28
Law and Human Behavior, Vol. 22, No. 3, 1998 A Study of Juror and Jury Judgements in Civil Cases: Deciding Liability for Punitive Damages Reid Hastie,1-4 David A. Schkade,2'4 and John W. Payne3-4 A study was conducted to investigate civil juries' decisions concerning defendants' liability for punitive damages in tort cases. A total of 121 six-member mock juries composed of jury-service-eligible citizens were presented summaries of previously decided cases and given a comprehensive instruction on the defendant's liability for punitive damages. Most of the mock juries decided that the consideration of punitive damages was warranted, although appellate and trial judges had concluded that they were not warranted. The tendency to find the defendant liable was partly due to jurors' failure systematically to consider the full set of legally necessary conditions for the verdicts they rendered. Individual differences in the jurors' backgrounds were not strongly related to their verdicts; income and ethnicity were weakly related to judgments. The social processes in deliberation on civil juries were similar to the dynamics of deliberation that have been observed in criminal juries. The study of civil-jury decision making is still in its infancy (Hensler, 1988; Litan, 1993; MacCoun, 1989; Saks, 1992). How do individual jurors comprehend evidence and instructions on the law, and reason to a verdict? What are the dynamics of the jury deliberation process? Are there systematic habits of judgment or biases in reasoning that might produce departures from just and proper verdicts in civil cases? Although we know little about the decision processes in civil juries, the out- comes of civil jury decisions have received considerable scholarly and public scrutiny during the past decade, with many observers concluding that the system is in a state of crisis (e.g., Huber, 1988; Mahoney & Littlejohn, 1989; and see reports in Adler, 1994; Daniels & Martin, 1986). Other policy makers and scholars have re- sponded to the apparent crisis with arguments and evidence that the system, at least "on the average" or in comparison to the practical alternative of bench trials, is working properly and that the data cited by critics who are concerned about excessive verdicts are incomplete or of limited generality (Clermont & Eisenberg, 'Psychology Department, University of Colorado, Boulder, CO 80309. 2Department of Management, University of Texas, Austin, TX 78712. 3Fuqua School of Business, Duke University, Durham, NC 27708. 4Correspondence may be addressed to any of the authors. 287 OI47-7307/98;0<iOO-0287$15.00/l @ 1998 American Psychology-Law Society/Division 41 of the American Psychological Association

A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

Embed Size (px)

Citation preview

Page 1: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

Law and Human Behavior, Vol. 22, No. 3, 1998

A Study of Juror and Jury Judgements in CivilCases: Deciding Liability for Punitive Damages

Reid Hastie,1-4 David A. Schkade,2'4 and John W. Payne3-4

A study was conducted to investigate civil juries' decisions concerning defendants'liability for punitive damages in tort cases. A total of 121 six-member mock juriescomposed of jury-service-eligible citizens were presented summaries of previouslydecided cases and given a comprehensive instruction on the defendant's liability forpunitive damages. Most of the mock juries decided that the consideration of punitivedamages was warranted, although appellate and trial judges had concluded that theywere not warranted. The tendency to find the defendant liable was partly due to jurors'failure systematically to consider the full set of legally necessary conditions for theverdicts they rendered. Individual differences in the jurors' backgrounds were notstrongly related to their verdicts; income and ethnicity were weakly related to judgments.The social processes in deliberation on civil juries were similar to the dynamics ofdeliberation that have been observed in criminal juries.

The study of civil-jury decision making is still in its infancy (Hensler, 1988; Litan,1993; MacCoun, 1989; Saks, 1992). How do individual jurors comprehend evidenceand instructions on the law, and reason to a verdict? What are the dynamics ofthe jury deliberation process? Are there systematic habits of judgment or biases inreasoning that might produce departures from just and proper verdicts in civilcases? Although we know little about the decision processes in civil juries, the out-comes of civil jury decisions have received considerable scholarly and public scrutinyduring the past decade, with many observers concluding that the system is in astate of crisis (e.g., Huber, 1988; Mahoney & Littlejohn, 1989; and see reports inAdler, 1994; Daniels & Martin, 1986). Other policy makers and scholars have re-sponded to the apparent crisis with arguments and evidence that the system, atleast "on the average" or in comparison to the practical alternative of bench trials,is working properly and that the data cited by critics who are concerned aboutexcessive verdicts are incomplete or of limited generality (Clermont & Eisenberg,

'Psychology Department, University of Colorado, Boulder, CO 80309.2Department of Management, University of Texas, Austin, TX 78712.3Fuqua School of Business, Duke University, Durham, NC 27708.4Correspondence may be addressed to any of the authors.

287

OI47-7307/98;0<iOO-0287$15.00/l @ 1998 American Psychology-Law Society/Division 41 of the American Psychological Association

Page 2: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

1992; Daniels & Martin, 1995; Lempert, 1993; Saks, Hollinger, Wissler, Evans, &Hart, 1997; Schuck, 1993; Vidmar, 1995).

What is incontrovertible is that in a nonnegligible number of cases juries ren-der distinctively harsh antidefendant judgments and set dramatically large awards.Virtually all of these extreme verdicts are substantially reduced or reversed by ap-pellate courts; the reduction rate is over 50% for many classes of awards (e.g.,American Law Institute, 1991; Broder, 1986; Daniels & Martin, 1990; Landes &Posner, 1986; Ostrom, Hanson, & Daley, 1993; Peterson, Sarma, & Shanley, 1987).These observations imply that, under some conditions, trial juries render verdictsthat are consistently discrepant from legal principles and precedents.

One focus of the controversy over the performance of the civil jury is onpunitive damage awards. Historically, punitive damages were introduced into thecommon-law tort system to punish and deter conduct that approached criminal lev-els of egregiousness and maliciousness. At first punitive damages were rarelyawarded, but since the late 1970s, juries have imposed punitive damages with in-creasing frequency and in escalating amounts (the largest punitive damages awardin the history of civil jury trials was the $5,000,000,000 award assessed against theExxon Corporation in 1994, following the Exxon Valdez oil spill).

In a typical jury trial in which punitive damages are sought, there is an initialdetermination of liability with three possible outcomes: the defendant is not liablefor any damages; the evidence demonstrates the defendant's conduct was negligentand only compensatory damages may be warranted; or the defendant's conduct wasmore than negligent and reached levels of reckless or malicious conduct and bothcompensatory and punitive damages may be warranted (Ghiaridi & Kircher, 1995;Roeca, 1984). The legally necessary conditions that must be established to concludethat punitive damages are warranted identify conduct that approaches criminal lev-els of reprehensibility (indeed, some states do not allow punitive damages in civiltrials, but instead apply criminal sanctions to the same types of misconduct). Atypical set of requirements to conclude that a defendant's conduct was recklessenough to warrant punitive damages includes subjective consciousness of a fore-seeable and probable grave danger consequent on the conduct, conscious disregardfor the danger by the defendant, conduct that was a gross deviation from an ordi-nary level of care, and the occurrence of the dangerous outcome (Eades, 1993).The application of these instructions requires jurors to make several difficult, sub-jective judgments of intangible conditions.

The Supreme Court appears to be trying to develop a uniform, principleddoctrine for the assessment of punitive damages. Some justices are concerned thatcurrent procedures do not provide sufficient guidance and control of jury verdicts(e.g., O'Connor dissenting in Pacific Mutual Life Insurance Co. v. Haslip, et al, 1991;Kennedy concurring, and O'Connor, White, and Souter dissenting in TXO Prod.Corp. v. Alliance Resources Corp., et al., 1993; Steven's lead opinion in Honda MotorCo., Ltd., et al, v. Karl L. Oberg, 1994; and Breyer concurring in BMW of NorthAmerica, Inc., v. Ira Gore, Jr., 1996). The explicit goals of punitive damage awardsare to deter and punish the defendant's conduct; thus, they are a mixture of civiland criminal law traditions. Jurors can be expected to have a variety of interpre-tations of these concepts and to exhibit considerable variability in their expression

288 Hastie, Schkade, and Payne

Page 3: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

of these concepts on monetary damage scales. The jurors' decision problem is com-pounded by the lack of clear instructions on these concepts or the provision ofexamples or standards for award amounts (the situation seems even less controlledthan that confronting jurors when they are presented with criminal sentencing de-cisions; Brown, 1996; Luginbuhl, 1992; Wiener, Pritchard, & Weston, 1995). JusticeStevens writing a majority opinion commented on the Court's primary concernabout punitive damage awards: "The problem that concerns us ... is that a jurywill not follow those instructions and may return a lawless, biased, or arbitrary ver-dict" (Honda Motor Co. v. Oberg, 1994, p. 2341). A series of relevant decisions hasaddressed the issues of prescribing fair jury trial procedures and providing concep-tual tests to determine when awards are excessive (most notably Pacific Mutual Lifev. Haslip, [1991; BMW of North American v. Gore, 1996; Honda Motor Co. v. Oberg,1994; and TXO Prod. Corp. v. Alliance Resources Corp., 1993). However, no uniformpolicy has yet emerged.

In recent written opinions, the Court has focused on procedural issues con-cerning the extent to which the trial jury's decision adheres to rational or consti-tutional principles of due process. Do jurors and juries consider the legallyprescribed conditions and factors when they determine liability for punitive damagesand set awards? Are unwarranted or excessive verdicts common? Can jury instruc-tions be designed that will properly constrain and guide jury decisions?

The present research is designed to study the jury's decision of whether adefendant has engaged in malicious or reckless conduct that would warrant theconsideration of a punitive damage award. We have selected the punitive damagesaward decision to study because it is controversial and there is some evidence thatjudgments by jurors and by higher authorities (i.e., appellate courts) are discrepant(Finkel, 1995; Robinson & Darley, 1995). But our primary goal is to sketch aninitial picture of the juror and jury cognitive and social processes when making atypical serious civil decision.

We study the behavior of jury-eligible citizens making decisions in realistic,representative civil cases that turn on the question of whether a defendant's conductis reckless or not. We use a mock-jury simulation method that has proven usefulin studies of jurors' decision processes and in providing accurate projections to theoutcomes of real jury trials (see Bray and Kerr, 1979, Hastie, Penrod, and Pen-nington, 1983, and MacCoun, 1989 for explication and defense of the method). Weshould note two differences between our methods and the procedures in actualtrials: First, our mock jurors did not first decide on compensatory damages awardsbefore deciding on punitive damages, as they usually would in trials that involvepunitive damages. Second, our case materials were veridical summaries of the evi-dence from the original trials written in an expository format, not in the form ofa mock trial with judge, attorney, and witness speaking roles (see example casematerials in the Appendix).

The four stimulus cases were based on cases that had been subjected to ju-dicial review, so we can assess the extent to which our mock jurors' verdicts matchthe legally endorsed verdicts. Second, there are many suggestions that civil jurors'verdicts are more predictable than criminal jurors' from characteristics such as in-come and political orientation (Hastie, 1991). Which background characteristics of

Liability Decisions 289

Page 4: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

jurors predict their verdicts? Are there any general predictors that exceed the 10%"variance accounted for" in verdicts that is usually observed in studies of criminaljury decisions (Hastie, et al, 1983; Penrod, 1979)? Third, we will examine jurors'self-reported reasons for their verdicts and the contents of their discussions in de-liberation to shed light on their mental and social decision processes. Because wehave the conclusions from judicial reviews, we will attempt to identify the differ-ences in processes that discriminate between juror decisions that are consistent andinconsistent with the decisions of judges.

METHODS

Participants

Participants, jury-eligible adult citizens, were sampled from the Denver areaby a marketing research firm and paid $50 for participation in the study. The par-ticipants were sampled representatively by residential address and contacted bymail. Approximately 85% of those contacted agreed to participate in the study of"lifestyles and opinions." A final sample of 726 deliberating mock jurors was ob-tained with the compositional statistics summarized in Table 1.

Materials

Four legally significant cases that required jurors to distinguish between neg-ligent and reckless conduct were selected as the basis for stimulus materials (An-derson v. Whittaker Corp., 1987; Harper v. Zapata Off-Shore Co., 1983; Jardel Co.Inc., v. Hughes, 1987; and In re Marine Sulphur Queen, 1972). These cases wereselected because they represented a variety of fact situations, because they are fre-quently cited as precedents, and because the proper action on the issue of punitivedamages had been decided as a matter of law by trial or appellate court review of

290 Hastie, Schkade, and Payne

Table I. Demographic Characteristics of the Participants

Sex45% men55% women

AgeMedian = 47 years (range: 25-73)

Race/ethnicity70% White (European American)15% African American12% Hispanic3% Asian American, Native American, and other

Income (annual family)Median = $40,000 (approx.)(6% less than $15,000; 10% over $75,000)

EducationMode = high school diploma + vocational(1% no high school diploma;23% college degree plus vocational or professional)

Page 5: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

Liability Decisions 291

Table 2, Summary of Stimulus Cases

Case Court, Action on punitivename Case description year damages

Anderson

Harper

Jardel

Queen

Four boaters lost in defectivepower boat after inadequaterecallInjured seaman denied part ofmaintenance pay, after hiringlawyerShopping mall owner failed toprevent abduction and rape ofemployee in parking lot39 seamen lost in dangerouslydesigned and operated moltensulfur carrier

DistrictW.D. Mich.1987/88Fifth Circuit1984

Supreme Courtof Delaware 1987

Second Circuit1970/72

Denied

Reversed jury verdict

Reversed jury verdict

Affirmed denial

the original proceedings. In each case, the trial court or a higher court establishedthat the facts did not warrant the consideration of action on punitive damages; twoof the cases were originally tried by juries and two were "bench trials," althoughall could have been tried by juries if the parties had not waived that right. (Table2 provides a summary of the case citations and dispositions on the issue of punitivedamages). Written summaries of the facts were prepared, ranging in length fromapproximately 1,000 to 1,500 words (exclusive of the judge's instructions). Care wastaken to adhere to the courts' summary statements as closely as possible; thus thecase summaries are composed mostly of excerpts from the court opinions.

Jury instructions (approximately 500 words long) on liability and on proce-dures such as the standard of proof, the unanimity quorum requirement, and theselection of a presiding juror were excerpted from In re The Exxon Valdez, whichin turn were derived primarily from the instructions in Jardel Co. Inc. v. Hughes.The instructions provide a detailed prescription of the standards for finding "reck-less disregard" and they are consistent with currently accepted standards for de-ciding this issue (see Appendix). The instructions include definitions of maliciousconduct and reckless or callous disregard for others, and carefully specify the dif-ferences between negligent and reckless conduct.

The case summaries and instructions were used as scripts for videotapes rep-resenting the four cases (an example case, Jardel Co. Inc. v. Hughes, is included inthe Appendix, copies of all case materials are available from the authors). One ofthese summaries served as the stimulus case for each mock jury. Jurors watchedthe videotape (durations ranged from 11 to 15 min) and then read a written sum-mary before beginning deliberation.

Procedure

The study was conducted in a conference hotel located in a Denver suburb.Participants registered with the researchers and went to a waiting area until the formalstudy began. Then, participants assembled in an auditorium (in groups ranging in sizefrom 38 to 52 participants) and were introduced to their mock-jury task. They com-

Page 6: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

pleted a short background information questionnaire and were instructed that theywould be asked, "To make a legal decision just like the ones that jurors make in legaltrials." Then, they were shown one of the videotaped stimulus cases on a projectionscreen with an image approximately 2 x 2 m in size (participants' seating placed themfrom approximately 3 m to 15 m from the screen). They were told that the judge'sinstructions on the tape were the law and that they should adhere to them in makingindividual and group decisions. They were provided with a written copy of the to-be-judged case including jury instructions, note pads, and pencils. After the videotape,they were given 10 min to study the written script and to take handwritten notes.Then they completed a questionnaire indicating their individual predeliberation ver-dicts by checking "Yes," proper to award punitive damages; "No," not proper to awardpunitive damages, or "Undecided." They were also asked to indicate their confidencein the decision on a 6-point scale labeled from "50%, No idea I am just guessing,"to "100%, Absolutely certain my verdict is correct."

Following the trial presentation, the mock jurors were grouped into six-personmock juries and went to conference rooms to deliberate. They were provided witha copy of the judge's instructions and reminded that the judge had instructed themto select a presiding juror and that their verdict must be unanimous. Their delib-erations were videotaped, with their knowledge and permission (one videotape waslost due to an equipment malfunction and that jury is excluded from content analy-ses of the deliberation process). They completed deliberation, with a verdict or bydeclaring their jury deadlocked. (Four of the 121 juries were unable to reach aconclusion after 2 hr of deliberation, but did not declare themselves deadlocked.Their deliberations were terminated by the experimenters and they will be classifiedas "Hung" in our analyses.) After concluding deliberation, jurors were asked tocomplete a second questionnaire that asked for their personal verdicts, probed thereasons for their judgments, and tested their memories for information from thejudge's instructions (copies of the original instructions were not available at thistime). Then jurors were paid for their participation and excused.

292 Hastie, Schkade, and Payne

RESULTS

Verdicts

Table 3 provides a summary of the individual verdict preferences measuredbefore and after deliberation. The 95% confidence intervals on the proportions of"Yes" verdicts are listed for the individual pre- and post-deliberation verdicts. Table4 presents the group jury verdicts. The best estimate of the rates at which the mockjuries rendered, "Yes, liable for punitive damages," verdicts is based on the pro-portions of "Yes" juries out of the total number of verdict-rendering juries, exclud-ing undecided juries. The assumption is that undecided juries would tend to splitaccording to the proportion represented by verdict-rendering juries or, should thecases be retried, decide at the rate estimated by the other verdict-rendering juries.Thus, we have calculated 95% intervals around the proportions of "Yes" jury ver-dicts with the "undecided" verdicts excluded from the totals. The rate at whichindividuals and juries favored "Yes" verdicts is far above zero for all cases; over

Page 7: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

Liability Decisions 293

Table 3. Juror Verdicts

Anderson Harper Jardel MSQ Overall

Predeliberation individual juror verdictsYes

No

Undecided

95% confidenceintervals(around "Yes")

62% 53%(116) (98)30% 33%(56) (62)8% 14%(14) (26)

55%-69% 46%-60%

59%(99)33%(55)8%(14)

52%-66%

76%(142)17%(31)7%(13)

70%-82%

63%(455)28%(204)9%(67)

60-67%

Postdeliberation individual juror verdictsYes

No

Undecided

95% confidenceintervals(around "Yes")

52% 67%(97) (124)43% 32%(83) (59)3% 1%(6) (3)

45%-59% 60%-74%

60%(101)39%(65)1%(2)

53%-67%

81%(150)16%(30)3%(6)

75%-87%

65%(472)33%(237)2%(17)

62%-69%

Table 4. Jury Verdicts

Jury verdictsYes

No

Undecided

Anderson Harper

35% 68%(11) (21)42% 32%(13) (10)23%(7) (0)

Jardel

57%(16)29%(8)14%(4)

MSQ

71%(22)13%(4)16%(5)

Overall

58%(70)29%(35)13%(16)

Jury verdicts, hung verdicts excludedYes

No

95% confidenceintervals(around "Yes")

46% 68%(11) (21)54% 32%(13) (10)

26%-66% 52%-84%

67%(16)33%(8)

48%-86%

85%(22)15%(4)

71%-99%

67%(70)33%(35)

58%-76%

50% of individual jurors favored "Yes" after deliberation, and the "Yes" jury verdictrates were over 50% for three of the four cases. Recall that, in each case, a trialcourt or appellate review concluded that the facts did not warrant a finding of reck-less or malicious conduct.

Relations between Individual Differences and Verdict Preferences

Simple correlation coefficients were used to explore relationships betweenbackground factors such as participants' sex, age, education, income, and compre-

Page 8: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

hension of the judge's instructions and their predeliberation and postdeliberationverdicts and confidence ratings. Simple probability levels associated with signifi-cance tests were calculated and the p < .01 level was applied to select relationshipsfor further consideration. No effort was made to correct for multiple tests to esti-mate "true" family-wide probabilities (e.g., a Bonferroni inequality adjustment;Snedecor & Cochran, 1980). Thus, the simple probabilities doubtless overestimatethe reliability of individual relationships. This procedure was followed to avoid con-fusion over exactly which family of variables was the basis for an individual adjust-ment; we believe the simple probabilities are easier to interpret in such cases andno attempt to "inflate" the number of significant relationships reported is intended.

Two measures of verdict preferences were used: (i) the simple verdict selected("no" the defendant was not liable, coded as "1" and "Yes" the defendant was liablefor punitive damages, coded as "2"; when this measure was used in analyses, undecidedjurors were excluded from the calculations); and (ii) a score ranging from 0 ("No" thedefendant was not liable with a confidence of 100%) to 100 ("Yes" the defendant wasliable with a confidence of 100%); a confidence rating of 50% with either verdict wasassigned a score of 50 and undecided jurors were included in the analysis and assigneda score of 50. Some data analysts would use a logistic regression model, with a logistictransformation applied to the binary-valued dependent variable, rather than the ordi-nary least squares models we used in our analyses. However, following Neter, Wasser-man, and Kutner (1983, p. 361) we believe that there is little to be gained (andsomething to be lost in comprehensibility) from the logistic approach if the mean re-sponses range between .20 and .80, as they do in our results.

Combining the four cases, jurors' income predicted verdicts (r = -.10, p < .01for the unweighted predeliberation verdict measure), with higher income participantslikelier to decide "Not liable." Effects of the individual difference factor of ethnicbackground or race were difficult to assess because of the small sample of non-Whitejurors. Rather than attempting to make fine distinctions among racial categories,analyses were conducted to compare the majority White jurors with minority jurors(primarily African-American and Hispanic jurors). White jurors were likelier to favora "No liability" verdict than minority jurors (predeliberation verdict: r = -.13, p <.01; confidence weighted predeliberation verdict: r = -.10, p < .01). Exploratory re-gression analyses were also performed and the results supported the correlation co-efficient summaries. For example, the most sensible equation to predict the confidenceweighted predeliberation verdict score on a 0-100 scale was: -1.27 (income) - 6.30(race) + 72.66; R2 = .015; F(2, 715) = 5.31, p < .01, with both coefficients and theconstant significant at the p < .05 level. The equation to predict the simple unweightedverdicts (1 = not liable, 2 = liable, with undecided jurors excluded from the analysis)was -02(income) — .12(race) + 1.88; (R2 = .024; F(2, 649) = 8.12,p < .001, withboth coefficients and the constant significant at the p < .05 level.

Comprehension of the Judge's Instructions

Participants were asked to recall portions of the judge's instructions on liabilityby means of specific questions on each important element relevant to their decisions

294 Hastie, Schkade, and Payne

Page 9: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

(e.g., "What is the legal definition of 'reckless or callous disregard for the rightsof others?' [If you cannot remember the legal terms, try to state this definition inyour own words]"). Written responses to these questions were "scored" based onaccuracy of recall of nine elements or legal distinctions from the instructions. Tworesearch assistants were trained to code the responses in terms of lenient criteriain which recall of the gist of each element was scored correct if paraphrases pre-served the approximate meaning of the original instruction. These scores were con-verted to a percent correct metric with possible scores ranging from 0% to 100%correct. Intercoder reliability was high as assessed via a simple correlation betweencodes for 48 double-coded questionnaires (r = +.86); for the 48 questionnaires,90% of the paired scores were within 5% of one another for the two coders.

The overall level of performance on this recall-comprehension test was low:The median score was 5% correct (the mean was 9%); 30% of the participantsreceived a score of 0 correct and the high score was 67% correct. These low scorescannot be attributed to nonresponses due to low motivation to comply with thedecision task; there were almost no blank response forms and the 0 comprehensionscores were usually assigned to lengthy, but incorrect responses. A simple wordcount of the contents of the responses on the memory-comprehension test yieldeda range of from 0 words to 167 words; there were only 16 jurors (2%) whose re-sponses were blank (0 words); and the median number of words written was 43(the mean was 41).

Several individual difference background factors predicted comprehension ofthe instructions: juror age (younger jurors' comprehension was better: r = -.24, p< .001), juror education (better educated jurors' comprehension was better: r =+.32, p < .001), juror income (higher income jurors' comprehension was better: r= +.16, p < .001), and juror ethnicity (White jurors' comprehension was betterthan minority jurors: r = +.14, p < .001). Exploratory regression analyses werealso performed and the results supported the correlation coefficient summaries. Forexample, the most sensible equation to predict individual comprehension levels ona 0-100 scale was: -0.25(age) + 3.34(education) + 3.26(race) + 7.21; R2 = .19;F(3, 705) = 54.98, p < .001, with all coefficients and the constant significant atthe p < .001 level.

The instruction recall-comprehension score was significantly correlated withthe confidence weighted predeliberation verdict preference for only the Andersoncase (/• = -.21, p < .01; r = -.23, p < .01 for the simple unweighted verdict pref-erence with undecided jurors excluded from the analysis). Participants who werebetter able to comprehend and recall the instructions were also likelier to decide'•not liable."

The Presiding Juror Role

There was a tendency for the juror selected to preside over deliberations topossess "dominant" social characteristics: the presiding juror was more likely to bemale (r = -.19, p < .01) and to be better educated (r = +.13, p < .01). Exploratoryregression analyses were also performed and the results supported the correlation

Liability Decisions 29S

Page 10: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

coefficient summaries. For example, the most sensible equation to predict who wouldbe elected foreperson {0-1 scale) was: -.13(sex) + .04{education) + .22; R2 = .047;F(2, 723) = 17.85, p < .001, with all coefficients and the constant significant at thep < .01 level. The presiding juror also tended to have a better memory for the in-structions (r — +.11, p < .01). Presiding jurors were slightly likelier than nonpresidingjurors to favor a "No liability" verdict before deliberation (predeliberation verdict, r= -.10, p < .01; confidence weighted predeliberation verdict, r = -.08, p < .04; thesecorrelations were only significant when all four stimulus cases were combined in theanalysis, although the pattern of correlations was the same for each individual case).

Did the presiding juror have a greater influence over the outcome of delibera-tion than the nonpresiding jurors? One answer to this question is provided by com-paring the degree to which the presiding juror's predeliberation verdict preferencepredicted the jury verdict compared to the predictive power of nonpresiding jurors'predeliberation preferences. The correlation between the individual jurors' predelib-eration verdict and the group verdict (excluding undecided jurors and hung juriesfrom the analysis) was +.29 for presiding jurors and +.26 for nonpresiding jurors.This difference is not large or statistically significant, implying that presiding jurorsdid not play a specially decisive role in determining the outcome of deliberations.

Predicting Who Will Change during Deliberation

A change score was calculated to indicate which jurors changed opinions onthe verdict during the course of deliberation. Jurors who were undecided beforeor after deliberation were excluded from the analysis. Overall, 23% (153 of 659)of the jurors who had reached definite verdicts before deliberation changed verdictsduring deliberation. No distinction was made between jurors changing from "Yes"to "No" (56% or 85 of the 153 who changed) and from "No" to "Yes" (44% or68 of the 153 who changed) in our analysis of predictors of change.

When the analysis was conducted by combining data from all of the stimuluscases, the jurors' predeliberation confidence that their verdicts were correct wasthe primary predictor of change during deliberation (correlation coefficients relatingthe change/no-change variable to juror background variables will be reported here:r = -.19, p < .01, indicating more confident jurors were less likely to change).Annual family income (r = -.10, p < .014) was a marginally significant predictorof change. However, when the same analyses were conducted for the four stimuluscases separately, only predeliberation confidence was a significant or marginally sig-nificant predictor for each of the cases. One other relationship emerged at theindividual case level: juror age was significantly correlated with whether or not thejuror changed during deliberation on the Harper case only (r = -.22); older jurorswere less likely to change than younger jurors.

The "Dynamics" of Jury Deliberation

Jury deliberation times ranged from 5 min to 99 min, with an average timeof 43 min (median of 42 min). Four jury deliberations were terminated by the re-

296 Hastie, Schkade, and Payne

Page 11: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

Table S. Relationship between Individual Juror's Predeliberationand Postdeliberation Verdicts (All Cases Combined)

Predeliberation Postdeliberation verdictsverdicts Yes No Undecided Total

Yes

No

Undecided

Total

81%(370)30%(62)60%(40)65%(472)

18%(79)67%(136)33%(22)33%(237)

1%(6)3%(6)7%(S)2%(17)

100%(455)100%(204)100%(67)

100%(726)

searchers because the end of the session had been reached, but no verdict hadbeen rendered (these juries are treated as "Hung" in our analyses). The individualpre- and postdeliberation verdict preferences and confidence ratings show that de-liberation had the general effect of shifting jurors from the "Undecided" state ofmind to a verdict preference (summary statistics in Table 5). Overall, 9% of thejurors declared themselves "undecided" before deliberation and 2% after delibera-tion. Similarly, confidence ratings moved up from pre- to postdeliberation: averagepredeliberation confidence was 83% (on the 50-100% scale) and postdeliberationit moved up to 90%; t(646) = 12.88, p < .0001 with "Undecided" jurors excludedfrom this analysis; means of 80% and 89% with undecideds included, t(725) =16.84, p < .0001. Significant differences were obtained in all of these comparisonsfor each case analyzed separately.

There was also a general tendency for jurors to shift to the "Yes, liable forpunitive damages" verdict during deliberation. Predeliberation "Undecided" jurorswere about twice as likely to shift from "Undecided" to "Yes, liable for punitivedamages" than to "No, not liable" (of the 67 initially "Undecided" jurors, 60%shifted to "Yes" and 33% to "No," 7% remained "Undecided"). This shift is bestexplained as due to coalition size effects during deliberation, larger factions tendto prevail in deliberation; and "Yes" factions were larger at the start of deliberationin most juries.

If we consider the jurors with definite predeliberation verdict preferences, wesee again an advantage for the "Yes, liable" verdict: of the jurors who started with"Yes" verdicts, the probability of changing to "No" was .17 (79/455); while for jurorsstarting with the "No" verdict, the probability of changing to "Yes" was higher: .30(62/204). Again, we would interpret this "Yes" advantage as primarily an effect ofthe larger coalitions favoring the "Yes" verdict (455 initially favored "Yes," 204 fa-vored "No," and 67 declared themselves "Undecided").

These global conclusions need to be qualified for the particular case beingjudged. For the Anderson case, there was a tendency for the shift across deliberationto favor the "No" verdict: the probability of a "Yes" juror changing to "No" was.25, versus the probability for a "No" juror changing to "Yes" of .14. For the Harpercase, the probabilities of changing favored the "Yes" verdict: from "Yes" to "No".19 versus from "No" to "Yes" .42. For the Jardel case, the probabilities of changingfavored the "Yes" verdict: from "Yes" to "No" .23 versus from "No" to "Yes" .34.

Liability Decisions 297

Page 12: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

298 Hastie, Schkade, and Payne

Fig. I. The relationship between individual juror's verdict preferences at the start of deliberationand jury verdicts.

And for the Marine Sulphur Queen case, again in favor of "Yes": from "Yes" to"No" .06 versus from "No" to "Yes" .32.

Another view of these dynamics is provided by the individual jurors' confi-dence-weighted verdict scores (ranging from 0, 100% confident the verdict is "No,"to 100, 100% confident the verdict is "Yes"; "Undecided" jurors were assigned avalue of 50 on this variable). Here there is no overall difference across all casesfrom pre- to postdeliberation, but the patterns are systematic for each case con-sidered separately. For the Anderson case, there was an overall shift toward "No"verdicts; the average predeliberation score was 63, the postdeliberation score was54; t(185) = 3.92, p < .0001. For the Jardel case there was almost no overall change;the average predeliberation score was 60, the postdeliberation score was 61; t(185)= -.31, n.s.). In the Harper and Marine Sulphur Queen cases the shift was towards"Yes"; average predeliberation scores were 57 and 72; postdeliberation scores of65 and 79; ts(185) = -2.91 and -3.41, ps < .0001.

Another insight into the deliberation process is provided by measuring therelationship between the distribution of individual juror's verdict preferences at thestart of deliberation to the jury's verdict. Each jury was classified according to thenumbers of jurors declaring themselves favoring "No," favoring "Yes," or undecidedon the predeliberation questionnaire. Then, for each combination of "No" and"Yes" verdict coalition patterns (e.g., 2 jurors for "No," 6 jurors for "Yes"; 3 jurorsfor "No," 2 for "Yes," 1 juror "Undecided") the proportion of jury verdicts wastabulated.

The results of this tabulation are presented in Figure 1 where the proportionof juries rendering a "Yes, punitive damages are warranted" verdict is plottedagainst the size of the coalition of jurors initially favoring the "Yes" verdict. Onegraph is provided for juries in which all 6 members had initial verdict preferencesand one is provided for juries in which 1 member declared himself or herself "Un-decided" (for simplicity, 13 juries [of 121] in which more than one member was"Undecided" are excluded from this summary). Although sample sizes for the cal-culation of some of the proportions are small, a clear pattern emerges in both

Page 13: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

graphs. There is an obvious "strength in numbers" effect, such that the larger thecoalition favoring "Yes" at the start of deliberation, the higher the probability thatthe jury will decide "Yes." The relationship is not exactly "proportional" (i.e., anidentity relationship between the initial proportion of jurors favoring "Yes" and theprobability of the jury deciding "Yes"), and it seems that the "No" verdict has aslight advantage in deliberation. Overall, when the "No" and "Yes" factions areequal, the jury tends to conclude with the "No" verdict. However, this result shouldbe labeled tentative. First, because most individual jurors for all the cases initially(and finally) favored the "Yes" verdict, there are very few observations underlyingthe calculations of mean probability of verdict values for juries with only 2 or 3jurors initially voting "Yes." Second, almost all of the observations supporting thedominance of larger "No" coalitions come from the Anderson stimulus case. Thisis not to suggest that there is no effect here, just that the initial finding is basedon a small, selective sample of data and must be considered to be tentative.

Jury Deliberation

To allow a detailed analysis of the nature of group deliberations, the vide-otaped jury deliberations were transcribed into written text. Each statement by adeliberating juror was coded into one of 63 categories designed to capture the es-sential decision-relevant elements of the deliberation process (a summary of thecoding system is available from the authors). The categories covered the followingtopics: beliefs and inferences about the case fact situation; legal requirements forconsideration of punitive damages (presented in the judge's instructions); expres-sions of sympathy for the plaintiffs; causal attributions including blame and respon-sibility; personal attitudes and experiences relevant to the case; statements ofconclusions on legal issues; and group management directives (e.g., voting, admo-nitions to follow instructions, etc.).

Coders were told only that they were helping analyze a "study of jury decisionmaking" and they were not provided with detailed information about the researchprocedures or hypotheses. First, a coder segmented each transcript into codablestatements (61,104 statements total, approximately 509 statements per jury). Sec-ond, each statement from every transcript was separately classified independentlyby two coders; the assignment of coders to juries was counterbalanced so everycoder and pair of coders analyzed equal numbers of juries and order of jury codingwas determined randomly for each coder separately. Coders agreed with one an-other on classification of over 70% of the individual statements across the full setof 63 coding categories. Disagreements between coders were dealt with by randomlyselecting one of the two assigned codes. Three random independent selections wereproduced and all basic data analyses were repeated to verify that there were nodifferences in the results.

Not surprisingly, the most common topic of discussion was recalling and in-terpreting the specific details of the case's fact situations (Table 6). It is no exag-geration to note that these civil juries, much as in criminal juries, spend most oftheir time in deliberation trying to construct a common narrative story summary

Liability Decisions 299

Page 14: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

30® Hastie, Schkade, and Pavae

Table 6. Percentage of Statements in MajorCoding Categories

Facts and inferences about the caseDiscussion of legal issuesSympathy for the plaintiffAttributions of blameGroup process issuesGeneral attitudesExperiences, examples, analogiesEvaluations and conclusionsMiscellaneous/uncodableTotal

36.0%14.60.34.4

10.41.66.37.1

19.3100.0%

of the evidence (Pennington & Hastie, 1991,1993). Juries also spent a large portionof their time discussing legal issues (induding the judge's instructions) and drawingconclusions about these legal issues. However, the coverage of legal conditions forpunitive damages presented in the judge's instructions was usually incomplete; onaverage groups discussed 3.13 of 5 and drew conclusions about 2.78 of the 5 legallyrelevant conditions. The other most frequent activities were various group manage-ment activities (who should speak next, taking votes, etc.) and sharing examples,personal experiences, and speculating about hypothetical situations.

Predicting Jury Verdicts from the Contents of Deliberation

We performed a series of regression analyses in which the unanimous juryverdict was the criterion dependent variable ("Yes, award punitive damages" vs."No, do not award punitive damages," with the hung juries excluded). The predic-tors were the frequencies of statements classified into each of the coding categories.Control variables were included in every analysis representing the average prede-liberation verdict of the individual jurors on each jury and "dummy variable" casefactors (Anderson, Harper, Jardel, or Marine Sulphur Queen). Thus, the analysis ac-counts for the variance in group verdicts that remains after these two "control vari-ables" have been entered into the predictive equation.

A first picture of the explanatory power of the deliberation contents can beobtained from a stepwise regression, with the controls forced into the equation first.The controls alone account for a substantial amount of the variance in verdicts (R2

= .42), implying that coalition size and the implications of the evidence were theprimary determinants of verdicts, as they should be. However, adding 15 delibera-tion content variables selected by the stepwise procedure nearly doubles the vari-ance explained, (R2 = .76, incremental F(15, 83) = 3.28, p < .01).

This result illustrates the power of the deliberation contents to capture im-portant variation in verdicts, but such a procedure capitalizes on chance variation,and the particular variables should not be overinterpreted. (The stepwise proceduresimply "stepped forward" adding predictors that increased the amount of variance ac-counted for by significant amounts, p < .05, but the significance tests in such a pro-cedure are of uncertain validity, the procedure adds some variables that are ultimatelynot significant predictors, and there are equally defensible alternate procedures.) To

Page 15: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

determine which variables were the most robust predictors, we conducted severalvariations on this type of analysis and found that 8 of the initial 15 coding categoriesappeared repeatedly and always had significant coefficients with the same signs (Ta-ble 7). Including these 8 variables in a regression with the controls produced anoverall R2 of .68, and a highly significant increment in variance explained over thetwo controls alone, F(9, 88) = 4.52, p < .001. The directions of influence of mostof these predictors are no surprise; in a sense they function as "face validity" checkson the coding categories. For example, a jury that spent more time blaming thedefendant for bad outcomes was more likely to vote for punitive damages. In con-trast, juries that questioned the motives of other parties and stated that more in-formation was needed were more likely to conclude that punitive damages werenot warranted.

Legally Significant Predictors of Verdicts

One of the most striking differences between juries rendering "Yes" verdictsand those concluding with "No" or "Hung" was how thoroughly their delibera-tions considered the legally necessary conditions for punitive damages stated inthe judge's instructions. There are two general justifications for awarding punitivedamages: (1) finding that the defendant was malicious or (2) finding that thedefendant's actions exhibited reckless or callous disregard for risks to others. Thefirst justification is sufficient by itself, but the second requires that a conjunctionof four conditions be present. Figure 2 summarizes the percentages of juries thatdiscussed or drew conclusions about each of these five conditions, separated into"Yes"- and "No"- verdict-rendering juries. For each of the five legal conditions,"Yes" juries were less likely to either discuss or state conclusions about each ofthe conditions.

The relationship between thoroughness of deliberation on the law and verdictsis also observed when summary measures of jury thoroughness are examined. Our

Liability Decisions 301

Table 7. Deliberation Content Categories Related to Group Verdicts

RegressionCoding category" coefficients6

Purpose of punitive damages is to deter others +.139 (.147)Defendant is to blame for a specific event +.033 (.268)Conclusions about whether the defendant was malicious +.038 (.225)Discussion of third party motives -.163 (-.216)

(motives of someone other than plaintiff or defendant)Discussion of meaning of negligigence -.048 (-.269)Discussion of required burden of proof -.046 (-.164)Admonition to follow the judge's instructions -.104 (-.181)Acknowledgment of uncertainities -.016 (-,160)

"Frequency of each category-verdict relationship consistently significant at orbeyond p < .05.

h+ means a jury was more likely to award punitive damages as the number ofstatements in the category increased; - means the inverse relationship wasobserved; standardized coefficients are in parentheses; the complete regressionequation also included a constant term (-1.089).

Page 16: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

382 Hastie, Schkade, and Payne

Fig. 2. The percentages of "Yes" and "No" verdict-rendering juries that mentioned (discussed orreached conclusions) during deliberation the legally required conditions for a judgment that pu-nitive damages were warranted.

Fig. 3. The relationship between the percentages of juries awarding punitive damages and thethoroughness of their coverage during deliberation of the legally required conditions for a judgmentthat punitive damages were warranted.

Page 17: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

most "lenient" global measure of thoroughness was created by indexing whethereach jury discussed or reached conclusions about each of the five legally requiredconditions and adding up these numbers to obtain a measure of thoroughness ofcoverage (ranging in value from 0, no coverage, to 10, both discussion and a statedconclusion for every condition). No juries received a perfect thoroughness score of10, discussion and conclusions on each of the five legal issues. By this measure,"Yes" juries were significantly less thorough, mean = 5.17, than "No" and "Hung"juries, mean = 6.92; t(118) = -3.54, p < .001. Figure 3 depicts the relationshipbetween thoroughness indexed as either discussion or conclusions on each of thefive legal elements and the percentage of juries rendering "Yes" verdicts. The pat-tern is again dramatic, with 100% of the juries who did not cover any legal issuesrendering "Yes" verdicts, dropping to 45% "Yes" verdicts for the most thoroughjuries.

Reinforcing this theme of cursory treatment of legal issues, our quantitativeanalyses of the contents of discussion showed that juries were significantly less likelyto award punitive damages if someone suggested that the judge's instructions shouldbe followed (see Table 7). Greater discussion of the required burden of proof alsoled to fewer punitive damage awards. In contrast, more discussion of extralegalreasons to award punitive damages (i.e., considerations not mentioned in the judge'sinstructions) led to more "Yes" verdicts. Jurors may have difficulty separating thedifferent legal concepts that apply to punitive and compensatory damages andtherefore may treat them as interchangeable, especially when the plaintiff is therecipient of a potential punitive award. In our juries, when someone commentedin deliberation that compensatory damages had already been paid to the plaintiff(a fact that was stated in each case description), the jury was less likely to awardpunitive damages. Similarly, negligence (relevant to compensatory damages) andrecklessness (relevant to punitive damages) may be confused in the minds of jurors.We found that when the meaning of negligence was discussed, the jury was lesslikely to award punitive damages. In both of these cases, confusion about the dis-tinctions between compensatory and punitive damages and between negligence andrecklessness would be likely to produce a bias against the defendant. Overall, dis-cussion of issues that were legally relevant reduced the tendency to decide, incor-rectly, that punitive damages were warranted.

Finally, we examined individual jurors' written postdeliberation explanationsfor their verdicts and rated the extent to which each juror relied on the judge'sinstructions when justifying his or her verdict. A score of 0 indicated no referenceto the instructions, 1 indicated some reliance, and 2 indicated heavy reliance onthe instructions. As for all measures of instruction comprehension and usage in thisstudy, the overall rate of reliance on the judge's instructions was surprisingly low.Forty-nine percent of the jurors made no reference at all to the judge's instructionswhen justifying their verdicts. The extent to which individual jurors referred to thejudge's instructions when justifying their verdicts was strongly correlated with theirjury's verdict (r = -.43 p < .001); jurors who relied more heavily on the instructionswere likelier to be on juries reaching the correct "No" verdict.

Liability Decisions 303

Page 18: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

DISCUSSION

The present experiment provides a coherent picture of juror and jury decisionprocesses when making an important liability decision. For the most part the pat-terns of behavior match those in analogous criminal decisions and in actual jurytrials (e.g., Bray & Kerr, 1979; Diamond, 1979; Dillehay & Nietzel, 1980; Hastie,Penrod, & Pennington, 1983; MacCoun, 1989). One dramatic finding was the lowlevel of comprehension and memory for information conveyed in instructions onthe law. Overall comprehension, indexed with a relatively lenient coding scheme,was low; less than 10% of the most relevant portions of the judge's instructionscould be correctly recalled within minutes of the conclusion of deliberation. Fur-thermore, in the present study the instructions were relatively clear, succinct, andrealistic, they were distinctively prominent in the context of our summarized cases,and written instructions were available during deliberation. Poor comprehensionand memory for instructions is the usual finding in studies of real and mock-jurydecision making, even when researchers provide more specific recall cues or usemultiple-choice or recognition-memory test formats (e.g., Luginbuhl, 1992; Reifman,Gusick, & Ellsworth, 1992; Wiener et al., 1995). We believe that the essential find-ing, disturbingly poor comprehension, is a valid description of what occurs in actualjury trials, although our mock-jury simulation is far enough from courtroom con-ditions that we cannot make general claims about absolute levels of comprehensionand memory.

Jury deliberation did not provide a procedural corrective to individual juror'sfailure to comprehend and apply the judge's instructions. Our content analyses ofthe deliberation process showed the typical jury failed to consider even half of thefive legally prescribed conditions for a finding that punitive damages were war-ranted, even though approximately 65% of these same juries concluded that thedefendant was liable. And we found that juries concluding, "Yes, punitive damageswere warranted," were reliably less thorough in their coverage of the legally pre-scribed conditions than juries concluding that punitive awards were not justified.

The low levels of comprehension by individual jurors and the lack of thor-oughness in jury deliberation imply that juries rely on legal concepts only partly inmaking their decisions on liability for punitive damages (Wiener, Habert, Shkodri-ani, & Staebler, 1991). An analysis of jurors' postdeliberation justifications for theirverdicts also suggests that their reasoning is at best incomplete and often error-prone with reference to the law. For example, almost no jurors, even those con-cluding that punitive damages were warranted, mentioned the element of "grossdeviation from an ordinary standard of care" when explaining the grounds for theirdecisions. Finally, there is a large correlation (r = -.43; the "Not liable" verdictwas coded as a 0, "Liable" as a 1, in this data analysis) between jurors' postdelib-eration references to legal factors and their juries' tendencies to conclude with alegally correct verdict. Thus, there is substantial and varied evidence converging onthe conclusion that jurors and juries make their decisions without careful consid-eration of the law. Furthermore, the less they rely on the law, the more likely theyare to render verdicts that disagree with judges' opinions.

304 Hastie, Schkade, and Payne

Page 19: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

Because of the practical importance of individual differences during the juryselection process, it is interesting to note that we did not find many relationshipsbetween jurors' backgrounds and their predeliberation verdict preferences. Incomeand ethnicity were weakly (but reliably) correlated with predeliberation verdicts.Other researchers have observed stronger relationships between jurors' backgroundcharacteristics (e.g., jurors' income) and award amounts (Zeisel & Diamond, 1976,p. 169). But, since the jurors in the present study made only the liability decisionand did not set awards, we cannot assess this relationship.

The dynamics of the deliberation process in our mock juries was also consis-tent with prior reports and theories of the consensus-seeking process in other kindsof juries. The relationship between individual verdict preferences at the start ofdeliberation and the jury's verdict is best described by a proportionality process inwhich the initial proportion of jurors favoring a verdict is directly related to theprobability of the jury reaching that verdict (Davis, 1973; Stasser, Kerr, & Davis,1989). The dominant role of confidence in predicting the likelihood a juror wouldchange verdicts during deliberation is also consistent with previous observations andtheories of juror opinion change during deliberation (Hastie, Penrod, & Pennington,1983; Stasser & Davis, 1981).

An interesting finding was the observation that in juries which started delib-eration with large coalitions of jurors (3 or more out of 6) favoring the "No, punitivedamages are not warranted" verdict, the "No" verdict was rendered by the jurywith a relatively high probability. Such juries were likelier to reach the "No" verdictthan were juries to render "Yes" verdicts when they were composed of comparablylarge coalitions of "Yes" jurors. This probability (of rendering a "No" verdict) wasalso larger than would be expected from a "proportionality" group decision rule.Again, emphasizing the tentativeness of this observation, one interpretation is thatit results from the asymmetry in the burdens of proof on plaintiff and defendantand from the fact that several elements must all be independently satisfied by apreponderance of the evidence to find for the plaintiff.

The social decision process underlying this interpretation is based on the fol-lowing assumptions: (i) Jurors' predeliberation "individual reasoning" is less thor-ough than the "group reasoning" that occurs during deliberation (see commentbelow on the individual jurors' low comprehension and memory for the jury in-structions). Specifically, the elements that must be decided to render a proper ver-dict are considered less thoroughly by individual jurors than by the jury, (ii) Whena substantial coalition of prodefendant jurors is present in a jury, the deliberationis more likely to review thoroughly and systematically all of the elements necessaryto properly render a verdict, (iii) When a thorough review of the elements isachieved, there is a high probability that the jury will determine that some of theelements are not supported by a preponderance of the evidence and its implications.Thus, jury deliberation is more thorough than individual reasoning, and the morethorough the decision process, the likelier it is to conclude for the defendant, sothere is a (slight) shift toward prodefense verdicts from the beginning to the endof jury deliberation.

This interpretation accrues credibility from the analogous interpretation ofthe "leniency shift" in criminal jury decisions (MacCoun and Kerr, 1988, concluded

Liability Decisions 30S

Page 20: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

that the "leniency shift" was caused by the asymmetry between prosecution anddefense burdens and standards of proof in criminal trials). However, the empiricalevidence is still tentative and the prodefendant leniency shift would only emergein some cases and for jury configurations that include a substantial coalition ofdefendant-favoring jurors (see Lempert, 1981, 1993, for a similar conclusion con-cerning the behavioral implications of special questions when there are many to-be-decided elements in civil cases).

We do not have any firm basis to explain the somewhat lower rate of pro-plaintiff verdicts in the Anderson case compared to the other three cases. We canspeculate that some of the critical differences arise from the fact that this is theonly case in which our mock jurors discussed the possible contribution of the victims(experienced sailors who were killed when their pleasure craft sank in a storm) totheir own fate and it is the only case in which the defendant company definitelytook an action to remedy the problem that was putatively the cause of the accident(the manufacturer mailed a recall/repair notice to dealers and first owners of theboats). However, the patterns of jury deliberation were indistinguishable acrosscases in terms of the content analysis we performed.

In the cases we studied individual jurors exhibited a persistent tendency tofavor the plaintiffs, concluding that punitive damages were warranted when judgeshad concluded they were not. These verdicts are not anomalies; they were consis-tently obtained for the factual circumstances and with standard instructions on thelaw from four representative civil cases. Obviously, we cannot claim that these re-sults, from a sample of only four cases presented in a summary format, will gen-eralize to all punitive damages judgments. But we have established a consistentdiscrepancy between jurors' and some judges' judgments for these cases. We believethat poor comprehension and application of judge's instructions on liability is partof the explanation for the tendency to disagree with the judges. In support of thisinterpretation, we observed a corrective effect in deliberation such that the juriesthat more thoroughly considered the judge's instructions were likelier to renderverdicts in favor of the defendant.

Implications for the Civil Justice System

The subject of punitive damages is surrounded by controversy and the con-troversy is intensified by the high economic stakes and the social significance ofthe issues involved. One fundamental problem is that there is not consensus onthe foundations of the institution: Should the civil justice system serve the goals ofpunishment, retribution, and deterrence in addition to the traditional goal of com-pensating those who have been injured for their losses? This is not primarily an"empirical question"; the answers rest on moral and ethical assumptions about whatwe want, not considerations of what is the truth. Nonetheless, we think that em-pirical research can advance the debate over the necessity and nature of a punitiveawards system by eliminating some nonissues and by clarifying the relations betweensome proposed changes in the system and their likely consequences. Obviously thepresent research is most relevant to proposals to change the procedures in jurytrials.

306 Hastie, Schkade, and Payne

Page 21: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

The proposals to improve the punitive damages institution span the spectrumfrom abolition to "it's not broken, don't fix it" (Galanter & Luban, 1993; Rustad& Koenig, 1993). The proposals for change can be divided into three categories:First there are proposals to change the nature of the institution; these include abo-lition (presumably the functions of the punitive damages award would be trans-ferred to regulatory controls and to the criminal courts), dropping the goal ofretributive punishment so that assessment of an award would involve only consid-erations of deterrence, or shifting the grounds for liability to exclude simple reck-lessness and to emphasize maliciousness. Second there are proposals to improvethe appellate review process, most of these involve clarifying the principles thatidentify excessive awards or provide more specific catalogues of "mitigating andaggravating" considerations. Third there are proposals to improve the performanceof trial courts including providing limits or caps on the award amounts, raising thestandard of proof, and improving instructions to the jury. Suggestions for improve-ments in instructions include adding more exhortations to perform conscientiously,more comprehensible definitions of basic concepts in the current instructions—andmore controlling procedures such as the use of interrogatories and special questionsverdict forms.

The picture of the jurors and the jury that we developed in the present re-search is not reassuring. We believe juries are motivated to perform their task ac-cording to the judge's instructions—at least there is no evidence that they arewillfully disobedient—but they do not apply those instructions correctly (Finkel,1995; Horowitz & Willging, 1991; Myers, 1979; Robinson & Darley, 1995; Wiener,et al., 1991). The main visible departure from proper procedure was that jurorssimply did not consider some legal requirements at all that would have reducedtheir proclivity to decide in favor of the plaintiff on the issue of liability for punitivedamages. The jurors' inability to comprehend legal concepts, the lure of reasoningin terms of everyday notions of responsibility and recklessness, or possibly theirsympathy for the injured plaintiffs, led them to ignore essential legal considerations.

There is also substantial disagreement among jurors (and then juries) on theverdict for every one of the four cases we presented. Our empirical results indicatethat at least part of the explanation of this disorder is that jurors do not fully exe-cute their legally prescribed task and there is also considerable variation in people'sinterpretations of the concepts cited in the instructions on liability. Furthermore,the cases that go to trial (from which our stimulus cases were sampled) are likelyto be those where both parties have convincing arguments. And, in the background,reflecting the controversies over the institution of punitive damages, the everydayconcepts of responsibility, liability, deterrence, and punishment that jurors bringinto the jury box are topics on which there is substantial disagreement among mem-bers of our society.

Our interpretation of what is happening is that jurors are faced with a verydifficult task: their everyday concepts of liability, punishment, and deterrence donot map neatly onto their legal counterparts; their everyday habits of assessingblame do not match the legal procedures of verifying that a series of requisite "ele-ments" have each been established to a specified (and unfamiliar) "standard ofproof"; and when they get to the jury deliberation part of their task they discover

Liability Decisions 307

Page 22: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

that there is substantial disagreement among their fellow citizens about conceptualand factual issues. Under these circumstances, the jurors fall back on their rough-and-ready everyday reasoning habits, probably influenced by their sympathies forone party or the other, and they fail to complete the task presented to them intheir instructions from the judge.

We are not arguing for elimination of the jury trial or abolition of the punitivedamages in civil trials. We believe that clarification, by authorities like the SupremeCourt, of a theory or underlying policy for the institution will have major benefits.The efforts to improve the acuity of appellate review by clarifying the underlyingtheory and by providing more detailed guidelines to identify inappropriate verdictsshould also increase the reliability of the system. The present findings suggest thatproviding more direct guidance and control over the jury decision process—for ex-ample, through a device such as requiring answers to interrogatory questions onthe elements warranting each general verdict or by returning special verdicts-should help (see May, 1995, and Wiggins and Breckler, 1990, for concurring opin-ions). Especially in cases in which there is a clear structure of conjunctive anddisjunctive relationships on elements that must be decided to properly render thegeneral verdict, an interrogatory questionnaire could insure that each element re-ceived some consideration in the decision process. Furthermore, if, as we have sug-gested, there is a bias in jury decisions against unsympathetic defendants and infavor of injured plaintiffs, increasing the requisite standard of proof should helpreduce the rate of improper verdicts.

But there may still be additional problems for the jury deciding liability forpunitive damages that will not be corrected by forcing juries to heed more carefullythe law given to them in the judge's instructions. It is not surprising to find thatjurors are confused by their instructions on judging liability for punitive damagesgiven that there are few, if any, coherent theories of the nature of the institutionat the highest levels of legal policy analysis. Even the juries that most thoroughlyconsidered the relevant legal requirements still rendered, "Yes, punitive damagesare warranted" verdicts about half (45%) of the time, demonstrating substantialdisagreement among jurors and between jurors and judges. Supreme Court JusticeO'Connor expressed the critical implication of our empirical findings: "Many courtscontinue to provide jurors with skeletal guidance that permits the traditional guar-antor of fairness—the jury itself—to be converted into a source of caprice and bias"(TXO Prod. Corp. v. Alliance Resources Corp., 1993, p. 2742). Here again we believethat further empirical research can inform and focus the debate over whether andhow we should assess punitive damages.

APPENDIX

Jardel Co., Inc. v. K. Hughes

This case involves the following two parties: the plaintiff, Kathleen Hughes,a shopping mall employee who was abducted from the mall parking lot, beaten,

308 Hastie, Schkade, and Payne

Page 23: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

and raped; the defendant, Jardel Co., Inc., owner and operator of the shoppingmall.

The specifics of the case are as follows:On July 18, 1980, shortly after the 9:00 p.m. closing time, the plaintiff, an

employee of the Woolco store in the Blue Hen Mall, left her place of employment.She walked through the interior arcade of the mall and exited through the rearentrance adjacent to the parking lot where her vehicle was located. Near the en-trance to the Fox theater, the plaintiff was accosted by two scruffy-looking youngmen who had been denied entrance to the theater because Fox personnel believedthem intoxicated. When these individuals asked plaintiff for a cigarette, she brieflyspoke with them, gave them a cigarette and walked on. The men followed the plain-tiff to her car, which was parked approximately 50 feet from the mall entrance.When she attempted to enter her car, they forced their way into her car with herand drove away.

After being driven to a remote site adjacent to the mall, the plaintiff wasbeaten and raped by both assailants. While she was unconscious and lying on theground, her assailants attempted to ran her over with her car and later set fire tothe car while she was lying in it. Plaintiff regained consciousness and staggerednude and bloodied onto a nearby highway where a passing motorist found her andtook her to a nearby hospital.

Plaintiff remained hospitalized for 6 days She was treated for severe contu-sions to the face and scalp, a cerebral concussion and a permanent skull fracture.This latter injury has resulted in a permanent displacement of her left eye anddouble vision. In addition to her physical injuries the plaintiff, at the time of trialfour years later, still suffered psychological effects from her traumatic experience.

Since 1979, Jardel had contracted with Globe Security Systems, Inc. to provideguards for the mall exterior. Jardel paid according to the number of guards providedby Globe. Jardel decided to use one guard for each nightly shift beginning at 9p.m. Globe personnel testified that they had suggested to Jardel that the mall ex-terior was too large for one guard and that more guards were needed. Jardel deniedreceiving a specific recommendation but did acknowledge that Globe requested,on more than one occasion, an increase in the number of guards. In response,Jardel asked Globe to justify the additional expense.

Globe said they wanted an additional guard because t h e r e was no radio con-tact with the Dover police. The police allowed direct radio access only to policeofficers. Globe wanted a second guard inside the mall, who would be in radio con-tact with the outside guard and could telephone the police. Jardel countered thatmaintenance personnel who already worked inside the mall throughout the nightcould contact the police if necessary.

When Hughes was forced into her vehicle in the rear parking lot, the singleGlobe guard was in his patrol vehicle, parked in the front parking lot.

Mali owners are not compelled to provide security. But Jardel chose to pro-vide security, presumably to render the mall more attractive to its tenants, andtheir customers and employees. Once Jardel did provide a security program, Jardelbecame obligated to do it in a reasonable manner, with a view toward the dangersto which the program was directed. The standard that governs this situation says,

Liability Decisions 3W

Page 24: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

"A possessor of land who holds it open to the public for business purposes is liableto the public while they are on the land for such purpose, for physical harm causedby ... intentionally harmful acts of third persons. If past experience is such thatthe owner should reasonably anticipate criminal conduct on the part of third per-sons, either generally or at some particular time, he may be under a duty to takeprecautions against it, and to provide a reasonably sufficient number of servantsto afford reasonable protection."

A Dover police officer presented evidence of the number of police responsesto the Blue Hen Mall in the 2 1/2 years preceding the incident. Of 394 incidentsreported, over 90% were property or nonpersonal crimes. The remaining incidents,however, involved a kidnapping at gun point, an armed robbery, a sexual molesta-tion, indecent exposures, and purse snatchings, with approximately one-half of theseincidents occurring in the mall parking lot.

The court concluded that under the circumstances, the protection providedby Jardel was an insufficient response to the known history of criminal activity.

Hughes was compensated in full for her expenses, lost income, and pain andsuffering. In addition, Hughes claimed she was entitled to punitive damages.

Plaintiff Argument in Favor of Punitive Damages

Despite the history of violent crimes in the Blue Hen Mall, and despite therepeated requests from Globe Security to add another guard because of the largesize of the parking lot, Jardel chose to save money by having only one exteriorguard on duty. Punitive damages should be awarded against Jardel to punish themfor flagrantly avoiding their duty of protecting mall employees and customers, andto deter them, and other mall owners, from such future behavior.

Defendant Argument Against the Award of Punitive Damages

Although we all wish crime could be prevented, we know it cannot be. In thiscase, Jardel voluntarily provided a reasonable level of security, consistent with thehistory of crime on their mall property. Jardel's behavior was not even close to themalicious or reckless and callous disregard for the rights of others that is requiredby the Judge's instructions for the award of punitive damages.

Jury Instructions

The judge has given you the following instructions that you are required by lawto use in deciding whether or not to award punitive damages.

The purposes of punitive damages are to punish a defendant and to deter adefendant and others from committing similar acts in the future.

Plaintiff has the burden of proving that punitive damages should be awardedby a preponderance of the evidence. You may award punitive damages only if youfind that the defendant's conduct

(1) was malicious; or(2) manifested reckless or callous disregard for the rights of others.

310 Hastie, Schkade, and Payne

Page 25: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

Conduct is malicious if it is accompanied by ill will, or spite, or if it is forthe purpose of injuring another.

In order for conduct to be in reckless or callous disregard of the rights ofothers, four factors must be present. First, a defendant must be subjectively con-scious of a particular grave danger or risk of harm, and the danger or risk mustbe a foreseeable and probable effect of the conduct. Second, the particular dangeror risk of which the defendant was subjectively conscious must in fact have even-tuated. Third, a defendant must have disregarded the risk in deciding how to act.Fourth, a defendant's conduct in ignoring the danger or risk, must have involved agross deviation from the level of care which an ordinary person would use, havingdue regard to all the circumstances.

Reckless conduct is not the same as negligence. Negligence is the failure touse such care as a reasonable, prudent, and careful person would use under similarcircumstances. Reckless conduct differs from negligence in that it requires a con-scious choice of action, either with knowledge of serious danger to others or withknowledge of facts which would disclose the danger to any reasonable person.

To "establish by a preponderance of the evidence" means to prove that some-thing is more likely so than not so. In other words, a preponderance of the evidencein the case means such evidence as, when considered and compared with that op-posed to it, has more convincing force, and produces in your minds belief that whatis sought to be proved is more likely true than not true.

In your decisions on issues of fact, a corporation is entitled to the same fairtrial at your hands as a private individual. All persons, including corporations, part-nerships, and other organizations, stand equal before the law, and are to be dealtwith by the judge and jury as equals in a court of justice.

The verdict must represent the considered judgment of each juror. In orderto return a verdict, it is necessary that each juror agree thereto. Your verdict mustbe unanimous.

Upon retiring to the jury room, you will select one of your number to act asyour presiding juror. The presiding juror will preside over your deliberations.

Question

Kathleen Hughes (the plaintiff), a shopping mall employee who was abducted,beaten, and raped, while going to her car after work in the mall, has sued themall's owner, Jardel Co., Inc., (the defendant) asking for punitive damages. Youhave been given a legal definition of when punitive damages are proper. You mustuse that definition whether you agree with it or not. Based on that definition wouldit be proper to award punitive damages against Jardel Co., Inc., yes, or no?

ACKNOWLEDGMENTS

Support for the research was provided by funds from the National ScienceFoundation (Grant No. SBR 9410288) and from the Exxon Co., USA. The authors

Liability Decisions 311

Answer: Yes No

Page 26: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

would like to thank Richard Berk, Valerie Hans, Charles Judd, Janet Kabili, GaryMcClelland, and the staff of Bodaken Associates for assistance and advice on vari-ous aspects of the research. Of course, the views expressed in this paper shouldonly be attributed to the authors.

REFERENCES

Adler, S. J. (1994). The Jury. New York: Random House.American Law Institute. (1991). Reporter's study: Enterprise responsibility for personal injury: Vol. 2.

Approaches to legal and institutional change. Philadelphia: American Law Institute.Anderson v. Whittaker Corp., 692 F. Supp. 734 (W.D. Mich., 1987).BMW of North America, Inc., v. Ira Gore, Jr., 116 S.Ct. 1589 (1996).Bray, R. M., & Kerr, N. L. (1979). Use of the simulation method in the study of jury behavior. Law

and Human Behavior, 3, 107-119.Broder, I. E. (1986). Characteristics of million dollar awards: Jury verdicts and final disbursements.

Justice System Journal, 11, 349-359.Brown, D.K. (1996). Structure and relationship in the jurisprudence of juries: Comparing the capital

sentencing and punitive damages doctrines. Hastings Law Journal, 47, 1255-1323.Clermont, K. M., & Eisenberg, T. (1992), Trial by jury or judge: Transcending empiricism. Cornell Law

Review, 77, 1124-1177.Daniels, S., & Martin, J. (1986). Jury verdicts and the 'crisis' in civil justice. Justice System Journal, 11,

321-348.Daniels, S., & Martin, J. (1990). Myth and reality in punitive damages. Minnesota Law Review, 75, 1-64.Daniels, S., & Martin, J. (1995). Civil juries and the politics of reform. Evanston, !L: Northwestern

University Press.Davis, J. H. (1973). Group decision and social interaction: A theory of social decision schemes.

Psychological Review, 80, 97-125.Diamond, S. S. (1979). Simulation: Does the microscope lens distort? [Editor's introduction to special

issue]. Law and Human Behavior, 3, 1-4.Dillehay, R. C, & Nietzel, M. T. (1980), Constructing a science of jury behavior. In L. Wheeler

(Ed.), Review of personality and social psychology (Vol. 1, pp. 246-264). Beverly Hills, CA:Sage.

Eades, R. W. (1993). Jury instructions on damages in tort actions (3rd ed.). Charlottesville, VA: Michie.Finkel, N. J. (1995). Commonsense justice: Jurors' notions of the law. Cambridge, MA: Harvard University

Press.Galanter, M., & Luban, D. (1993). Poetic justice: Punitive damages and legal pluralism. American

University Law Review, 42, 1395-1463.Ghiaridi, J. D., & Kircher, J, J. (1995). Punitive damages: Law and practice. Deerfield, IL: Clark,

Boardman and Callaghan.Harper v. Zapata Off-shore Co., 563 F. Supp. 576 (1983).Hastie, R. (1991). Is attorney-conducted voir dire an effective procedure for the selection of impartial

juries? American University Law Review, 40, 1501-1524.Hastie, R., Penrod, S. D., & Pennington, N. (1983). Inside the jury. Cambridge, MA: Harvard University

Press.Hensler, D. R. (1988). Researching civil justice: Problems and pitfalls. Law and Contemporary Problems,

51, 55-65.Honda Motor Co. Ltd., et al., v. Karl L. Oberg, 114 S.Ct. 2331 (1994).Horowitz, I. A., & Willging, T. E. (1991). Changing views of jury power: The Nullification Debate,

1787-1988. Law and Human Behavior, 15, 165-182.Huber, P. W. (1988). Liability: The legal revolution and its consequences. New York: Basic Books.In re Marine Sulphur Queen, 460 F.2d 89 (1972).

In re The Exxon Valdez, No. A89-0095-CV (D. Alaska Sept. 1994).Jardel Co., Inc. v. Hughes, Del. Supr., 523 A. 2d 518 (1987).Landes, W. M., & Posner, R. A. (1986). New light on punitive damages. Regulation,

1986(September/October), 33-36, 54.Lempert, R. O. (1981). Civil juries and complex cases: Let's not rush to judgment. Michigan Law Review,

80, 68-132.

312 Hastie, Schkade, and Payne

Page 27: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

Lempert, R. O. (1993). Civil juries and complex cases: Taking stock after twelve years. In R. E. Litan(Ed.), Verdict: Assessing the civil jury system (pp. 181-247). Washington, DC: Brookings Institute.

Litan, R. E. (Ed.) (1993). Verdict: Assessing the civil jury system. Washington, DC: BrookingsInstitute.

Luginbuhl, J. (1992). Comprehension of judge's instructions in the penalty phase of a capital trial. Lawand Hitman Behavior, 16, 203-218.

MacCoun, R. J. (1989). Experimental research on jury decisionmaking. Science, 244, 1046-1050.MacCoun, R. J., & Kerr, N. L. (1988). Asymmetric influence in mock jury deliberation: Jurors' bias for

leniency. Journal of Personality and Social Psychology, 54, 21-33.Mahoney, R. J., & Littlejohn, S. E. (1989). Innovation on trial: Punitive damages versus new products

Science, 246, 1395-1399.May, C. N. (1995). 'What do we do now?': Helping juries apply the instructions. Loyola of Los Angeles

Law Review, 28, 869-902.Myers, M. A. (1979). Rule departures and making law: Juries and their verdicts. Law and Society Review,

13, 781-798.Neter, J., Wasserman, W., & Kutner, M. H. (1983). Applied linear regression models. Homewood, IL:

Richard D. Irwin.Ostrorn, B., Hanson, R., & Daley, H. (1993). So the verdict is in—What happens next? The continuing

story of tort awards in state courts. Justice System Journal, 16, 97-115.Pacific Mutual Life Insurance Co. v. Haslip, el al., 499 U.S. 1 (1991).Pennington, N., & Hastie, R. (1991). A cognitive theory of juror decision making: The Story Model.

Cardozo Law Review, 13, 519-557.Pennington, N., & Hastie, R. (1993). A theory of explanation-based decision making. In G. Klein, J.

Orasanu, R. Calderwood, & C. E. Zsambok (Eds.), Decision making in action: Models and methods(pp. 188-204). Norwood, NJ: Ablex.

Penrod, S. D. (1979). A study of attorney and "scientific" jury selection models. Unpublished doctoraldissertation, Harvard University.

Peterson, M., Sarma, S., & Shanley, M. (1987). Punitive damages: Empirical findings. Santa Monica,CA RAND Corp.

Reifman, A., Gusick, S. M., & Ellsworth, P. C. (1992). Real jurors' understanding of the law in realcases. Law and Human Behavior, 16, 539-554.

Robinson, P. H., & Darley, J. M. (1995). Justice, liability, and blame: Community views and the criminallaw. Boulder, CO: Westview Press.

Roeca, A. F. (1984). Damages. In G. Z. Nothstein (Ed.), Toxic torts: Litigation of hazardous substancecases (pp. 494-524). New York: McGraw-Hill.

Rustad, M., & Koenig, T. (1993). The historical continuity of punitive damages awards: Reforming thetort reformers. American University Law Review, 42, 1284-1304.

Saks, M. J. (1992). Do we really know anything about the behavior of the tort litigation system—andwhy not? University of Pennsylvania Law Review, 140, 1147-1292.

Saks, M. J., Hollinger, L. A., Wissler, R. L., Evans, D. L, & Hart, A. J. (1997). Reducing variabilityin civil jury awards. Law and Human Behavior, 21, 243-256.

Schuck, P. H. (1993). Mapping the debate on jury reform. In R. E. Litan (Ed.), Verdict: Assessing thecivil jury system (pp. 306-340). Washington, DC: Brookings Institute.

Snedecor, G. W., & Cochran, W. G. (1980). Statistical methods (7th ed.). Ames, IA: Iowa State UniversityPress.

Stasser, G., & Davis, J. H. (1981). Group decision making and social influence: A social interactionsequence model. Psychological Review, 88, 523-551.

Stasser. G., Kerr, N. L., & Davis, J. H. (1989). Influence processes and consensus models indecision-making groups. In P. B. Paulus (Ed.), Psychology of group influence (pp. 279-326). Hillsdale,NJ: Erlbaum.

TXO Prod. Corp. v. Alliance Resources Corp., et al., 113 S.Ct. 2711 (1993).Vidmar, N. (1995). Medical malpractice and the American jury: Confronting myths about jury incompetence,

deep pockets, and outrageous damage awards. Ann Arbor, MI: University of Michigan Press.Wells, G. L, & Gavanski, I. (1989). Mental simulation of causality. Journal of Personality and Social

Psychology, 56, 161-169.Wiener, R. L., Gaborit, M., & Pritchard, C. C. (1994). Counterfactual thinking in mock juror assessments

of negligence: A preliminary investigation. Behavioral Sciences & the Law, 12, 89-102.Wiener, R. L., Habert, K., Shkodriani, G,, & Staebler, C. (1991). The social psychology of jury

nullification: Predicting when jurors disobey the law. Journal of Applied Social Psychology, 21,1379-1401.

Liabiiity Decisions 313

Page 28: A Study of Juror and Jury Judgments in Civil Cases: Deciding Liability for Punitive Damages

Wiener, R. L., Prichard, C. C, & Weston, M. (1995). Comprehensibility of approved jury instructionsin capita] murder cases. Journal of Applied Psychology, 80, 455-467.

Wiggins, E. C., & Breckler, S. J. (1990). Special verdicts as guides to jury decision making. Law &Psychology Review, 14, 1-41.

Zeisel, H., & Diamond, S. S. (1976). The jury selection in the Mitchell-Stans conspiracy trial. AmericanBar Foundation Research Journal, 1, 151-174.

314 Hastie, Schkade, and Payne