessler and Levine  study should be established. Following this, the positive and negative presentations of recurrence risks should be studied, such as 25% for an affected child as opposed to 75% for a healthy child. An important issue raised here is the right way to present recurrence risks, the implication being that there is a right and wrong way of perceiving recurrence risks. This assumption is not valid, since risk perception is a multidimensional and personal process [Vlek, 19871. The right way must be tailored for the individual counselee and to specific counseling situations as part of the counseling process. Moreover, the counselor may choose to present the information in more than one way during the counseling process, according to the different needs and concerns expressed by the counselee. Multimethods of presentation may serve as a powerful technique for clarifying risk information and for helping clients gain a better understanding of the personal meaning of the information. Divergent presentations are also recommended as a means to overcome judgmental biases and to enhance objectivity in communication [Tversky and Kahneman, 19811. Finally, in this study we dealt with the communicational properties of risk perception. A complete picture of this issue should take into consideration the personal level as well, and more research is required to determine the interactions between presentation modes and other relevant factors for risk perception, such as prior expectations [Shiloh and Saxe, in press], personal cognitive styles [Kessler and Levine, 19871, or emotional states [Antley, 19791. REFERENCES Antley RM (1979):The genetic counselor as facilitator of the counselee’s decision process. In Capron AM, Lapp6 M, Murray RF, Powledge TM, Twiss SB, Bergsma D (eds): “Genetic Counseling: Facts, Values, and Norms.” New York Alan R. Lisa, Inc., for the National Foundation-March of Dimes. BD:OAS XV (2):137-168. Bryant GD, Norman GR (1980): Expressions of probability words and numbers. N Engl J Med 302411. Budescu DV, Weinberg S, Wallsten TS (1988): Decisions based on numerically and verbally expressed uncertainties. J Exp Psycho1 [Hum Percept] 14(2):281-294. Carter CO, Fraser Roberts JA, Evans KA, Buck AR (1971): Genetic clinic: A follow-up. Lancet 1:281-285. Cohen M, Jaffrey JY, Said T (1985): Individual behavior under risk and under uncertaintly: An experimental study. Theory Decision 18203228. Clark HH, Clark EV (1977): “Psychology and Language: An Introduction t o Psycholinguistics.” San Diego: Harcourt Brace Jovanovich. Emery AEH (1977): Changing patterns in a genetic counseling clinic. In Lubs HA, de la Cruz F (eds): “Genetic Counseling.” New York: Raven Press. pp 113-120, Ere” I, Cohen BL: Verbal numerical probabilities: Efficiency, biases and the preference paradox. Organ Behav Hum Perform, in press. Framing Effects in Risk Perceptions Fechner G T (1860): “Elements der Psychophysik.” Leipzig: Breitkopf und Hartel. Fischhoff B, Slovic P, Lichtenstein S (1979): Improving intuitive judgment by subjective sensitivity analysis. Organ Behav Hum Perform 23:339-359. Hsia YE (1979): T h e genetic counselor as information giver. In Capron AM, Lapp6 M, Murray RF, Powledge TM, Twiss SB, Bergsma D. (eds): “Genetic Counseling: Facts, Values, and Norms.” New York: Alan R. Liss, Inc., for the National Foundation-March of Dimes. BD:OAS XV(2):169-186. Jones EE, Nisbett R E (1971): “The Actor and the Observer: Divergent Perceptions of the Causes of Behavior.” New York General Learning Press. Kahneman D, Tversky A (1973): On the psychology of prediction. Psych Rev 80:237-251. Kahneman D, Tversky A (1979): An analysis of decision under risk. Econornetrica 47263-291. Kahneman D, Tversky A (1984): Choices, values, and frames. Am Psychol 39:341-350. Kessler S (1979): T h e genetic counselor as psychotherapist. In Capron AM, Lapp6 M, Murray RF, Powledge TM, Twiss SB, Bergsma D (eds): “Genetic Counseling: Facts, Values, and Norms.” New York: Alan R. Liss, Inc. for the National Foundation-March of Dimes. BD:OAS XV(2):187-200. Kessler S (1980): T h e psychological paradigm shift in genetic counseling. SOCBiol 27:167-185. Kessler S (1981): Psychological aspects of genetic counseling: Analysis of a transcript. Am J Med Genet 8137-153. Kessler S, Levine EK (1987): Psychological aspects of genetic counseling. IV. The subjective assessment of probability. Am J Med Genet 28361-370. Lichtenstein S, Newman JR (1967):Empirical scaling of common verbal phrases associated with numerical probabilities. Psychonomic Sci 9:563, 564. Lippman-Hand A, Fraser FC (1979a): Genetic counseling: Provision and reception of information. Am J Med Genet 3:113-127. 135 Lippman-Hand A, Fraser FC (1979b): Genetic counseling: The postcounseling period: I. Parents’ perceptions of uncertainty. Am J Med Genet 4:51-71. Lyon D, Slovic P (1976):Dominance of accuracy information and neglect of base rates in probability estimation. Acta Psychol (Amst) 40:287298. McNeil B, Pauker S, Sox H, Jr, Tversky A (1982): On the elicitation of preferences for alternative therapies. N Engl J Med 306:1259-1262. Pearn J (1979): Decision making and reproductive choice. In Hsia YE, Hirschhorn K, Silverberg RL, Godmilo L (eds): “Counseling in Genetics.’’ New York: Alan R. Liss, Inc., pp 223-238. Shiloh S, Saxe L (1988): Perception of risk in genetic counseling. Psychol Health: An International J (in press). Slovic P, Fischhoff B, Lichtenstein S (1982): Response mode, framing, and information-processing effects in risk assessment. In Hogarth R (ed): “New Directions for Methodology of Social and Behavioral Science: Question Framing and Response Consistency.” San Francisco: Jossey-Bass, pp 21-36. Thaler R (1980): Toward a positive theory of consumer choice. J Econ Behav Organ 1:39-60. Tversky A, Kahneman D (1981): The framing of decisions and the psychology of choice. Science 211:453-458. Vlek C (1987): Risk assessment, risk perception and decision making about courses of action involving genetic risk: An overview of concepts and methods. In Evers-Kiebooms G, Cassiman JJ, Van den Berghe H, d’Ydewalle G (eds): “Genetic Risk, Risk Perception, and Decision Making.” New York: Alan R. Liss, Inc., for the National Foundation-March of Dimes BD: OAS XXIII(2):171-207. Wertz DC, Sorenson J R (1984): Genetic counseling and reproductive uncertainty. Am J Med Genet 18:79-88. Wertz DC, Sorenson JR, Heeren T C (1986): Clients’ interpretation of risks provided in genetic counseling. Am J Hum Genet 39253-264. Zimmer AC (1983): Verbal versus numerical processing of subjective probabilities. In Scholz R (ed): “Decision Making Under Uncertainty.’’ Amsterdam: North Holland Publishing Co., pp 159-182.