how well do jurors reason?

4 downloads 573 Views 756KB Size Report
schools, a trade school, and a local uni- versity Adult subjects received modest flnancial compensation for their time Of the onginal 160 subjects, 152 partici-.
PSYCHOLOGICAL SCIENCE

Research Report HOW WELL DO JURORS REASON? Competence Dimensions of Individual Variation in a Juror Reasoning Task Deanna Kuhn, Michael Weinstock, and Robin Flaton Columbia University issues of competence, as well as individAbstract—Significant individual variation IS observed m how people reason as ual vanauon In addiUon, we are conjurors At the saUsficing end of a contin- cerned with generality Are the same difuum we identify, the juror draws on evi- ferences identifiable as well beyond the dence selectively to construct a single juror task"* story of what happened, with no acIndividual differences in jurors' thinkknowledgment of discrepant evidence or mg are likely, given the substanUal cogalternative possibilities A contrasting mUve demands of the task Penmngton theory-evidence coordination mode of and Hastie's model specifies a threeprocessing entails construction of multi- phase process First, evidence is drawn ple theories (story-verdict constellations) on to construct a plausible narrative that are evaluated against the evidence structure (story) of what happened Secand against alternatives Individual dif- ond, the set of verdict altemauves (and ferences Influence task outcome, the sat- cntena associated with each) must be isficing mode being associated with learned Finally, the story representation more extreme verdict choices and very IS compared with representaUons of the high certainty verdict categones to find an acceptable match Pennington and Hastie's research shows high correspondence between the The juror task is one of the most cogstory constructed and verdict choice nitively complex tasks that ordinary citWe undertake an lmUal conceptualizens are routinely asked to perform It is lzauon of potenUal mdividual vanabdity also one of the most socially significant by identifying two alternatives repreRecent research has advanced undersenUng contrasUng extremes of a compestanding of the cogmUve processes intence-related continuum In what we call volved In particular, Pennington and a satisficing model, the juror draws on HasUe (1993) have made an impressive the presented evidence to construct a case for then- story model of juror reasingle plausible story, omitting from fursoning Of available models, it is the ther consideration any evidence that remost elaborated and exphcit in terms of sists integration into this scenano This specifying the cognitive processes instory IS compared with representaUons volved and has the most extensive body of one or more of the verdict categones of empuical evidence to support it In If a saUsfactory match is found, that verthe present research, we extend they dict IS chosen work, focusing on the quesUon of indiAt the other end of this contmuum is vidual vanabUity Initial efforts to invesa model of opumal performance that we tigate the cogniUve processes involved m term a theory-evidence coordination juror reasoning have been centered ap- model In this case, the juror draws on propriately on characterizing in general conflictmg evidence to construct muluterms how jurors reason Could a model ple stones Different stones correspond be constructed to descnbe the cogniUve most closely to different verdict categoprocesses of any juror engaged in the nes, as Pennmgton and HasUe (1993) task'> Here, we ask whether individual have shown However, the trial evidence differences can be identified, m particu- remains differenUated fh)m the stones in lar, we are interested in differences in the sense that the juror recognizes that how well jurors reason We thus address not all of the presented evidence fits any one story For each story (and correAddress correspondence to Deanna Kuhn, sponding verdict), there exists some evDcpaitnwnt of Psychology, Teachers College, idence that IS consistent and some that is discrepant To reach a verdict choice. Columbia University, New York, NY 10027 VOL 5, NO 5, SEPTEMBER 1994

the juror must evaluate each storyverdict constellation—or theory—both against the evidence and against the altemaUves These theones are then compared, and the verdict having the most consistent and least discrepant evidence associated with it is chosen ' The contrasUng satisficmg and theory-evidence coordination models were denved jointly from consideration of the juror task itself and from ourresearchon argumentive reasoning more broadly (Kuhn, 1991) In that work, subjects were asked to descnbe and justify thcir causai theones regarding social problems such as school failure and cruiunal recidivism Substantial individual vanaUon was found, with some subjects able to reflect on theu- own views m a framework of altemaUve theones and relevant evidence Others were satisfied simply to tell a story—"this is the way it happens"—with little recognition that it might be otherwise A juror reasomng task was admmistered to this same sample of subjects, enabling us to assess whether individual differences similar to those found for argumenuve reasoning would appear for juror reasoning We assessed the relauon between reasoning m the juror task and two outcome vanables, verdict choice and the certainty attached to that choice In the argument research, subjects who reasoned more adequately—who contemplated their theones m the framework of alternatives and evidence—were less certain that these theones were correct (Kuhn, 1991) We therefore anUcipated in the juror task the similar result that more adequate reasomng would be associated with lower certainty Along the same hne, we mvestigated the possibility that more adequate reasomng—in particular, the recogmUon of altemaUve 1 Although clearly more cogmUvely demanding than the satisficing model, this model still falls shoit of the "beyond a reasonable doubt" legal standard

Copyright O 1994 Amencan Psychological Society

289

PSYCHOLOGICAL SQENCE How Well Do Jurors Reason'> possibilities and confhcting evidence— would be associated with more moderate verdict choices

Subjects The original 160 subjects who participated in the argument research (see Kuhn, 1991, for a full descnption) were selected to represent average people across the life span, begmning with adolescents (ninth graders) and including young adults in their 20s, middle adults in their 40s, and older aduhs in their 60s Two education levels were included, generally high school versus at least some college education (these differences were prospective among the adolescent group) The groups were racially and economically diverse The two older age groups were drawn from various community groups, the two younger groups were drawn from several high schools, a trade school, and a local university Adult subjects received modest flnancial compensation for their time Of the onginal 160 subjects, 152 participated in the juror task, with roughly equal distnbution across sex, age group, and education level Matenals and Procedure The jtiror interview occurred at the !nd of the second of two lengthy individual mterviews (from 45 to 90 nun each) designed to assess argumentive reasoning skills (Kuhn, 1991) It began with presentation of a 25-min audiotaped retment of portions of the Commonwealth of Massachusetts v Johnson murder tnal This matenal was developed by Penmngton and Hastie, the verused was adapted by them from a longer (3-hr) videotaped reenactment of the tnal ^ The tape includes attorneys' opemng and closing sUtements, du-ect and cross-examinations of witnesses and the defendant, and the judge's instructions to the jury Subjects also received a copy of the judge's instructions to refer to (although not the practice m real tnals) These included an explanation of 2 We thank Penmngton and Hastie for permission to use these matenals

Fig 1 Verdict categones in the Commonwealth v Johnson tnal Repnnted from Penmngton inington (1981) reasonable doubt and of the cntena defining each verdict choice The four verdicts and associated cntena appear m Figure I In subsequent analyses, we group first-degree murder (Ml) and selfdefense (SD) together as extreme verdicts and the two intermediate verdicts of second-degree murder (M2) and manslaughter (MS) as moderate verdicts These labels refer to the extremity of the cognitive judgments involved and are intended only as comparative terms ' In Commonwealth v Johnson, the defendant, Frank Johnson, is charged with Ml The undisputed background events include a quarrel between Johnson and the victim, Alan Caldwell, m a bar early on the day of CaldweU's death At that time, Caldwell threatened Johnson with a razor That evening they were at the same bar, went outside together, and got into a fight Johnson knifed CaldweU, resulting in CaldweU's death Matters under dispute include whether CaldweU puUed a razor, wheth3 A justification for this categorization is that Ml requires premeditation and malice, whereas M2 requires only the latter, similarly, SD indicates absence of guilt, whereas MS unphes at least some guilt

er Johnson actively stabbed CaldweU or merely held his knife out to protect himself, how the two men got outside together, whether Johnson intentionally went home to get his knife, and why Johnson returned to the bar After the subject listened to the Upe and was offered the opportunity to look over the wntten version of the judge's asked 1 If you were one of the jurors, what verdict would you choose*> Why'' 2 What other factors went into your decision to choose that verdicf 3 Was there any other evidence that influenced you** 4 How certain are you that the verdict you chose is the nght one'' 5 Was there anything m the tiial that suggested this was not the nght verdicf Why didn't this information lead you to rule out this verdicf 6 I'd like to ask about each of the verdicts you didn't choose What about '' Why didn't you choose this verdicf (repeated for each verdict) VOL 5, NO 5, SEPTEMBER 1994

PSYCHOLOGICAL SCIENCE Deanna Kuhn, Michael Weinstock, and Robin Flaton The first three questions were designed to ehcit as fiiU a justification as possible of the verdict choice, the purpose of the fourth was to assess certainty, the fifth assessed awareness of counterevidence or counterarguments and how these were dealt with, and the sixth explored abihty to discount alternative verdicts

Coding of Protocols The transcnbed interviews yielded protocols consisting of an lmtial, openended section (based on Questions 1-3), followed by typically shorter answers to the more structured Questions 4, 5, and 6 The fu^t step was to segment the interview matenal into codable units A unit consisted of an assertion and any accompanying justification For example, these sentences form a unit "I really think he could have escaped I don't think he was cornered that badly that he couldn't have got away " In this unit, the assertion that the defendant could have gotten away is justified by the claim that he was not cornered badly Units ranged in length from a single sentence (minimally, an assertion with no accompanying justification) to three or four sentences in which a justification was elaborated Identification of units proved rehable, with an lnterrater reliability of 90%, based on a subset of 40 randomly selected protocols A coding system was constructed and revised through an iterative procedure The final system included three dimensions the function of the umt. Its quality, and its type Rehabihty, assessed on a subset of 40 protocols, proved very high (95%) for function and quality and adequate (76%) for type The function of the unit was the purpose It was intended to serve The function of most umts was to justify the choverdict Another possible function to discount an alternative verdict Two remaining functions were to discount the chosen verdict or support an alternative verdict These appeared pnmanly when the subject was asked for counterarguments (Question 5), but they also occasionally appeared earlier when the subject was discussing testimony discrepant with a preferred verdict The quahty of each unit was coded as sufficient or deficient To be coded as sufficient, the umt had to relate legitiVOL 5, NO 5, SEPTEMBER 1994

mate and relevant evidence to one of the verdict categones To be legitimate, the evidence had to either be drawn directly fi-om testimony or constitute a reasonable inference drawn fh)m testimony To be relevant, the evidence had to either mcrease or decrease the probability that the cntena associated with the verdict were met Subjects vaned widely in the number of codable units contained in their protocols (ranging from 0 to 14, with a median of 6) and in the percentage of units assessed as sufficient (vs deficient) in quality Among 37% of subjects, all units were coded as sufficient, among another 32%, at least two thirds were coded as sufficient Among the remaining 31%, less than two thirds were coded as sufficient (Five subjects exhibited no units coded as sufficient) Each unit coded as sufficient was categonzed into one of the eight types in Table 1 Responses to Question 4 were categonzed into four categones of certainty low (e g , responses such as "not really," "not very," "not that sure," and "not at all"), medium (e g , "medium," "a htde," "sort of," and "75%"), high (e g , "fairly," "pretty," "quite," and "90%"), and very high (e g ,"very," "positive," "absolutely," and "100%")

Dimensions of Reasoning SkiU Application of the coding system to the 152 protocols yielded a data base amenable to a vanety of analyses Here, we focus on several summary dimensions most relevant to the theoryevidence coordination model The model stipulates that the endorsed theory (story-verdict constellation) be evaluated against alternative theones and against evidence To evaluate a chosen theory against alternatives means that such alternatives are recogmzed and addressed To evaluate a theory agamst evidence implies that the evidence must be reflected on and evaluated, rather than simply assimdated, making possible the recognition of evidence that does not fit the theory With respect to awareness of alternatives, the most obvious indicator is units

devoted to the function of discounting alternatives Another indicator that the theory is considered m the bght of alternatives IS the presence of counterarguments, because these are in effect alternatives to the subject's theory, whether they take the form of supporting arguments for an alternative theory or discounting arguments against the subject's theory (the two types observed) With respect to evaluation of evidence, the most obvious indicators are the judgmental types—importing and credibility (Table 1>—which occur within both supporting and discounting arguments Table 2 contains a summary of performance with respect to these key mdicators Considerable individual vanation is evident on each dimension The 26% of the sample who offered only discounting arguments in effect argued for a verdict by presenting arguments against alternative verdicts (hence failing to conform to the legal mandate that verdict cntena be established as met beyond a reasonable doubt) Among the 56% of counterargument units coded as successful, 39% consisted of spontaneously generated counterarguments, and 61% were elicited (in response to Question 6) Only 35% of counterarguments coded as successful were genuine counterarguments in the sense of being arguments against the chosen verdict The remaining 65% referred to evidence supporting an alternative verdict The protocols of most subjects who showed judgmental use of evidence also included units coded as factual or as factual and narrative No subjects showed exclusive use of the narrative type Associations among the dimensions m Table 2 provide some evidence of an overall pattern of reflection on evidence to weigh and discount alternatives SucI in counterargument was empincally related to ehcited but not spontaneous discounting Among subjects able to discount at least two of the three verdicts, 64% successfully generated a counterargument, in contrast to 47% among remainmg subjects, x^(l) = 4 42, p < 05 Where relations of other dimensions to evidence types were significant, they were significant for overall judgmental use of evidence and importing, but not credibihty, results for importing and credibihty are therefore presented separately Importing was related to 291

PSYCHOLOGICAL SCIENCE

How WeU Do Jurors Reason-^

fe 1 Coding of argument types

Supporting Noryudgmental

The subject uses testimony directly and uncnucally The subject refers to pieces of evidence drawn directly fix>m testimony, without elaborauon, to support the verdict choice Testimony is drawn on to construct a narrauve It with testimony but elaborating it

The subject reflects on the evidence, evaluating It rather than merely drawing on it uncnucally, hence enhancing its strength as supportive of the verdict choice The subject imports real-world knowledge that the evidence ts compared against,* with the outcome that the evidence is judged more supportive of the verdict choice

It was self-defense CaldweU struck Johnson first, sending him to the floor It was hke he said He [Johnson] went home with his wife and his fnend came and said let's go to this place where they usually go at times, and they went there And Johnson called him out and he figured he was going to apologize He was going to make it up, you know, patch things up, and then he [CaldweU] pulled out his razor and came forth And so he [Johnson] feared for his life and naturally put up his kmfe that he carries all the time, not just one Ume And he held it in front of him, not, you know, pointing it or menacing So, I really do believe it was self-defense because he just held the knife trying to ward him off

He intended to fight If he were intending to just talk, he could have talked tnside just as easily [verdict choice second-degree murder]

The subject evaluates the source of the evidence (i e , the witness who provided the testtmony), with the outcome that this evidence is judged more hkely to be accurate (and hence supportive of the verdict choice)

I thought that Johnson's testtmony was very clear and very lucid and it seemed there was nothing hazy about it

Notyudgmental (factual}

The subject refers to pieces of direct testimony alleged to be discrepant with the verdict being discounted

Before they went in the bar, they made sure he was not in the bar That shows he was not looking to kill the guy [discounting of first-degree murder]

Judgmental (importing)

The subject imports real-world knowledge* to support the argument that the particular evidence is discrepant with the verdict

If It actually was premeditated and he actually wanted to kill the guy, he would probably have brought something bigger and more effective than just a little fishing knife

Discounting

Discounting of evidence

The subject discounts evidence alleged to support the verdict

Judgmental (importing)

The subject unports real-world knowledge* that the evidence ts compared against, with the outcome that the evidence is judged less supportive of the verdict choice

Another witness said that he probably had Ume to put the razor back in his pocket If he has been stabbed m his heart, I don't think he would really think about putting the razor back m his pocket

Judgmental (credibility)

The subject evaluates the source of the evidence (i e , the witness who provided the testimony), with the outcome that this evidence is judged less likely to be accurate (and thus supportive of the verdict choice)

The ofiTicer said he raised the knife but his view was obstructed by the car and it was dark I don't thmk his testimony can be relied on too much

*A11 jurors, of course, draw on their real-worid knowledge to interpret testimony The dunension identified here is whether a subject does so in an expbcit way, indicating that he or she is aware of the apphcaUon of that knowledge and can reflect on its relevance

VOL 5, NO 5, SEPTEMBER 1994

PSYCHOLOGICAL SCIENCE Deanna Kuhn, Michael Weinstock, and Robin Flaton

Table 2 Individual variability tn key reasoning skills Types of performance

SkiU Spontaneous discounting of alternative verdicts Elicited discountmg of alternative verdicts

Counterargumen Judgmental use of evidence

No discounting arguments (supportive only) 16% No discounting arguments 13%

One discounting argument 35%

Not attempted 32% None 20%

Only discounting arguments 26%

Credibility 40%

Two discounting arguments 40%

Attempted but unsuccessful 12% Importing 63%

Supportmg and discounting arguments 58% Three discounting arguments 13% Successful 56% Credibibty and importing 23%

Note Each cell miry shows the percentage of the total sample in that category

three verdicts) were more likely than unspontaneous and elicited discounting but successful subjects to choose a moderate not counterargument Among subjects showing both supporting and discounting verdict—61% versus 34%, x^d) = 10 89, arguments, 81% used importing, com- p < 001 The association for spontanepared with 49% among those showing ous discounting was similarly significant, only discounting arguments and 42% with 75% of subjects showing no spontaamong those showing only supporting ar- neous discounting choosing an extreme verdict, versus 48% of remaining subguments, x^(2) = 19 89, p < 001 Among subjects largely successful in jects, x^(l) = 5 88, p < 05 The associelicited discounting (two or three ver- ation between counterargument and verdicts), 77% used importing, versus 52% dict did not reach significance Finally, associations were again significant for among the remainder, x^(l) = 9 90, p < 01 (In contrast, associations involving importing but not credibility Among subjects using importing to evaluate the credibility were not significant) There were no relationships between the di- evidence, 56% chose a moderate verdict, versus 31% among remaimng subjects, mensions in Table 2 and subjects' age group or sex, but supenor performance X^(l) = 8 4 3 , p < 01 was observed at the higher education Certainty level Subjects were more hkely to report high certainty (31%) or very high certainty (30%) in then- verdict choice than Outcome Vanables to report low (26%) or medium (13%) certainty Subjects choosing extreme Verdict The percentage of subjects choosing verdicts were more hkely than other suban exti^me verdict (Ml or SD) was 52%, jects to report very high certainty (40% whereas 48% chose a moderate verdict did so, vs 23% among subjects choosing (M2 or MS) Compared with less compe- moderate verdicts, x^[l] = 4 57, p < 05) tent subjects, subjects exhibiting more More competent reasoning was assocompetent reasoning more often chose a moderate verdict Subjects largely suc- ciated with lower certainty Subjects showing no spontaneous discounting of cessful in elicited discounting (two or VOL 5, NO 5, SEPTEMBER 1994

alternative verdicts were most likely to express very high certainty—59% did so, among remaining subjects, those who showed only discounting were least hkely to report very high certainty (12%, compared with 38% among subjects showing both supporting and discounting), x^(2) = 13 97, p < 001 Subjects successful in ehcited discounting were less likely than other subjects to express very high certainty (24%, vs 40% among unsuccessfiil subjects—a marginally sigmficant difference, x^[l] = 3 78, p < 06) Similarly, among subjects producing a successful counterargument, 25% expressed very high certainty, compared with 42% for remainmg subjects, x^d) = 4 33, p < 05 Certainty dtd not vary significantly as a function of importing or credibility

DISCUSSION The present results are consistent with Pennington and Hastie's (1993) claim that story construction is a central component of juror decision making At the same time, the results indicate significant individual vanation in the manner in which people approach the juror task This variation, moreover, is readily conceptualized along a competence dimension, considenng alternatives and reflecting on evidence is clearly more competent than not doing so In addition, the vanation has implications for task outcome To provide a qualitative sense of this vanation. Table 3 presents excerpts from the protocols of subjects at contrasting ends of the continuum we have identified Both S16 and S149 constiucted a single story of what happened (although S16 showed greater use of a narrative form), with little evidence of considenng alternatives, both were highly certain of their verdict choice (SD) In contrast, S69 treated the task as one of evaluating and weighing evidence for and against a set of alternatives, a process that left him less certain that his verdict (MS) was correct These findings have implications pertaining to the juror task and to informal reasonmg Many subjects showed inappropriately high certamty regarding their verdict choices, raising the question of how they construed the juror task In our 293

PSYCHOLOGICAL SCIENCE

How WeU Do Jurors Reason''

TaUe 3 Protocol excerpts illustrating c,

I—imUal jusUfication 2-probe

S—counterargument 6—elicited discounting of Ml

1—initial justification

2—probe 5—counterargument 6—elicited discounting of Ml

1—initial jusUfication

2—probe

4—certainty

isting approaches

among 3 subjects

S16 Because they say that Caldwell first hit him in the face and he [Johnson] fell to the floor and then Caldwell took out his razor So he (Johnson] thought he [Caldwell] would stab him, so he had to take out his fishing knife to defend himself Because Johnson was home with his wife and his two kids, and if somebody came over and asked him if he wants to go to the restaurant and have a dnnk, he says sure, because he didn't thmk Caldwell was going to be there, but he went anyway When CaldweU walked in, you know, Caldwell was very nice and fnendly and he said he wanted to talk So Johnson says sure So they went outside and when Caldwell hit him in the face, he had to defend himself WeU, they said that he did n deliberately, but in my opinion he didn't do it debberately—he did it just to protect himself He didn't want to do that because he has a wife and children to support, so he probably would not just kill a man S149 Because Caldwell was threatening him before and later dunng the day and attacked him in the evening So what he was trying to do was to defend himself from that He just walked with the knife like he was going fishing or something bke that So, since he drew out the razor from his pocket and started to you know, he was trying to defend himself so he takes a knife to defend himself Because the guy was threatening him with the razor No, not to me No, I don't thmk first-degree murder would be nght because he was trying to defend himself, so he can't be charged for first-degree murder S69 There are several things 1 find it incredible that a man who is dying would put a razor back in his pocket, and they did find it m his pocket He may never have taken it out at the Ume he was seen attacking Johnson I do believe in those circumstances As one who has spent a good part of my life on I49th Street and 8th Avenue [m New York City], I have bved through things like this I have actually seen fights among people, and I have seen the razors, and I have seen terrible slashing, and under those circumstances I would say that probably 85 to 90% of the people carry some kind of a weapon They are afraid of being mugged They are afraid of being attacked Life IS filled with fear So I can understand both of them carrying weapons The other part is that several saw Johnson bemg hit in the face, and the instinct is to respond with a weapon, especially as he knows that the other one has a weapon, which he saw earlier in the day He had been threatened earlier in the day with it I would not put It as murder in the first or second because he was m an area where he goes all the time He was there fu°st Caldwell came in afterwards Apparently, they were talking in a fairly fnendly fashion when they went out Of course, in the back of his mind might have been the fact that a battle might ensue, but there was a possibility that it would end up calmly I would not call it an out-and-out self-defense because he did return to a scene where he had been threatened earlier If he wanted to avoid violence entirely, he would have stayed away from that area He would not have been m that place, and when he was invited to go outside, he would either have refused or bterally nin away, assuming that he saw that something was about to happen Therefore I cannot say it is stncUy self-defense because when he went back to the place and when he went outside, there was a possibility of a physical confrontaUon I do beheve that the main attack came from Caldwell, in spite of the fact that he [Caldwell] did not have his razor out, because he was seen whatever was seen the first violence was his punching Johnson IS69 at this point offers arguments against the MI cntenon of malice, focuses on testimony that Johnson was a peaceful man and that no comparable testimony was presented regarding CaldweU, and repeats an argument against self-defense, regarding the plausibility of a dying man returning a razor to his pocket} I do not believe that CaldweU ran onto the knife I bebeve he was stabbed, but I bebeve that this was because Johnson had been attacked and the instinct is to use whatever weapon you have, knowmg that the other one has a we^xm With all the evidence, we reaUy do not know who was the real aggressor outside It might have been Johnson, but there is no visual evidence of that

*See the Method secUon for the fiiU text of the quesuons Ml = first-degree murder

VOL 5, NO 5, SEPTEMBER 1»4

PSYCHOLOGICAL SCIENCE

Deanna Kuhn, Michael Weinstock, and Robui Flaton argument research, many of these same subjects saw complex questions, such as why pnsoners return to crime, as having single, certain answers (Kuhn, 1991, chap 7) The high certainty expressed in the juror task suggests a sumlar absolutism, with the task conceived of as identifying a smgle, certain truth, rather than weighing alternatives whose truth can never be known with certamty People who construe the juror task as one of identifying a single truth wdl be less disposed to contemplate the evidence in a framework of alternatives Instead, they are likely to search for a plausible scenano of "what happened"—the story cential to Pennmgton and Hastie's (1993) model—and to seek evidence allowing them to attach certainty to this story This approach establishes optimum conditions for the biased assinulation (of evidence) prevalent m research on reasoning (Baron, 1988, Kunda, 1990) The associations reported here between reasoning and verdict choice are not of a magmtude to make the verdict predictable from pnor assessment of jurors' reasoning skill Yet such associations are cntical in establishing that the quality of reasoning a juror engages in is relevant to task outcome Why might less competent reasoners—those toward the satisficing (single-theory) end of the continuum we have identified—be more likely to choose extreme verdicts'' Verdict cntena in most legal tnals are complex and poorly understood by a majority of jurors (Smith, 1991) As a result, the narrative, storyhke features of a verdict may be more salient than the legal cntena that define it •* Exti-eme verdicts (Ml and SD) may be chosen more readily because it is easier to construct Itones corresponding to these verdicts than It IS to construct stones corresponding to the moderate verdicts '

Choice of one of the two mtermediate verdicts, in contrast, implicates discounting of altemaUves It is difficult to envision the choice of M2 without first considermg and ruhng out Ml In contrast, one can readily envision construction of an "Ml story" in the absence of any consideration of alternatives Consistent with this interpretation are the associations we found between verdict choice and discounting and between verdict choice and importing, with subjects choosing the extreme verdicts more likely to simply draw on the evidence (to construct a story), rather than evaluate it cntically An additional factor favonng extreme-verdict stones is the fact that these stones are to a considerable extent constructed for the juror in the respective attorneys' opening and closing statements (in contrast to the moderate verdicts, which are represented for the juror only in the judge's instructions) This typical difference in attorneys' treatment of extreme versus intermediate verdicts in muitiple-verdict tnals may interact with and amplify preexisting proclivities of some jurors toward the sattsficing form of processing Do the contrasting forms of processing complex information identified here extend beyond the juror task'' Although limited space precludes descnption of the vanous dimensions of reasoning skill assessed in our argument research (Kuhn, 1991), we did observe relations between subjects' performance in the juror task and their performance on those measures, in particular, those dimensions most directly related to contemplation of theones in a framework of alternatives This association with argumentive reasoning more broadly is important in establishing that we are dealing with more than task-specific vanation and makes clear the relevance of the present results to the growing study of informal reasomng (Galotti, 1989, Kuhn, 1993, This IS likely to be true for discounted as Rogoff & Lave, 1984, Smnott, 1989, Stemberg & Wagner, 1986, Voss, Perwell as supported verdicts, as illustrated by S16, who discounted Ml on the grounds that kins, & Segal, 1991) That hterature has the defendant had a wife and children to sup- been consistent in revealing significant port (Table 3) individual variability Yet a long-stand5 It IS also, of course, easier to discount ing issue m the study of reasomng has :treme verdicts (because conflicung evi- been whether it is warranted to idendence IS more readily available), but subjects tify some forms of reasonmg as lnfenor the sausfking (single-story) end of the conluum we have identified are less likely than to others (Cohen, 1981), especially when more competent subjects to engage m dis- one takes into consideration the trade-off between effort and gain Perhaps satiscounting of altemabves VOL 5, NO 5, SEPTEMBER 1994

ficing forms of reasoning arc m fiact more rational than cognitively effortful ones that reqture contemplation (rf all possibilities (Baron, 198S, 1988, Moshman, in press) Here, however, we have presented a case in which the exti^ processing is without question worthwhile and the more cogmtively complex way of executing the task a superior one, the approach we would want anyone engaged in the task to adopt Having established that this cogmtive vanation influences task outcome, we have demonstrated a respect in which how, and how well, people reason makes a tangible difference to society If a juror settles on a verdict of Ml or SD without ever having considered the mtermediate alternatives laid out in the judge's instiiictions, the tnal process has been undermined If, following presentation of testimony, jurors enter into deliberation with an lnappropnate conception of the task as one of identifying a single truth, and with an accompanying inappropnately high certainty in their own predehberation verdict choices, the deliberation process will be adversely affected On the positive side, the jury process was designed to remedy deficits of just this sort Other jurors will have made initial choices of different verdicts and will interpret the evidence in different ways Social exchange with these other minds may thus go a long way in providmg the corrective to what we have found missing in the thinking of individual jurors It could prove fruitful to study jury process from this perspective The compensatory role that individual jurors may play relative to one another in the social context of the jury has been noted with respect to retention of information from the testimony (with other jurors likely to remember what one has forgotten) This compensatory function may occur as well at the deeper level of how testimony IS represented and used, rather than merely whether it is represented and retained More broadly, and speculatively, the present findings raise the possibility of juror education du-ected toward the goal of improving juror reasomng Our research on argumentive and inductive reasoning more broadly (Kuhn, 1989, 1991, Kuhn, Schauble, & Garcia-Mila, 1992) suggests that exercise—m contem295

PSYCHOLOGICAL SOENCE

How Well Do Jurors Reasoti''

plating alternatives and evaluating evidence--may be sufficient to induce change, in the absence of more direct instruction This raises the possibility of a juror "learning curve," with more expenenced jurors performing more competently—a possibihty open to empuical mvestigation The associations observed between education level and reasoning in our own and other people's work on argumentive reasonmg (Kuhn, 1991, Perkins, 1985, Voss et al , 1991) do not offer a great deal of insight into process, but they do suggest that some of the broad expenential variables associated with higher education levels are relevant to the development of informal reasoning skill Individual cognitive processes are only one component of the jury process They are nonetheless a cntical component, without their contnbution, no social exchange among jurors could occur The present work helps to establish that the skill with which these cogmtive processes are executed is of cntical significance to the jury process It IS faith m the

power and ultimate tnumph of reason that makes societies willmg to entrust an individual's fate to the collective reason of peers This widely shared faith makes understanding these reasoning processes a partictilarly important goal

REFERENCES Baron, J (1985) Rationality and mtelUgence New York Cambndge University Press Baron, J (1988) Thinkmg and deciding New York Cambridge University Press Cohen L J (1981) Can human lmttionality be expenmentally demonstrated'^ Behavioral and Bram Sciences 4 317-331 Galotu K (1989) Approaches to studying formal and everyday reasoning Psychological Bulletin 105 331-351 Kuhn D (1989) Children and adults as intuiUve scientists Psychological Review, 96 674-689 Kuhn,D (1991) The skills of argument New York Cambndge Umversity Press Kuhn, D (1993) Connecting scientific and informal reasoning Merrill-Palmer Quarterly [Special issue on rationahty and cntical thinking] i9(l), 74-103 Kuhn D Schauble, L , & Garcia-Mila, M (1992) Cross-domain development of scientiTic reasoning Cognition and Instruction 9 285-327

KuDda Z (1990) The case for motivUed reasooing Psychological BuUetin 108 480-498 Moshman, D On press) Reason, reasons, and reasoning A constnictivist account of human raUonahty Theory and Psychology n, N (1981) Causal reasoning and decimahng The case ofjuror decisions Un- I published doctoral dissertation. Harvard University, Cambndge, MA Penmngton N , & Hastie, R (1993) The story model for juror decision making In R Hastie (Ed ), Inside the juror (pp 192-221) New York Cambndge University Press Perkins, D (1985) Postpnmary education has bttle unpact on informal reasoning Journal of Educational Psychology, 77, 562-571 Rogcrff, B &Lave J (Eds) (1984) Everyday cogmtion Its development in social context Cambridge, MA Harvard University Press Sinnott,J (1989) Everyday problem solvmg Theory and applications New York Praeger Smith V (1991) Prototypes m the courtnxim U y representations of legal concepts Journal of Personality and Social Psychology 61, 857872 Stemberg, R & Wagner, R (Eds) (1986) Practical intelligence Nature and origins of competence in the everyday world New York Cambndge Umversity Press Voss, J , Perkins D & Segal, J (Eds ) (1991) Informal reasoning and education HiUsdale NJ Eribaum (RECEIV

3 8/4/93, R E I

i ACCEPTEl

3/2/94)

VOL 5, NO 5, SEPTEMBER 1994