The “Hired Gun” Effect: Assessing the Effect of Pay, Frequency of Testifying, and Credentials on the Perception of Expert Testimony

Law and Human Behavior, Vol. 24, No. 2, 2000

The ‘‘Hired Gun’’ Effect: Assessing the Effect of Pay,Frequency of Testifying, and Credentials on thePerception of Expert Testimony

Joel Cooper1 and Isaac M. Neuhaus1

Three experiments addressed the proposition that jurors use short cuts in processinginformation when confronted with expert scientific testimony. The results of the firsttwo studies demonstrated that experts who are highly paid for their testimony andwho testify frequently are perceived as ‘‘hired guns.’’ They are neither liked norbelieved. The results of the third experiment replicated the hired gun effect and showedthat it is most likely to occur when the testimony is complex and cannot be easilyprocessed. The results were discussed in terms of the theoretical differences betweencentral and peripheral processing of persuasive messages in a legal context.

INTRODUCTION

In our increasingly litigious society, the role of expert scientific witnesses incourts of law has become correspondingly important. Ordinary citizens, serving intheir role as jurors, are often asked to make critical decisions on the basis of highlytechnical and scientific information. Product liability, wrongful death and personalinjury cases are typical of civil disputes in which jurors may be asked to sift throughtestimony from the physical and life sciences. Gross and Syverud (1991) reportthat, of the 529 civil trials they surveyed, 86% utilized expert testimony. More oftenthan not, jurors are subjected not only to the testimony of a single expert, butrather to opposing experts for the plaintiff and defense. Are jurors, who usuallyhave no special scientific training, capable of determining which expert to believeand which to discount? How do jurors make decisions when confronted with oppos-ing opinions, especially when the jurors do not possess the special technical knowl-edge that the expert witnesses possess?

Research in the social psychology of persuasion may provide the appropriateanalytical tools. A useful distinction is between what Chaiken (1980) calls systematicand heuristic processing of persuasive messages, or what Petty and Cacioppo (1986)

1Department of Psychology, Princeton University, Princeton, New Jersey 08540.

149

0147-7307/00/0400-0149$18.00/1 2000 American Psychology-Law Society/Division 41 of the American Psychology Association

150 Cooper and Neuhaus

refer to as the central and peripheral routes to persuasion. In the absence of eitherthe motivation or the ability to engage in thoughtful consideration of the contentof a persuasive message, people often resort to shortcuts or heuristics to helpthem assess the degree to which they should believe a particular message. In trialsinvolving expert scientific witnesses, who typically speak in language far moretechnical than most jurors are accustomed to, it is likely that jurors will engage inheuristic rather than systematic processing of the expert’s message. They will findsimple guides to help them make decisions about what they have heard. Of course,information from the message itself may serve as such a guide, but it is independentof the quality or accuracy of the message. For example, people may use the numberof arguments presented by a witness, rather than the quality of those arguments,to make an adjudication about the testimony.

Quite often, source cues are used for peripheral route processing. For example,the credentials or the appearance of the communicator may determine the persua-siveness of a particular message. Petty, Cacioppo, and Goldman (1981) found thatwhen people’s motivation to engage in central processing was low, they were morestrongly influenced by the expertise of the source making the argument rather thanthe content of the message itself. Ratneshwar and Chaiken (1991) asked subjectsto read a persuasive message that, in some cases, was too complex for them tocomprehend. When unable to process the message’s content, the subjects used theexpertise of the communicator to determine if they were persuaded by the message.Mackie and Worth (1989) found that when subjects were put under considerabletime pressure, they were more likely to use heuristic shortcuts rather than processthe message’s arguments.

Courtroom experts who serve as witnesses can also be thought of as communica-tors delivering persuasive messages to the audience—in their case, a jury. Due tothe specialized language and knowledge used by scientific experts, it is likely thatjurors will find their testimony difficult to comprehend and process (Cecil, Hans, &Wiggins, 1991). Based on Petty and Cacioppo’s (1986) and Chaiken’s (1980) distinc-tions, we can predict that jurors are likely to use peripheral cues in deciding onthe persuasiveness of a scientific expert’s testimony.

What kinds of cues will juries use to determine their belief in an expert’stestimony? One category of cues relates to the knowledge that an expert is likelyto have. Thus, information relating to the expert’s credentials may reflect on hisor her credibility and ultimately to the jury’s belief in his or her testimony. Cooper,Bennett, and Sukel (1996) showed that when mock jurors were exposed to highlycomplex scientific testimony in a civil case, their verdicts and beliefs about theissues were functions of the credentials of the experts. Despite identical testimony,a scientist from a highly prestigious university whose degree was also from a presti-gious university was significantly more effective in convincing a jury than a scientistwhose credentials were more moderate.

The use of heuristic cues for persuasion probably affects the persuasivenessof all of the witnesses who communicate to the jury. Their appearance, their dress,and their manner of speech are among the cues that may affect the degree to whichthey are effective communicators. However, expert witnesses differ from most otherwitnesses in criminal and civil trials on at least two dimensions: First, as already

The ‘‘Hired Gun’’ Effect 151

pointed out, their testimony may require special knowledge to be comprehended.Second, expert witnesses are paid.

Does the level of compensation affect the way expert testimony is regardedin the courtroom? In the wake of complex scientific testimony, jurors may seize onthe amount of money paid to the witness to assess the degree to which they believehis or her testimony. It is not immediately apparent which direction the use of payas a heuristic cue will take. It is possible that jurors will use a market economyheuristic and assume that an expert who is highly paid must be worth that amountof money. Such logic would lead to greater persuasiveness by a highly paid expert.However, the highly paid expert may be considered a ‘‘hired gun,’’ selling his orher testimony for the highest fee. A factor that causes a jury to suspect the motivationof the witness may reduce that witness’ persuasiveness (Hans & Ivkovich, 1994).

We conducted a study in which we orthogonally manipulated two potentialheuristic cues: the level of pay an expert received for his testimony and, consistentwith the study by Cooper et al. (1996), the level of credentials the expert possessed.Mock jurors listened to complex scientific testimony in a civil case and were sub-jected to one level of each variable. We predicted that, after listening to complexscientific testimony, jurors would be affected by level of credentials and by themagnitude of the fee the witness received for testifying.

EXPERIMENT 1

Method

Participants

Participants were 80 residents of central New Jersey. Twenty-seven were male;53 were female. They ranged in age from 18 to 65 years. All of the subjects werejury-eligible residents obtained from advertisements in local newspapers. Of thesubjects, 64% had previously been called for jury duty and 21% had served as jurors.All participants were paid $7 for their participation.

Procedure

Participants served in small groups of 2–7. They were told that they wereserving as mock jurors in a summary civil trial. In accord with actual practice inthe State of New Jersey, it was explained that courts often make use of partial or‘‘summary trials’’ because they provide a way of giving the plaintiffs and defendantsan opportunity to assess the strength of their cases. Judgments in summary trialsare not binding, but often lead to negotiated settlements prior to jury trials. Partici-pants were told that our interest was in understanding how jurors make decisionsin such summary trials.

Participants were told that the case they were going to hear was an ideal casefor summary trial because both sides had agreed to the facts of the case and therewas only one issue to be resolved. As they would soon discover, the case revolvedabout the scientific issue of whether a chemical to which the plaintiff had been


exposed was the immediate cause of his cancer. We explained that, as mock jurors,they would be told about the stipulated facts, given a brief summary of the openingand closing arguments, and then listen to the crucial testimony from the scientistson the single, contested question at trial. They were asked to behave as if theywere jurors deciding the case. Participants were instructed to attend closely to theevidence in order to come to a reasoned and rational decision.

The testimony was adapted from a case created by Cooper, Bennett, and Sukel(1996) that was based on a series of product liability cases in New Jersey concerningexposure to toxic substances. It was introduced to mock jurors as Mark A. Stevensv. Craven Laboratories, 1995. The researcher explained the context of the case tothe mock jurors. The jurors were told that Mr. Stevens had worked for CravenLaboratories for 10 years. During that time, he had conducted chemical research forthe laboratory, and that he had worked directly with the chemical polychlorinatedbiphenyls (PCBs). In that time, he contracted colon cancer. Plaintiff was suingdefendant, alleging that the exposure to PCBs was the proximal or direct cause ofhis cancer. Mock jurors were told that the defense had stipulated to all of the factsand these would not be contested at trial. Defendant had agreed that Mr. Stevensdid have colon cancer, that he did work at the laboratory, and that there werePCBs in the lab and in the soil around the laboratory. They did not contest theclaim that he had been exposed to PCBs nearly every day that he worked in thelaboratory. They did contest the claim that exposure to PCBs causes cancer and,specifically, that PCBs were the proximal cause of Mr. Stevens’ cancer. Mock jurorswere instructed to listen to the key testimony in the summary trial: expert witnessesarguing for and against the claim that PCBs were the proximal cause of Mr. Stevens’cancer. They were told that, in the end, they would be asked to render a judgmentfor plaintiff or defendant, depending on whether they believed that exposure toPCBs caused Mr. Stevens’ cancer.

Mock jurors heard a 30-min audiotape of expert testimony that they believedhad been delivered at the summary trial. Dr. Thomas Fallon, a biochemist, testifiedfor the plaintiff. His testimony was presented on direct examination from plaintiff’sattorney. He presented biochemical evidence from the research literature thatimplicated PCBs in causing cancer. Arguing for the defense, Dr. William Campbell,responding to direct examination from the defendant’s attorney, presented evidencethat indicated PCBs did not produce cancer and thus could not be the proximalcause of Mr. Stevens’ illness.

In reasonably complex testimony, the experts for each side presented theirsides of the argument. The testimony began with responses to questions askingabout the expert’s background and credentials. Then, in response to the directquestion, ‘‘In your opinion, did Mark Stevens contract cancer from his exposureto PCBs?’’ Dr. Fallon responded, ‘‘I believe that there is a very high probabilitythat Stevens contracted cancer as a result of his exposure to PCBs.’’ Respondingto 42 questions from the plaintiff’s attorney, Dr. Fallon went through the logic ofhis answer. He stressed the relationship between PCBs and cancer in research onanimals, then explained the relevance of animal studies to humans. In response tothe question, ‘‘What can these animal studies tell us about the effects of PCBs onhumans?’’ Dr. Fallon responded, ‘‘I don’t think there is any scientist out there who


can seriously challenge the idea that animal studies can tell us a lot about the effectsof chemicals on humans. . . . Almost every substance that has been shown to causecancer in humans was first shown to be a cause of cancer in animals.’’ Dr. Fallonthen went on to rule out the possibility of other probable causes of cancer. Theplaintiff did not smoke and did not come from a family with genetic predispositionto cancer. Finally, Dr. Fallon was asked, ‘‘can you give us your expert opinion asto the cause of Mr. Stevens’ colon cancer?’’

‘‘In my expert opinion,’’ he responded, ‘‘Mark Stevens’ colon cancer was causedby exposure to PCBs at the Craven Laboratories.’’

All of the mock jurors then heard from Dr. Campbell, the expert for thedefendant. His testimony was also elicited in response to 42 questions asked by theattorney for Craven Laboratories. The gist of his testimony was epidemiologicrather than biologic in order to avoid direct contradiction of factual aspects of thetestimony. He stated that PCBs have not been shown to cause colon cancer inhumans. Using epidemiologic data, he claimed the evidence suggests that PCBs dolead to liver damage and skin irritation, but not cancer. He dismissed Dr. Fallon’sanimal studies as invalid because rats and humans are completely different species.In the end, in response to defense attorney’s direct question, Dr. Campbell con-cluded, ‘‘I would say that it is highly unlikely that exposure to PCBs resulted inthe plaintiff’s illness.’’

The Pay Variable. In both the high and low-pay conditions, subjects heardthe plaintiff’s attorney ask Dr. Fallon, the plaintiff’s witness, ‘‘Doctor, could youtell the court how much you are being compensated for your appearance in courttoday?’’ In the high-pay condition, Dr. Fallon replied, ‘‘I am being paid $4,800 forthe time I’ll spend in court today.’’ In the low-pay condition, he stated that he wasbeing paid $75 for his time.

In order to set a standard, the defense’s witness, when asked the same question,always responded that he was earning $600 for his testimony. Although the numbersare arbitrary, they were established to make the highly paid expert’s fee eight timeshigher than the defense’s expert, while the low-paid expert received payment thatwas one-eighth the fee of the defense expert.

Credentials Variable. The level of credentials of the plaintiff’s expert wasvaried in a manner similar to the procedure of Cooper et al. (1996). In the high-credentials condition, Dr. Fallon’s Ph.D. in biochemistry came from a prestigiousinstitution, he was the author of several dozen articles, and he taught at anotherwell-known, prestigious institution. By contrast, in the modest-credentials condition,the institutions from which Dr. Fallon earned his degree, and at which he currentlytaught, were more obscure. He also had published far fewer articles. This informa-tion came in response to the initial set of questions posed to the witness by the at-torneys.

The defense expert’s credentials were established to be more impressive thanthe modest-credential expert, but less impressive than the high-credential expert.

Following the presentation of testimony, the mock jurors were asked to rendera verdict for Mark Stevens or for the Craven Laboratories. They were then askedto rate their confidence in that verdict on a 9-point scale and also asked to indicatethe probability that PCBs were the proximal cause of Mr. Stevens’ cancer. Partici-


pants then answered questions that measured specific attributions that they assignedto the experts. These questions measured the attorneys’ persuasiveness, likeability,believability, honesty, and trustworthiness. In order to assess how closely they paidattention to the testimony, subjects were shown a list of arguments that wereplausibly related to the scientific arguments. Subjects were asked to identify whicharguments had and had not actually been used by the experts during their testimony.

Participants were administered a thought-listing task consistent with the proce-dures introduced by Petty and Cacioppo (1986). They were asked to list as manythoughts that they could recall as they listened to the trial. They were asked to listall of the thoughts they had, whether relevant to the case or not. They were given3 min to complete this task.

Finally, participants were given a questionnaire that asked for their impressionsof the pay that the experts received. They were asked to recall which expert waspaid more. Then, they were asked if they thought that each of the expert’s testimonywas influenced by the money he received. Finally, they were asked an open-endedquestion, asking subjects to ‘‘explain if the money that each expert paid factoredinto your verdict.’’

Results

Checks on the Manipulation

Because there seemed little opportunity in the course of the study to ask mockjurors about their perception of the credentials of the witnesses, a small pilot testwas run with a separate sample of 10 participants drawn from the same populationas the mock jurors, with half run in the high-credentials and half run in the moderate-credentials conditions. Pilot subjects listened to the same testimony that was pre-sented in the audio stimulus tape. Consistent with the intended manipulation, Dr.Fallon was rated as a ‘‘better’’ expert in the high-credential compared to the moremoderately described Dr. Campbell, t(8) � 8.22, p � .0001. Fallon was also consid-ered to have more expertise than Campbell, t(8) � 3.83, p � .01. However, partici-pants who heard Dr. Fallon described in the modest-credential condition rated himas having less expertise than Dr. Campbell, t(8) � 9.0, p � .0001, and saw Dr.Campbell as the better expert, t(8) � 11.23, p � .0001.

The mock jurors in the main study were asked to recall which expert was paidmore. All jurors recalled correctly.

The Verdict

Mock jurors were asked to render their verdict individually. Overall, the casewas perceived evenly, with 52.8% of the subjects voting for the plaintiff. However,inspection of the verdict data presented in Fig. 1 reveals that the credentials andpay of the plaintiff’s expert made a difference in jurors arriving at their decisions.

The pay of the expert did not affect his testimony in any straightforward way.Overall, the highly paid expert was slightly less effective at convincing the jury(44% vs. 53%), but this difference was not significant. What apparently affected thejury was the combination of pay and credentials. The highly credentialed expert who


Fig. 1. Proportion of mock jurors voting for the plaintiff as a function of the pay and credentials ofthe expert.

was paid eight times what the defense’s expert was paid was markedly ineffective atgetting the jury to favor his client, securing a favorable verdict only 29% of thetime. When the expert with modest credentials was paid handsomely for his time,62% of the mock jurors sided with his client. Log linear analyses of variance forproportions (Fienberg, 1980), which is distributed as �2, was used to assess the data.As predicted, this analysis revealed that the pay � credentials interaction wassignificant, �2(1, N � 80) � 5.96, p � .05. In addition, the verdicts in high-credentials,high-pay condition were significantly different from the verdicts in the other threeconditions, �2(1, N � 80) � 6.02, p � .05.

Verdict Confidence

A scalar variable was created by using the subjects’ verdict and confidence inthose verdicts. The mock jurors were asked to rate how confident they were in theverdict they had given on a 1–9 scale. A positive score was given for a plaintiffverdict, whereas a verdict in favor of the defense was given a negative value. Thusverdict confidence ratings ranged from �9 (extremely confident that the defendantwas correct) to �9 (extremely confident the plaintiff was correct).


The results are presented in the first row of Table 1. The two-way pay �credentials ANOVA showed that the interaction between pay and credentials wassignificant, F(1, 76) � 6.98, p � .01. Contributing to the interaction was the highconfidence jurors had in the high-credentials, high-pay condition to vote for thedefense’s position. The planned comparison assessing the difference between themean in the high-pay, high-credentials condition and the combined mean of theother three conditions was significant, F(1, 76) � 5.25, p � .05.

Probability That PCBs Caused Plaintiff’s Cancer

The plaintiff’s main argument was that exposure to PCBs at the Craven Labora-tories was a direct and proximal cause of his cancer. Subjects expressed their beliefin the probability that Stevens’ cancer was caused by the exposure. The resultspresented in the second row of Table 1 show a similar pattern to the previous data,but are somewhat attenuated. The interaction of pay and credentials in the 2 � 2ANOVA was significant, F(1, 76) � 4.09, p � .05. The high-pay, high-credentialsexpert was least effective at convincing the jury of his argument. Jurors believedthat it was 31.8% likely that PCBs were the proximal cause of cancer when thehighly paid, highly credentialed expert was the source of the persuasive messagecompared to an average belief of 44.9% for the other communicators, F(1, 76) �4.41; p � .05. Within the high-pay condition, individual comparisons showed thatthe difference for credentials was highly significant, F(1, 76) � 8.33; p � .01.

Perception of Influence of Pay on Testimony

Mock jurors were asked how much they believed the pay that each expertreceived influenced his testimony. Participants reported their opinion using a 1 (lowinfluence) to 9 (heavy influence) scale. The third row of Table 1 reports the ratingsgiven for Dr. Fallon (who received $4,800 in the high-pay and $75 in the low-payconditions) and Dr. Campbell (whose fee was always $600.)

Two-way analysis of variance revealed a significant main effect for pay forboth witnesses. It was thought that the plaintiff’s witness was influenced by moneyas a direct effect of its magnitude. The witness paid $4,800 was seen as having his

Table 1. The Effects of Pay and Credentials (Experiment 1)

High pay Low pay

High Modest High Modestcredentials credentials credentials credentials

Verdict confidence �3.10a 1.92bc 0.56c 0.53c

Probability that PCBs caused .318a .546c .404b .396ab

cancerPerceived influence of pay on 5.29b (2.50d) 4.67bc (2.25d) 2.46d (4.77b) 2.85c (6.77a)

expert’s testimonya

Number of pay-oriented 0.71a 0.15b 0.15b 0.85a

thoughts

Note. Within each row, entries with different subscripts are different from each other by tests of plannedcomparison, p � .05.aScores for plaintiff’s expert are listed first. Scores for defense’s expert are in parentheses.


testimony influenced by pay more than the witness paid $75, (M � 4.88 vs. M �2.60), F(1, 76) � 8.44, p � .01. The defense’s expert was seen as being moreinfluenced by pay when his $600 fee was juxtaposed with the plaintiff’s expert’ssmaller fee (M � 2.38) compared to when it was juxtaposed with the plaintiff’sexpert’s large fee (M � 5.77), F(1,76) � 12.07, p � .01. Thus, the assumption thatthe expert’s testimony was influenced by money was leveled against the expertearning the larger fee rather than by the absolute value of that fee. Interestingly,confidence in deciding for the plaintiff was only slightly correlated with the influencepay had on Dr. Fallon’s testimony, r � .18, p � ns. The belief that the expert’stestimony was influenced by pay was not isomorphically related to jurors’ decisionsto vote against the ‘‘bought’’ expert.

Thought Listing

Participants’ thought listings were categorized as containing message-relevantand message-irrelevant thoughts. Planned comparisons showed that participants inthe high-pay, high-credential condition reported more irrelevant thoughts than didparticipants in the other three conditions, M � 6.0 vs. M(combined) � 4.2;F(1, 76) � 8.88, p � .01. However, this must be interpreted cautiously because theinteraction between pay and credentials did not quite reach statistical significance,F(1, 76) � 3.81, p � .10.

The subjects’ thought listing was also examined to assess the number of thoughtsparticipants specifically expressed about the experts’ pay. The results are presentedin the last row of Table 1. Mock jurors had more thoughts about pay in the high-pay, high-credentials condition and in the low-pay, modest-credentials conditionthan in the other two conditions. The two-way ANOVA showed that this producedan interaction of pay and credentials that was highly significant, F(1, 76) � 12.73,p � .001. Inspection of the content of the thoughts showed that mock jurors werefocused on the plaintiff’s expert in the high-pay, high-credentials condition and onthe defendant’s expert’s pay in the low-pay, modest-credentials condition.

The list of thoughts was also examined to assess attributions of the witnesses.However, there was not a sufficient number of thoughts using attributional languageto enable satisfactory analysis.

Attributions to the Witnesses

The mock jurors were asked to rate the experts’ characteristics on a numberof dimensions. The results are described in Table 2. Two-way pay � credentialsANOVAs performed on the data revealed significant differences on attributionsof trustworthiness, honesty, persuasiveness, believability, and likeability and on arating of how annoying the expert was perceived to be.

When Dr. Fallon received a large fee for his testimony, main effects showedhe was perceived as less trustworthy, F(1, 76) � 4.88, p � .05, and more annoying,F(1, 76) � 9.61, p � .01, than when he received a low fee. The specific combinationof high pay and high credentials rendered the expert less likable, F(1, 76) � 5.33,p � .05, less believable, F(1, 76) � 4.89, p � .05, less trustworthy, F(1, 76) � 8.84,p � .01, less honest, F(1, 76) � 5.26, p � .05, and more annoying, F(1, 76) � 10.18,


Table 2. Attributions of Plaintiff ’s Expert as a Function of Pay and Credentials (Experiment 1)

High pay Low pay

Attribution High credentials Modest credentials High credentials Modest credentials

Trustworthy 6.9a 8.0ab 11.3b 11.4b

Honest 6.6a 9.6ab 12.0b 11.4b

Persuasive 7.8a 9.0a 12.7b 10.9ab

Believable 7.5a 10.1ab 12.3b 10.2ab

Likeable 7.0a 10.6b 11.0b 11.1b

Annoying 12.2a 10.0ab 9.3b 9.1b

Note: Entries with different subscripts are different from each other by Neuman–Keuls analysis, p � .05.

p � .01, than the expert in the other three conditions. Because of the number ofcomparisons on the multiple attribution measures, post hoc tests of individual celldifferences were conducted using the Neuman–Keuls procedure and are reportedin Table 2.

Many of the attribution measures were significantly related to the jurors’ ver-dict-confidence rating. Verdict confidence was correlated significantly with the ex-pert’s likeability, r(78) � .35, trustworthiness, r(78) � .50, honesty, r(78) � .53,and believability, r(78) � .73. The combination of pay and credentials, therefore,were associated with the expert’s being perceived as more likeable, trustworthy,honest, and believable and those attributes in turn were correlated with the partici-pants’ verdict confidence.

Discussion

Does pay serve to reduce a witness’ effectiveness? The results of the firstexperiment reveal that it does, but in an interesting way. When given the opportunityto discount or augment an expert witness’s testimony when confronted with informa-tion about pay, subjects seemed to engage in complex processing. Rather than viewthe pay as a cue in itself, they combined that information with other informationto arrive at a conclusion. Witnesses who received high pay and who had highcredentials were not believed, not liked, and were not effective. However, the samehigh pay combined with less impressive credentials rendered the witness as effectiveas his less well-compensated counterpart.

What did the high pay combined with high credentials mean to our participants?During the debriefing session, many of our participants expressed the opinion thatthe expert from the highly prestigious institution who was also highly paid for histestimony probably testified frequently as a paid witness. By contrast, the jurorsseemed to assume that the highly paid expert with the modest credentials waschosen because he had expertise that was crucial for the case.

We asked 10 additional Princeton University students to listen to each of thetape-recorded conditions of Dr. Fallon’s testimony. They were asked one question:How many times do you think Dr. Fallon has testified in court? Those who listenedto Dr. Fallon in the high-pay, high-credentials condition estimated an average of14.0 times. In the high-pay, moderate-credentials; low-pay, high-credentials, and


low-pay, moderate-credentials conditions, it was estimated that Dr. Fallon testified1.8, 2.1, and 1.7 times, respectively. From this modest pilot test, it appeared thatthe scientist with high credentials who comes to court compensated handsomelywas the one perceived as the ‘‘hired gun.’’ This is the expert who jurors suspecthires him or herself out to the highest bidder and this is the expert who is notsuccessful at persuading his or her audience.

EXPERIMENT 2

We conducted a second study in order to assess the impact of the frequencyof the expert’s court appearances. We systematically varied information the juryhad about the number of times the experts had testified in court as well as themagnitude of the expert’s pay. The method was similar to the first experiment,except that all subjects were run in the high-credential condition. The magnitudeof pay was varied as before and the number of times the expert had previouslytestified was manipulated to be as close as possible to the means provided by theuniversity subjects in the pilot study. This resulted in a two- (frequency of testifying)by-two (magnitude of pay) factorial design.

Method

Participants

Forty residents from central New Jersey communities volunteered for a studyon ‘‘jury processes.’’ They ranged in age from 19 to 72 years. All of the participantswere jury-eligible.

Procedure

The procedure followed the outline of experiment 1. Participants were toldthat they would serve as mock jurors in a summary trial of Mark A. Stevens v.Craven Laboratories, 1993. The meaning of a summary trial was explained and thecontext of the experts’ testimony was established just as in the first experiment.Doctor Fallon was the witness for the plaintiff; Dr. Campbell was the witness forthe defense.

All of the participants heard Dr. Fallon described with high credentials as inthe high-credential condition of experiment 1. In response to the introductoryquestions asked by plaintiff’s attorney, Dr. Fallon responded that he was earningeither $75 or $4,800 for his time. He was then asked, ‘‘Professor, how many timeshave you testified in either a civil or criminal trial?’’ In the novice-testimony condi-tion, Fallon answered that this was his second time testifying. In the Frequent-testimony condition, he responded that this was his 14th appearance.

As in experiment 1, Dr. Campbell was presented as having moderate credentialsand moderate pay ($600). He was not asked how many times he had testified. Itwas decided not to choose a number of appearances for the defense’s witness forlack of a proper metric. We cannot be certain that the number of times testifying


is perceived to be linear rather than there being a number above which participantsconsider frequent and below which they perceive to be novice.

Participants were then asked for a verdict and their confidence in their verdict,and were asked to rate the probability that exposure to PCBs was the proximalcause of Mr. Stevens’ cancer. Participants were also asked to rate Dr. Fallon andDr. Campbell on a series of trait dimensions that included honest, trustworthy,believable, likeable, and annoying. They were asked to rate how ‘‘good’’ eachwitness was. Finally, as checks on the perception of the manipulated variables,participants were asked to recall how many times Dr. Fallon indicated that hetestified in court, how much money each witness received, and how many timesthey believed the defense witness, Dr. Campbell, had testified previously.

Results

Check on the Manipulation

Almost all participants correctly recalled the number of times the novice andthe professional had testified. Eighteen of the 20 participants in the novice conditionand 17 of 20 in the frequent condition recalled the numbers accurately. The meannumber of occasions testifying was recalled as M � 13.4 for the frequent testifierand M � 2.2 for the novice.

Participants also recalled the amount of pay effectively. High-pay participantsrecalled Dr. Fallon’s pay as M � $4,548, while low-pay participants recalled hispay as M � $88, t(39) � 4.31, p � .001. They recalled Dr. Campbell’s pay as M �$665. The two pay conditions did not differ in their recollection of Campbell’s fee,t(39) � 1.

Dependent Measures

The participants in each of the four conditions (n � 10) were asked to votefor the plaintiff or the defendant in Stevens vs. Craven. The results are presentedin Fig. 2. These data replicate and extend the results of experiment 1. The verdictdata reveal a significant main effect for professional experience. Log linear ANOVAperformed on the data show that expert witnesses who had only appeared oncebefore (novice condition) were believed significantly more than those who testifiedregularly, �2(1, N � 40) � 5.09, p � . 05. There was no main effect for pay, butpay interacted with the frequency variable, �2(1, N � 40) � 4.82, p � .05 Whenthe expert was paid a large amount of money in the professional witness condition,he was not an effective communicator for his position. In such circumstances, theverdict went for the plaintiff only 25% of the time. However, the highly paid expertwas considerably more effective when he was testifying as a novice, resulting in averdict for the plaintiff 58.3% of the time. In the low-payment condition, the profes-sional witness was also less effective than the novice, but this difference was quitesmall (41.7% v 50%).

The difference found for the verdict-dependent measure was paralleled by theverdict-confidence measure. The scalar variable comprised of the confidence andverdict data were submitted to a 2 � 2 ANOVA. The analysis reveals a highly


Fig. 2. Effect of pay and frequency of testifying on verdicts for plaintiff.

significant interaction between frequency of testifying and the magnitude of pay,F(1, 36) � 9.52, p � .01. Individual comparison tests reported in Table 3 show thatthe expert who produced the most confidence for the plaintiff’s position was thehighly paid, novice expert (M � 2.11), whereas the least confidence in the plaintiff’sposition was provided by the highly paid, frequently testifying expert (M � �3.07).

Similarly, participants’ estimation of PCB exposure being the proximal causeof cancer also reveals the interaction pattern between pay and frequency. Table 3shows that the mean estimate for all participants of the probability that PCBs were

Table 3. The Effect of Pay and Experience on Verdict, Verdict Confidence, andOpinions About PCBs Being the Proximal Cause of Cancer (Experimental 2)

Frequent testifier Novice testifier

Dependent measure High pay Low pay High pay Low pay

Verdict confidencea �3.07a 0.51b 2.11c 0.63b

Proximal causea .25a .42b .52c .48bc

Attributionsb

Honesty 5.3a 9.2b 9.0b 9.6b

Trustworthy 6.1a 10.3b 10.6b 9.9b

Believable 7.0a 10.1b 10.7b 10.8b

Likeable 5.8a 9.7b 9.4b 9.5b

Annoying 11.1a 8.3b 8.0b 7.7b

aEntries with different subscripts are different from each other by planned compari-sons, p � .05.

bFor attribution measures, entries with different subscripts are different from eachother by Neuman–Keuls tests, p � .05.


the proximal cause of Mr. Stevens’ cancer was M � .44. The highly paid, frequentlytestifying expert who spoke on the plaintiff’s behalf produced a mean estimate ofonly M � .25, whereas the highly paid novice produced an estimate of M � .52.Experts who were not highly paid produced estimates close to the grand mean(M � .42 for the frequent testifier; M � .48 for the novice expert). ANOVA showedthat the combination of pay and frequency produced a highly significant statisticalinteraction, F(1, 36) � 9.22, p � .01.

Measures of attribution also paralleled the results of experiment 1. Plannedcomparisons showed that participants in the high-pay, frequent-testimony condi-tion were perceived as less trustworthy, F(1, 36) � 5.82, p � .05, less honest,F(1, 36) � 5.25, p � .05, less believable, F(1, 36) � 4.99, p � .05, and less likeable,F(1, 36) � 9.75, p � .01, than participants in the combination of the other threepay and frequency conditions. The less likeable, r(38) � �.41, honest, r(38) ��.52, trustworthy, r(38) � �.56, and believable, r(38) � �.39 the expert was, theless likely the mock jurors were to side with the expert’s position.

Discussion

The results of the second experiment show that mock jurors discount thetestimony of an expert who receives high pay, but only if other information suggestshe is an expert for hire—that is, a ‘‘hired gun.’’ For the highly paid, frequenttestifier, the rate of pay implies that the expert is a hired gun. In addition, the mockjurors make personal attributions to the hired gun expert, indicating that they didnot like him, found him less believable, less honest, and less trustworthy. Moreover,the attributional inferences were all correlated with the verdict-confidence measure.The less likeable, honest, trustworthy, or believable the expert was, the less likelythe mock jurors were to side with the expert’s position.

We have argued that our participants were engaged in peripheral processingand thus used cues such as pay, credentials, and frequency of testifying as a guideto persuasion. This is a theoretical claim that cannot be supported unequivocallyfrom the conditions in the first studies. We know that participants used the cuespresented in order to discount the testimony of the expert they perceived as a hiredgun. We do not know, however, whether mock jurors in all conditions engaged inperipheral processing or whether only the participants in the high-credential, high-pay condition of experiment 1 and the high-pay, frequent-testifier condition ofexperiment 2 engaged in peripheral processing. In other words, it is possible thatthe circumstances that created the impression of the hired gun short-circuited thesystematic processing that mock jurors may have engaged in the other conditions.

Conceptually, people are thought to engage in peripheral processing whenthey lack the motivation or ability to process centrally. We intended to push ourparticipants toward peripheral processing through the use of a relatively complexmessage, thus blocking their ability to engage in central processing. However,because we only had one level of complexity of the testimony, it is difficult to knowwhether participants had the ability to engage in message elaboration throughcentral processing. The following experiment remedies this problem. It is designedto bolster the theoretical assumption that the use of cues that imply an expert is a


hired gun only occur when mock jurors engage in peripheral rather than central pro-cessing.

Another problem that our third study seeks to address is whether participantsmay have made assumptions about the experts not only because of their high orlow fees, but also because of the exaggerated magnitude of those fees. A scientistwho was paid only $75 for his court appearance may have his ability, standing, orjudgment questioned. Similarly, a fee of $4,800 for testifying for a day in court maybe seen as so extraordinarily large that jurors needed to find some explanation,regardless of whether they were engaging in central or peripheral processing. Inthe third experiment, we established the peripheral cue in a manner more in linewith what might be expected in a court of law. Moreover, we used three magnitudesof pay in order to examine if attributions and verdicts approximated a linear functionof the magnitude of pay.

EXPERIMENT 3

A third experiment was conducted to assess the effects of pay and frequencyof testifying in conditions in which the conceptual distinction between peripheraland central processing could be assessed. This was accomplished by having twosets of testimony, one straightforward and easy to understand, the other morecomplex and difficult to understand. We predicted that the effects we found forthe joint impact of pay and frequency would occur when the testimony was complex,but would not occur when the testimony was easier to comprehend. In the lattercase, we expected to find fewer differences between the pay and frequency variablesbecause we predicted that participants would use central processing to assess thewitness’ testimony. Using central processing, mock jurors were expected to basetheir judgment on the substance of the testimony rather than the informationprovided by the peripheral cues of expertise and frequency.

Method

Participants

Sixty jury-eligible citizens of central New Jersey between the ages of 19 and53 years volunteered for the experiment. Twenty-six were male; 34 were female.As in the previous studies, they responded to advertisements in local newspapers.They were paid $7 for their participation.

Procedure

Participants served in groups of 3–6. As in the prior studies, they were toldthat they were serving as mock jurors in a summary civil trial. A summary of thefacts of the case and a summary of the opening arguments were presented. Asbefore, the mock jurors understood that the issue at trial was whether PCBs,stipulated to be present at the Craven Laboratories, were the proximal cause ofplaintiff’s cancer.


Varying Complexity. In order to manipulate the probable use of central versusperipheral processing, two different versions of plaintiff’s expert’s testimony wereused. It was reasonable to expect that if the testimony was extremely complex, thejurors would be inclined to use peripheral processing. If the testimony were madesimple, the jury—assuming they were so motivated—would be more able to engagein central processing. If they used central processing to elaborate and comprehendthe expert’s message, then cues to the expert’s frequency of testifying and magnitudeof pay would not be relevant and would not affect his persuasiveness. Fortunately,a complex and a simple version of the expert testimony were already availablefrom the study by Cooper et al. (1996). These versions were adopted for the presentstudy. Thus, the study consisted of a more complex version of Dr. Fallon’s testimonythan used in the first two studies and a simpler version of that testimony.

Following the procedure of the Cooper et al. (1996) study, plaintiff’s expert’stestimony in the Stevens v. Craven case was constructed such that all versions hadthe same substantive meaning, but varied on the linguistic complexity used toconvey the information. In the high-complexity version, the witness used specializedtechnical jargon and, in response to questions posed by the lawyer, responded withsimilarly technical jargon. In the low-complexity condition, the witness used agreater proportion of lay rather than scientific terms. The word count of the low-and high-complexity testimony was virtually equal and the organization of theparagraphs was identical. Four graduate students in molecular biology examinedthe content of the low- and high-complexity testimony and judged them to conveyequivalent information.

To illustrate the difference in the linguistic complexity of the two versions,plaintiff’s expert was asked whether there had been other studies conducted tomeasure the effect of PCBs on animals. In the low-complexity condition, the expertresponded as follows:

Definitely. In 1980, a scientist named McConnell, published a summary of the diseasesthat PCBs cause. He found that PCBs caused several different forms of liver disease inrats, mice, monkeys, and humans. In the rats and mice, PCBs caused not only liver disease,but also cancer of the liver. In addition to the liver damage, McConnell found diseasesof the immune system as well.

In the high-complexity condition, the expert responded as follows:

Definitely. In 1980, McConnell, publishing in the Elsevier Biomedical Press, reported asummary of the pathological findings due to the toxicity of PCBs. He reported tumorinduction in rats and mice. He also reported that not only rats and mice, but in monkeysas well, there was hepatomegaly, hepatomegalocytosis, and lymphoid atrophy in bothspleen and thymus.

The testimony of the defense’s expert did not vary by condition and, consistentwith the procedure of Cooper et al. (1996), was designed to be equal in complexityto that of the plaintiff in the high-complexity condition.

The Pay Variable. In the prior studies, the levels of pay were set extremelyhigh and extremely low relative to typical pay for a court expert. In the currentstudy, we use three levels of pay ranging from $750/hr for the highly paid expert,to $350/hr for the moderately paid expert, to $100/hr for the low-paid expert. Dr.Campbell, the defense’s expert, was always paid at the moderate rate of $350/hr.


These figures were obtained from several scientists who had served as expert wit-nesses.

In experiment 3, we held constant the level of credentials and the frequencyof Dr. Fallon’s court appearances. As in experiment 2, the credentials were set ata high level (consistent with the high credentials of experiment 1) and the frequencyof testimony was fixed at the high frequency level (14 prior experiences testifyingin court) used in the second experiment.

Mock jurors heard the testimony of the expert witnesses in the same manneras in the first two experiments. Following the presentation of the testimony, theywere asked to render a verdict for plaintiff or for the defendant. They were askedto rate their confidence on a 9-point scale, and were also asked to rate the degreeto which they believed that PCBs were the proximal cause of the plaintiff’s cancer.The mock jurors then responded to a series of questions that attempted to assessthe attributions that were made to the expert witnesses. Participants were askedto rate the experts’ honesty, trustworthiness, expertise, credibility, believability,and likeability on 15-point scales. The rating of ‘‘expertise’’ was added in thisexperiment because it is usually thought to be a part of the judgment of credibility.It had not been directly assessed in the prior two experiments. On an open-endedquestion, they were asked to explain the reasons that the plaintiff’s expert had fortestifying in this case. They were asked, ‘‘Why do you think Dr. Fallon testified incourt today? List as many reasons as you think appropriate. Please list the mostimportant reason first, the next most important reason second, and so forth.’’Participants were asked the same questions about Dr. Campbell, the expert whotestified for the defense.

Finally, the mock jurors were asked to respond to checks on the manipulationand were carefully debriefed.

Results

The Verdict

The verdict that the mock jurors reached is presented in Fig. 3. In the high-complexity condition, the data show that the expert who was paid a small amountfor his court appearance persuaded the jury to side with the defendant 57.3% ofthe time. This was not significantly different from the verdict reached in the low-complexity, low-pay condition (54.1% for the plaintiff). When pay was at a moderatelevel, high-complexity participants voted only 42.6% for the plaintiff and as payreached the high level, the verdict for the plaintiff in the high-complexity conditionwas only 19.1%. Log linear ANOVAs showed each of these differences to besignificant (p � .05 for high-complexity, low-pay vs. high-complexity, moderate pay,conditions; p � .001 for high-complexity, moderate-pay vs. high-complexity, low-pay conditions). In the low-complexity conditions, pay made no significant differ-ence. The percentages of mock jurors voting for the plaintiff when pay was low,moderate, and high were 54.1%, 55.3%, and 49.3% respectively. As predicted, thelog linear ANOVA revealed that the interaction of pay and complexity was signifi-cant, �2(2, N � 60) � 6.69, p � .01.


Fig. 3. Effect of witness’ pay and complexity of testimony on verdict for plaintiff.

Verdict Confidence

The results of the verdict-confidence measure are presented in Table 4. Confi-dence was rated on a 9-point scale and multiplied by �1 when the verdict was forthe plaintiff and �1 when the verdict was rendered for the defendant.

As can be seen, the results parallel the verdict measure. The results of

Table 4. Effects of Complexity of Testimony and Magnitude of Pay on Verdict Confidence, the ProbabilityThat PCBs Are the Proximal Cause of Plaintiff ’s Cancer, and Attributions to the Expert (Experi-

mental 3)

High complexity Low complexity

Low Moderate High Low Moderate HighMeasure pay pay pay pay pay pay

Verdict confidencea 2.47a �0.31b �3.42c 2.43a 2.55a 1.11ab

Probabilitya .62a .41b .27c .59a .53ab .48ab

Attributionsb

Trustworthy 9.8a 8.5a 6.2b 10.3a 9.9a 9.4a

Expertise 12.4a 12.1a 12.0a 11.3a 10.7a 9.8a

Credibility 11.3a 9.6b 5.6c 11.6a 10.5ab 10.4ab

Honest 10.3a 9.7a 5.9b 10.6a 10.1a 9.3a

Believable 11.0 10.3 7.4 10.2 9.7 9.4Likeable 8.4a 8.1a 6.6c 9.7b 10.0b 9.2ab

Annoying 9.7a 9.8a 10.1a 8.9a 9.4a 9.5a

aEntries with different subscripts differ from each other by tests of planned comparisons, p � .05.bFor attribution measures, entries with different subscripts differ from each other by Neuman–Keulstests, p � .05.


the two-way ANOVA show a significant main effect for the magnitude of pay,F(1, 55) � 4.97, p � .05, and a significant main effect for complexity, F(1, 55) �8.78, p � .01. These main effects are qualified by a highly significant interaction,F(1, 55) � 8.13, p � .01. The results of simple effects tests show that when complexitywas low, the magnitude of pay made no difference. However, when complexity washigh, mock jurors’ verdict confidence was indeed a function of pay. When complexityrendered the argument difficult to understand, the magnitude of pay was apparentlyused as an indicant of the expert’s believability.

Persuasion

Table 4 also shows the level of belief in plaintiff expert’s argument. As in theprior experiments, belief in whether PCBs are the proximal cause of the plaintiff’scancer parallel the verdict and verdict-confidence scores. The main effect for paywas significant, F(1, 55) � 5.11, p � .05, as was the main effect for complexity,F(1, 55) � 4.92, p � .05. Most important, the interaction between pay and complexitywas significant, F(1, 55) � 6.25, p � .05 such that the magnitude of pay made asignificant difference when the complexity of the arguments was high but not whenthe complexity was low.

Attributions

After they rendered their verdicts, their confidence, and their belief in whetherPCBs caused Mr. Stevens’ cancer, the mock jurors were asked to rate the expertson several dimensions. The participants were discerning on the use of the traitadjectives, revealing some interesting differences. The higher the pay, the lesstrustworthy they found the expert, F(1, 55) � 5.31, p � .05. In addition, plannedcomparisons showed that the highly paid expert who gave highly complex testimonyhad lower trustworthiness scores than did experts in the other three conditions ofthe experiment, F(1, 55) � 5.22, p � .05, although the overall interaction was notsignificant. However, perceived expertise did not vary as a function of pay. Therewas a tendency toward a main effect for complexity such that highly complextestimony made the witness seem more expert (M � 12.1) than low-complexitytestimony (M � 10.6), but this was not significant. Perhaps the expert’s impressivecredentials mitigated the differences on this measure. Credibility was also ratedand paralleled the main dependent measures on the verdict, verdict-confidence,and persuasion. ANOVAs revealed main effects for complexity, F(1, 55) � 4.37,p � .05, and for pay, F(2, 55) � 5.52, p � .05, and a significant interaction,F(2, 55) � 5.34, p � .05. That is, in the context of complex testimony, highly paidexperts were found to be the least credible witnesses. It is interesting that partici-pants clearly saw the expert’s credibility to be a function of his trustworthiness, nothis expertise. Mock jurors did not discount the expertise of the highly paid witnesswho used complex testimony, but they did discount his credibility because theyfound his trustworthiness suspect.

The expert giving straightforward testimony was liked more than the expertgiving complex testimony, F(1, 55) � 4.39, p � .05, but this was qualified by aninteraction between complexity and pay, F(2, 55) � 4.84, p � .05. The highly paidexpert giving complex testimony was highly disliked by the mock jurors.


It is interesting to see which attributional measures were related to the degreeof persuasion induced by the expert. Likeability was correlated with persuasion,r � .43, p � .05, and with verdict confidence, r � .59, p � .05. Similarly, trustworthi-ness was significantly correlated with persuasion, r � .51, p � .05, and with verdictconfidence, r � .44, p � .05. Ratings of credibility were also significantly correlatedwith persuasion, r � .47, p � .05, and with verdict confidence, r � .39, p � .05.However, the perceived degree of expertise of the witness did not correlate withthese dependent measures, r � .11. Without implying mediation, the successfulexpert was nonetheless seen as credible and likeable. The more the expert wasliked and the more credible he was perceived to be, the more convincing he was.The expert who was neither liked nor credible was not persuasive.

Participants had been asked to write down their impressions of the reasonsthat the expert had for testifying in court. The answers were divided into threecategories by two independent raters. The raters assessed the degree to which eachreason was for ‘‘personal reasons related to pay’’, ‘‘personal reasons other thanpay,’’ and for ‘‘professional reasons.’’ A verbatim example of the personal-paycategory was, ‘‘This is the way this guy gets rich.’’ An example of the personalnonpay category was, ‘‘He enjoys being in court.’’ The professional category wasbased on the expert’s scholarship and included, ‘‘He knows a lot about biology.’’Interrater reliability for category judgments was high, r � .88. Pay was mentionedmore in the high-pay conditions (91%) than in the moderate-pay (87%) or low-pay(31%) conditions.

A variable of pay importance was created by examining the proportion withineach condition that participants rated pay as the number one reason for the expert’stestifying in court. The condition in which pay was viewed most frequently as theprimary reason for testifying is the high-complexity, high-pay condition (50%),compared to the other experimental conditions (combined percentage � 12.5%).Results of log linear analysis of proportions reveals a main effect for pay, �2(2,N � 60) � 4.66, p � .05, and a significant interaction between pay and complexity,�2(2, N � 60) � 4.91, p � .05.

Discussion

Our theoretical proposition is that mock jurors who process a message peripher-ally will be guided by simple, heuristic rules to gauge the persuasiveness of themessage. In the first experiment, mock jurors were exposed to moderately complextestimony from the plaintiff’s expert. They were persuaded to believe in the plain-tiff’s position in all conditions except when an expert with high credentials appearedfor a very large amount of money. Analyses of measures of attribution showed thatthe expert was not liked, not found believable, not persuasive, and found consider-ably annoying, even though the words of his testimony were identical to the expert’stestimony in the other conditions. We concluded that the expert’s high credentialsand high pay made him seem like a hired gun, someone who will testify for thehighest bidder. This hypothesis received further support in the second study, inwhich we varied the frequency of the expert’s testimony as well as the magnitudeof the fee he received. Mock jurors made the same negative attributions to the


highly paid, frequent testifier as they had made to the highly credentialed, highlypaid expert in the first study. Similar to the first study, plaintiff’s expert was markedlyunsuccessful at convincing the jurors to decide the case in favor of the plaintiff.

The third experiment added credence to the use of the ELM model of persua-sion as the way to explain the persuasion and verdicts in the summary trial. Theoreti-cally, peripheral processing and the concurrent reliance on simple processing cuesoccur when people are either unmotivated or unable to scrutinize and understandthe content of a message. By varying the complexity of the testimony, we systemati-cally allowed some mock jurors to understand the substance of the testimony andmade it considerably more difficult for jurors in the highly complex testimonycondition to comprehend the testimony. As predicted by the ELM, the use of payand frequency as cues to persuasion occurred only when the testimony was complex.In that case, mock jurors used the combination of high pay and highly frequentcourt appearances to decide not to believe the expert’s testimony. They did notlike the expert, they did not find him believable, and they did not believe hisconclusion that PCBs were the proximal cause of the plaintiff’s cancer. When payand frequency did not coexist in such a way as to cause the expert to appear to bea hired gun, then the complex testimony did lead to verdicts in favor of the plaintiff.However, when the testimony was presented in language that was easier to compre-hend, mock jurors were not affected by any of the peripheral cues. Their decisionsin the case were similar, regardless of the expert’s pay and frequency of testimony.This pattern is consistent with the use of systematic or central processing by partici-pants in the simple testimony condition and peripheral processing by participantsin the complex testimony condition. In the end, there was only one expert whowas markedly ineffectual in persuading a jury: The expert whose highly complextestimony led to the use of peripheral processing and who presented cues that ledthe mock jurors to believe he was akin to a hired gun. In three experiments, thiswas the witness who failed convince the mock jurors that they should render averdict in favor of the plaintiff.

Although the results of our studies are consistent with a dual-process, cognitivemodel of persuasion, several fascinating theoretical issues remain. Abelson (1988,1995) pointed out that people sometimes resist persuasion not solely because thearguments in a message failed to convince them, but also because they are affectivlymotivated to resist. In essence, people can be irritated by arguments that aredesigned to influence them. Zuwerink and Devine (1996) examined participants’reactions to persuasion attempts on issues they considered important and unimport-ant. They found that when people held preexisting opinions that were importantto them, their resistance to persuasion was both affective and cognitive. Resistancewas mediated by the participants’ feelings of irritation as well as by cognitiveelaboration. Is it possible that negative affect rather than systematic or heuristicprocessing could have accounted for the reactions of the mock jurors in the currentstudy? Indeed, it is hard to rule out the possibility that the mock jurors were irritatedor had other negative reactions to the expert they considered a hired gun. However,our situation differs from that of Zuwerink and Devine’s (1996) because our studydid not involve preexisting, heartfelt attitudes. In the current study, it is likely thatmost of the jurors had not given much prior thought to the role of PCBs in producing


cancer, which renders it less likely that our results were mediated by negative affect.Nonetheless, Zuwerink and Devine’s work raises an interesting issue that shouldbe examined more carefully in the presentation of persuasive testimony in thecourtroom.

In addition, the theoretical derivation of the current research is based on a dualprocessing view of persuasion that treats thoughtful systematic processing and short-cut heuristic processing as separate and independent phenomena. Our results areconsistent with this approach. However, recent work in persuasion has shown occa-sions in which systematic and heuristic processing cooccur. The cooccurrence canoccur independently, with the two processes adding together to produce persuasion(Maheswaran, Mackie, & Chaiken, 1992) or interactively (Chaiken & Maheswaran,1994). For example, Chaiken and Maheswaran (1994) showed that information aboutcredibility of a source can alter the systematic processing of a message. This view ofpersuasion implies that information about the hired gun might not only have servedas a heuristic cue in the complex condition, but may also have affected the mock jurors’interpretation of the meaning of the evidence that the scientist presented. We believethat the possible interaction of heuristic and systematic processing is important andthat it may well exist in courtroom situations. However, previous research has alsoshown that such interactivity is most likely when the persuasive message is ambiguous.In the current study, the testimony presented by the scientist may have been complex,but it was by no means ambiguous. Participants did not have to infer which side thetestimony favored or how firmly the scientist advocated his particular position. Wesuggest that it is important for future research to assess possible interdependent andindependent joint effects of the two modes of processing courtroom advocacy, espe-cially where that testimony may be particularly ambiguous.

There are several features that exist in the world of real courtrooms that differfrom the experimental analog. Some of these differences may affect the theoreticalconstructs used in the current research. As we have commented, mock jurors’reliance on cues such as pay and frequency of testifying occur when the jurors useperipheral as opposed to central processing. A variety of factors that exist in ongoingcourtroom procedures may enhance people’s motivation to process the testimonycentrally. One such factor may be cross-examination. When hearing the expertanswer questions designed to weaken his testimony, jurors may work harder toprocess the substance of the testimony. Another factor may be the finality of theverdict. In the current case, although every effort was made to make the decisionsimportant to the mock jurors, the reality of the real courtroom makes the finaldecision more significant. This, in turn, may affect jurors’ motivation. Finally, judicialinstructions in tort cases can encourage more central processing if jurors are askedto decide certain elements of law in addition to the final verdict.

Although the experimental analog to the courtroom can never match therealism of a court of law, attempts were made to keep mock jurors’ motivationhigh. Mock jurors were convinced that the proceeding in which they were involvedwas not simply an experiment, but rather an attempt to assess the value of summarytrials that are used to help plaintiffs and defense come to more realistic decisions.Second, the dependent measures encouraged elaborated thinking by asking partici-pants not only for the verdict, but also about such issues as proximal cause for


plaintiff’s cancer. Our observation of mock jurors’ interest and motivation was thatit was considerably higher than what researchers normally experience with collegestudents in experimental research. Recognizing that these observations are not harddata, we can nonetheless say that our mock jurors, who were residents of thesurrounding community, were extremely motivated to do a good and thorough job.Debriefing was always lengthy, not because the researcher had that much to say,but because the participants had so much to say about the case that it was oftendifficult to get them to leave the laboratory.

Nonetheless, we must caution that our results must be interpreted within thelimits of the reactions of mock jurors rather than real jurors. We note, too, thatthe use of peripheral processing was created in the present studies not because thejurors were insufficiently motivated, but rather, especially in experiment 3, becausethe complexity of the testimony compromised their ability to process. We suspectthat it is not uncommon that scientific expert testimony is complex and difficult tofollow. As Petty and Cacioppo (1986) have pointed out, such complexity shouldpush people toward peripheral rather than central processing, and it is in thosecircumstances that experts may be perceived as hired guns, diminishing their effec-tiveness as witnesses.

REFERENCES

Abelson, R. P. (1988). Conviction. American Psychologist, 43, 267–275.Abelson, R. P. (1995). Attitude extremity. In R. E. Petty and J. A. Krosnick (Eds.), Attitude strength:

Antecedents and consequences (pp. 25–41). Hillsdale, NJ: Erlbaum.Cecil, J. S., Hans, V. P., & Wiggins, E. C. (1991). Citizen comprehension of difficult issues: Lessons

learned from civil jury trials. American University Law Review, 40, 727–774.Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus

message cues in persuasion. Journal of Personality and Social Psychology, 39, 752–766.Chaiken, S., & Maheswaran, D. (1994). Heuristic processing can bias systematic processing: Effects of

source credibility, argument ambiguity, and task importance on attitude judgment. Journal ofPersonality and Social Psychology, 66, 460–473.

Cooper, J., Bennett, E. A., & Sukel, H. L. (1996). Complex scientific testimony: How do jurors makedecisions? Law and Human Behavior, 20, 379–394.

Fienberg, S. E. (1980). The analysis of cross-classified categorical data. Cambridge, MA: MIT Press.Gross, S. R., & Syverud, K. D. (1991). Getting to no: A study of settlement negotiations and the selection

of cases for trail. Michigan Law Review, 90, 319.Hans, V. P., & Ivkovich, K. K. (1994). How jurors evaluate experts. Trial Lawyers Forum, 28, 5–9.Mackie, D. M., & Worth, L. T. (1989). Cognitive deficits and mediation of positive attitude change.

Journal of Personality and Social Psychology, 57, 27–40.Maheswaran, D., Mackie, D. M., & Chaken, S. (1992). Brand name as a heuristic cue: The effects

of task importance and expectancy confirmation on consumer judgments. Journal of ConsumerPsychology, 1, 317–336.

Petty, R. E., & Caciopppo, J. T. (1986). Communication and persuasion: Central and peripheral routesto attitude change. New York: Springer-Verlag.

Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal involvement as a determinant of argument-based persuasion. Journal of Personality and Social Psychology, 41, 847–855.

Ratneshwar, S., & Chaiken, S. (1991). Comprehension’s role in persuasion: The case of its moderatingeffects on the persuasive impact of source cues. Journal of Consumer Research, 18, 52–62.

Zuwerink, J.R. & Devine, P.G. (1996). Attitude importance and resistance to persuasion: It’s not justthe thought that counts. Journal of Personality and Social Psychology, 70, 931–944.

The “Hired Gun” Effect: Assessing the Effect of Pay, Frequency of Testifying, and Credentials on the Perception of Expert Testimony

Documents