Psychological Reports, 1982, 50, 259-266. Psychological Reports 1982 PSYCHOLOGICAL ISSUES RAISED IN THE CALIFORNIA SUPREME COURT CASE PEOPLE VS COLLINS 1 ROBERT M. KAPLAN AND CATHIE J. ATKINS San Diego State University Surnrnary.--In a California armed robbery trial one member of the jury was dismissed after a significant period of deliberation. Shortly after an alternate was added to the jury the defendant was found guilty. Upon appeal *he California State Supreme Court upheld the conviction but defined a new procedure requiring judges to instruct the jury to forget their prior deliberation and begin anew upon the substitution of a new juror. In a simulation of the situation, half of the subjects were assigned to juries in which two confederates argued for conviction while the other half were assigned to juries without confederates. A poll of the jury showed that the confederates were effective in persuading more subjects to favor conviction. Then a late juror was introduced and judges' instructions to begin deliberations anew were read. The new California procedure assumes that a judge's instructions can serve to negate the effect of prior deliberations. Contrary to this assumption the judge's instructions were not effective as a means of eliminating differences between the two types of jury. Alvin Leon Collins was convicted on three counts of first degree robbery in a 1975 San Diego jury trial. The case was submitted to the jury at 3:12 p.m. on the third day of trial. After no verdict had been reached by 4:30 p.m., the jury was excused for the day. The following morning one of the jurors explained that she would not be able to make a decision and asked to be excused from duty. The Court dismissed her and substituted an alternate juror. The substitution was made over the objection of the defense and was a subsequent point of appeal. This paper addresses the social-psychological implications of such substitution. In California the essential elements of the right to trial by jury require that a jury in a felony prosecution consist of 12 persons and that its verdict be unanimous. The requirement that 12 persons reach a unanimous verdict is not met unless those 12 reach their consensus through deliberations which are the common experience of a11jurors. In response to Collins' appeal, the Supreme Court ruled that, "A proper construction of section 1089 requires that deliberation begin anew when a substitution is made after final submission to the jury. This will insure that each of the 12 jurors, reaching the verdict, has fully participated in the deliberation." Despite this decision, the request for a new trial was denied. Subsequently the 1Address correspondence to Robert M. Kaplan, Psychology Clinic, San Diego State University, San Diego, California 92182. The authors gratefully acknowledge the assistance of Jeff Greet, Tara Howard, Terry Lunine, and Greg Balaban during many phases of this research, and to Tim Connor and Steve Perrillo for consultation on legal aspects of the case.
260 R. M. KAPLAN & C. J. ATKINS judge's instruction to begin new deliberations when a new juror is introduced has become known as the "Collins Instruction." This procedure rests on the assumption that jurors exposed to the Collins instruction witl be capable of disregarding all that was said during the initial deliberation. There is evidence that the Collins ruling has affected other important cases. The case was cited in an appeal (People Vs Guzn an) to the California State Court of Appeal. In the Guzman case a jury member attempted to influence other jurors by bargaining and bartering. After seven days of deliberations, the inappropriate juror was dismissed and replaced with an alternate juror. The new jury found Guzman guilty. In his appeal, Guzman requested that a mistrial be called because the judge did not instruct the jury to deliberate anew as specified in the Collins decision. A better known case involving the Collins instruction was the second trial of Manson family member Leslie Van Houton. In this trial a juror was replaced after three days of deliberation and the Collins instruction was given. The present experiment examines the assumption that a judge's instructions can wipe out the effects of deliberation. Specifically, we investigated the effect of asking a jury to begin new deliberations when a new juror is introduced. Relevant Research in Social Psychology Some research in social psychology argues against the reasonableness of the Collins instruction. Social influence resulting from deliberation should be difficult to reverse by a simple instruction from a judge. At least two lines of evidence support this argument: these are research on discussion-induced polarization and research on cognitive dissonance. Discussion-induced polarization.--some experiments have examined the psychological consequences of deliberation. These studies show that jurors' assessments of guilt or innocence polarize as a function of group discussion. Jurors have individual opinions prior to deliberation and, as a function of the deliberation, these opinions shift toward leniency or incrimination. Group polarization effects have been demonstrated for both simulated (Davis, 1973; Kaplan & Kemmeride, 1974) and actual juries (Kalven & Zeisel, 1966). Cognitive consistency theories.--under the California rule of People Vs Collins, instructions to jurors to forget about their prior deliberations are said to be used to undo the effect of deliberations which took place before the replacement juror was introduced. Theories of cognitive consistency, however, would argue that the judge's instructions should not be effective for this purpose. The process of decision has been of central importance in the numerous studies on dissonance theory. The theory assumes that information is neither consonant nor dissonant until a point of decision. After reaching one decision, information contrary to the decision will create cognitive inconsistencies or a state of cognitive dissonance. To maintain cognitive balance, this new information may not have
CHAINTGEIN INSTRUCTIONS TO JURY 261 much influence upon judgment. Numerous experimental studies show that information has less impact if it is acquired after a subject has come to accept certain contrary principles, see Bear and Hodum (1975) for example. Thus the arguments of a new.juror may not be influential if presented after most jurors had already reached a point of decision. In the present experiment subjects were exposed to the evidence in the Collins trial and then deliberated the issues in mock juries. In half of the juries, two confederates argued strongly that Collins should be convicted, and it was expected that these arguments would persuade the other jurors. After a first polling of the jury, a new juror was introduced and the Collins instruction was given. Thereafter the confederates made no further attempt to persuade other jurors. A test of the effectiveness of the Collins instruction is based on the results of jutt polling which occurred after a second deliberation period. The Collins instruction assumes that jurors can disregard the results of prior deliberations if they are instructed to do so. If both groups are given this instruction, then any differences created by the influence of the confederates should be washed out and the groups should not differ at the time of the second polling. A significant difference at the time of the second polling would indicate that the Collins instruction is not adequate as a means of nullifying the effect of earlier deliberation. Subjects METHOD The subjects were 68 female and 40 male undergraduate students at a large public university on the West Coast. In exchange for their participation, the subjects received credit in an introductory psychology course. Conditions The subjects were randomly assigned to one of two types of experimental juries. For the experimental groups, two confederates were placed in the juries along with naive subjects. During the deliberations, these confederates argued that the defendant was guilty by presenting specific rehearsed arguments. After 10 rain. of deliberations, the jury was interrupted and each juror was asked to record privately judgments about the guilt or innocence of the defendant. Then a new juror was added to the jury and the experimenter read the Collins rule. The subjects were told that the new juror had not benefited from the prior deliberations because of his late arrival Since the law requires that each juror have an equal chance to influence the others, all prior deliberations were to be disregarded. The importance of beginning deliberations anew was stressed. After the introduction of the alternate juror, the confederates were silent. The procedure for the control group was identical in every respect except
262 R.M. KAPLAN & C. J. ATKINS confederates were not involved in the deliberations. There were five experimental and five control juries which ranged in size from 9 to!2 participants. Materials Before deliberating, the subjects read and heard a tape recording of an abbreviated transcript from the actual Collins trial. The three-page transcript noted that Collins and two accomplices had allegedly picked up three young Marine hitch-hikers and Collins directed an accomplice to hold a knife to the throat of one victim and demand his wallet. The description also detailed the circumstances of Collins' arrest and the arguments presented on his behalf at the trial. For example, another convicted felon testified that he had been with Collins at a movie during the time of the alleged robbery. It was also noted that some of the prosecution testimony against Collins was provided by an accomplice who was testifying against Collins in exchange for having the charges against himself reduced from first degree to second degree robbery. 2 Before beginning the deliberations, the subjects read an abbreviated version of the actual instructions given to the jury in the Collins trial. An audiotape of these instructions was also played. The instructions defined first and second degree robbery, made the distinction between reasonable doubt and possible doubt, and discussed the standards of evidence. _ The second deliberation period usually ended after about 5 rain. Dependent Measures The juries were polled twice, once immediately before the introduction of the late juror and again after the second deliberation period. On each occasion, a brief questionnaire was distributed which asked whether, in the subject's judgment, the defendant was guilty or not guilty. A second question asked how sure the subject was of his own judgment. The alternatives were: very unsure, unsure, neither sure nor unsure, sure, and very sure. For scoring, the numbers 1 (for very unsure) through 5 (for very sure) were assigned to these ratings. The scores were multiplied by --1 if the defendant had been judged guilty and -_1 if he had been judged not guiky. Thus for each subject a composite judgment score with a possible range of --5 (guilty and very sure, --1 X 5 ----- --5) to 5 (not guilty and very sure, 1 X 5 -- 5) was possible. This same measure was administered again at the time of the second polling. On this occasion several other questions were also asked. The questionnaire asked for judgments of the credibility of testimony provided by the victims and the accomplices in the trial and questions concerning the subjects' regard for the legal system. In addition, the subjects were asked, "If you were the judge The names used in these materials were not the same as those in the real case. For instance, the name "Collier" was used in place of "Collins." *The summary of the testimony, judges' instructions and the instructions for the confederate are availablefrom the authors.
CHANGE IN INSTRUCTIONS TO JURY 263 and the jury returned a guilty verdict in this case, how much time would you recommend that the defendant spend in prison? The minimum sentence for armed robbery is 5 yr. and the maximum is 20 yr." RESULTS Table 1 presents the results of the jury pollings. These data were analyzed in a 2 )< 2 (Group X Time of Polling) analysis of variance with time of polling as a repeated measure. The unit of analysis for these comparisons was jury since individuals within juries could not be considered independent of one another. There were significant main effects for group (F_/ 5.43, p <.05) and for change between first and second polling of the jury (F_/7 -- 6.95, p <.05). The test of the hypothesis that the Collins instruction negates the effects of prior influence is given by the interaction term. Specifically, a significant interaction should be observed if the differences at the second polling are smaller than the differences at the time of the first polling. The interaction term representing a test of this hypothesis was nonsignificant (Fx/7 --.12, p >.08). To clarify the meaning of these results, separate group comparisons for the first and the second polling are presented. As expected, the experimental group was significantly more in favor of conviction at the time of the first polling on the composite measure (ts _- 2.71, p <.05). 4 Nearly 56% of these subjects, who had been exposed to the confederates favored a guilty verdict. These results suggest that the confederates were successful in persuading some jurors that the defendant was guilty. TABLE 1 JURY POLLING BEFORE AND AFTER THE COLLINS INSTRUCTION Polling Measure Experimental Control p Group Group Before Instruction M composite --.40 1.16 <.05 % favoring conviction 55.50 28.60 <.05 After Instruction M composite.27 1.62 <.05 % favoring conviction 41.80 22.40 <.05 If the Collins instruction works as indicated, the differences between groups should have been nullified by the time of the second polling. The instruction should have wiped out any effect of the initial deliberation. Since the groups were created by a random assignment, only chance difference should have remained. After the Collins instruction and the second deliberation period, how- *Four subjects were eliminated from the analysis because they failed to choose between guilty or not guilty.
264 R.M. KAPLAN & C. J. ATKINS ever, the difference between the groups was maintained (t8 -- 2.55, p <.05). There was a slight shift toward leniency as a result of the second deliberation. The correlation between the composite score from the first and second polling was.82 (p <.001). A major parr of the confederates' arguments was that the accomplices' word should not be taken seriously since they were receiving reduced sentences in exchange for their testimony. It was expected that faith in the testimony of the accomplices would mediate the tendency to judge the defendant as guilty. Results showed that those who had been exposed to the confederates saw the accomplices' testimony as less credible (M ---:2.53) than the control group (M --: 3.10) even after the judge's instructions to disregard earlier deliberations (t8 _- 2.43, p <.05). Personal attitudes and characteristics of the subjects were uncorrelated with judgments about the guilt or innocence of the defendant (all Rs <.15). In addition, there were no main effects or interactions associated with subjects' sex (all Fs < 1.0). The groups did not differ in their recommendations for a prison sentence. DISCUSSION The results of this experiment suggest that a judge's instructions to deliberate anew might be inadequate as means of nullifying the effect of a previous deliberation. Thus, the data argue against the reasonableness of the Collins instruction. The data do conform with the findings of previous psychological research. One potential problem with the experiment was that each subject-juror was polled twice. To maintain consistency, a juror may respond to the second poll as he had to the first. There is some evidence that subjects attempt to be consistent in the eyes of the experimenter (Rosenthal, 1966). This could be an alternative explanation of the nonsignificant interaction. This argument becomes less tenable, however, since there is a main effect for time of _eliberation. If jurors' positions become "fixed" as a result of the first polling, there should not have been a change between the first and the second pollings of the juries. Nevertheless, future research might avoid this problem by making the measurement of subjects' position a between- rather than a within-subjects factor. An earlier experiment on the Collins case demonstrated that the probability an alternate juror will vote for conviction is a function of the number of jurors who favor a guilty verdict at the time the alternate joins the jury (Atkins, Kaplan, & Howard, 1977). Subjects in this simulation were randomly assigned to hypothetical juries which systematically differed in the number of jurors already favoring conviction. In five conditions, the juries were described as 0-11, 2-9, 6-5, 9-2, and 0-11 for guilty and not guilty, respectively. The results of a planned comparisons showed that most of the variance between groups
CHANGE IN INSTRUCTIONS TO JURY 265 could be accounted for by a linear tendency for subjects to endorse conviction as a function of the number of other jurors already disposed in this direction. This experiment confirmed 25 yr. of research in social psychology suggesting that situations, such as that faced by the alternate juror in the Collins case, produce pressures to conform (Asch, 1952; Kiesler & Kiesler, 1969). Several of the methodological limitations in the present study have been addressed in a Master's thesis by Lunine (1979), who used many of the same materials. She asked what would have happened if the confederates had argued against (rather than for) conviction. In addition, she noted that in the Collins trial the confederate was replaced rather than dismissed. In our experiment no replacement occurred. In the Lunine study, a confederate argued either for conviction, against conviction, or took a neutral position. Then the confederate was dismissed, the Collins instruction was given, and a new confederate was substituted. The replacement juror also argued either for conviction, against conviction, or was neutral. In summary, the experiment used a 3 5< 3 factorial design with different confederates taking one of three positions prior to and after the Collins instruction. The results support those from the present experiment. When the substitute juror took a neutral position, the Collins instruction did not undo the attitude change created by the first confederate. In other words, the judge's instruction to deliberate anew and ignore the earlier deliberation did not eliminate the biasing effect of the first ddiberation. In conjunction with other experiments on the Collins case, the data suggest that there are at least two important factors which may operate to produce bias in post-deliberation jury substitution. First, the situation is one in which social pressures to conform are maximized. Further, it is unlikely that the alternate juror could have an equal influence upon the group decision since other jurors may have already made a decision when the new juror is introduced. Second, it is not likely that the Collins instruction can effectively wipe out the results of prior deliberation. Although these studies have several methodological limitations, we suggest that the Collins instruction, which has now become part of California procedure, be reviewed and reconsidered. Several authors have discussed the limitations of the jury-simulation method (Gerbasi, Zuckerman, & Reis, 1977 ). The present results must be considered in light of these limitations. REFERENCES ASCH, S.E. Social psychology. Englewood Cliffs, NJ: Prentice-Hall, 1952. ATKINS, C. J., KAPLAN, R. IV][., & HOWARD, T. Experimental tests of psychological issues in a California Supreme Court case. Presented at the meeting of the Western Psychological Association, San Francisco, 1977. BEAR, G., & HODUM, A. Implicational principles and the cognition of confirmation contradictory, incomplete, and irrelevant information. Journal of Personality and Social Psychology, 1975, 32, 594-604. DAVIS,J.H. Group decision and social interaction: a theory of social decision schemas. Psychological Review, 1973, 80, 97-125.
266 R.M. KAPLAN & C. J. ATKINS GERBASI, K., ZUCKERMAN, IV[., & REIS, H. Justice needs a new blindfold: a review of mock jury research. Psychological Bulletin, 1977, 84, 323-345. KALVEN, H., & ZEISEL, H. The American jury. Boston: Little, Brown, 1966. KAPLAN, IV[., & KEMMERICK, G.D. Juror judgment as information integration: combining evidential and nonevidential information. Journal o[ Personality and Social Psychology, 1974, 30, 493-499. KIESLER, C., & KIESLER, S. Conlormity. Reading, MA: Addison-Wesley, 1969. LUNINE, T. E. The effects of judge's instructions and juror substitution on decisionmaking in a simulated jury. Unpublished Master's thesis, San Diego State Univer., 1979. ROSENTHAL, R. Experimenter ellects in behavioral research. New York: Appleton- Century-Crofts, 1966. Accepted December 16, 1981.