The Effect of Pretrial Publicity on Trial Outcomes
Department of Speech Communication
California State University, Fullerton
Fullerton, CA 92834
William E. Loges
Department of Communication Studies
Waco, TX 76798
(254) 710-4466 (o)
Jon Bruschke (Ph.D., U of Utah, 1994) is a Lecturer in the Department of Speech Communication at California State University, Fullerton, and Executive Director of the Daniel Webster Foundation, which teaches debate skills to at-risk high school students in order to prepare them for college. His research in legal communication has appeared in Communications & The Law and Argumentation & Advocacy.
William E. Loges (Ph.D., U of Southern California, 1992) is Assistant Professor of Telecommunications at Baylor University. His previous research into media system dependency appeared in Communication Research and Journalism Quarterly, and he has published on the subject of human values in the Journal of Social Issues. He is presently studying the interpretation of presidential candidates references to values in campaigns.
The authors gratefully acknowledge the assistance of Kristen Walker, Jeff English, Nick van Kluyve, Christine Ford, and Christina Flora in the collection of these data, and the advice of Dennis Gouran on an earlier draft of this manuscript. This project was funded by a University Research Committee grant from Baylor University.
The issue of pretrial publicity is of tremendous social and theoretical importance and its prominence seems only to be growing since the O.J. Simpson trials. Extant research generally suggests that pretrial publicity influences trials. Past research has relied heavily on experimental methods and several commentators have questioned the ecological validity of the studies, and hence the value of the findings. The present study was the first to examine actual trials, and focused on all federal murder trials over a three year period. The results indicated that: (a) greater probability of conviction is associated with low rather than high levels of publicity, (b) defendants fared better under moderate rather than no publicity levels, and (c) for those defendants who are convicted, any degree of pretrial publicity is associated with longer sentences. In general, the results obtained from studying actual trials differ in important ways from the results obtained in laboratory research, which suggests that further study of actual trials with more refined measures could lead to more appropriate and relevant insights.
Key words: Pretrial publicity; media effects; legal communication
THE EFFECT OF PRETRIAL PUBLICITY ON TRIAL OUTCOMES
The issue of pretrial publicity is a maze of overlapping attentions and interwoven interests. Lawyers decry pretrial publicity while simultaneously raising their own career stock and hourly fee by accumulating more if it. The media both perpetrate and comment on the frenzy -- newspapers and television stations generate the publicity in the first place and then actively comment on the likely effect that the coverage will have on the trial. Litigants endure more and more egregious privacy intrusions and indignities and then cash royalty checks from what are increasingly lucrative book contracts that "tell their story." Many litigants have, ironically enough, taken the route of appearing on televised talk shows to lament media interest in their legal cases. The landscape is bizarre indeed. And while pretrial publicity has always been an issue for courts, scholars, and the polity to grapple with, the advent of the OJ Simpson spectacle seems to have made the issue even more poignant. Whatever may have been true of pretrial publicity before the Simpson trial, since Simpson press coverage has become part of the legal vernacular surrounding virtually any high profile case. Against this backdrop it is a good time to take stock of what scholars can contribute to the discussion.
The current state of thinking appears to be that pretrial publicity does influence the outcome of legal trials. A growing number of studies, utilizing a variety of methods ranging from jury simulations to telephone surveys, have concluded that pretrial publicity does in fact bias the trial outcome against the defendant (Constantini & King, 1980-81; Dexter, Cutler, & Moran, 1992; Greene & Wade, 1988; Hoiberg & Stires, 1973; Hvistendahl, 1979; Kramer, Kerr, & Carroll, 1990; Kerr, Kramer, Carroll & Alfini, 1991; Moran & Cutler, 1991; Ogloff & Vidmar, 1994; Otto, Penrod, & Dexter, 1994; Padawer-Singer & Barton, 1975; Rollings & Blascovich, 1977; Simon & Eimerman, 1971; Sue, Smith, & Pedroza, 1975). The publicity need not be direct to generate this effect; news about crime in general (Greene & Loftus, 1984; Greene & Wade, 1988) has been found to have a biasing effect. Neither does the news coverage have to be extensive, as even moderate levels of exposure have been found to influence juror opinions (Moran & Cutler, 1991). This suggests that anti-defendant feelings "bleed over" across cases and issues, a result that has been discovered in multiple-offense cases (Tanford & Penrod, 1982). Fulero (1987), following a review of literature, has summarized that: "[T]he body of research taken as a whole demonstrates an adverse effect of PTP [pre-trail publicity] on jurors" (p. 260).
Potential remedies do not seem to attenuate this effect. Jury instructions are ignored (Kramer et al., 1990; Hans & Doob, 1976; Sue, Smith, & Gilbert, 1974). Jury selection has produced more mixed reviews, with some scholars applauding the process for its ability to remove bias (Padawer-Singer & Barton, 1975), some scholars reporting equivocal results or admitting confusion (Carroll et al., 1986; Vidmar & Melnitzer, 1984; Zeisel & Diamond, 1978), but with the strongest opinions decrying its effectiveness (Dexter et al., 1992; Kerr et al., 1991; Padawer-Singer, Singer, & Singer, 1974). Other remedies, such as a continuance, have similarly failed to counteract publicity, biasing or actually exacerbated its influence (Kramer et al., 1990). The bulk of research to date appears to point to the conclusion that pretrial publicity biases trials against defendants and the remedies available in our current legal system are inadequate to counteract it.
Despite the seeming unanimity of findings to date, the studies that have been conducted have studied pretrial publicity in only very controlled conditions. As a result, we know a great deal about how pretrial publicity functions in very straightforward conditions of exposure, but we know very little about its influence in more equivocal circumstances. These less controlled circumstances might very well be more prevalent in actual courtrooms; at least one commentator has identified present lab studies as lacking realism (Carroll et al., 1986). This lack of realism has created a gap in our knowledge of how pretrial publicity affects trial outcomes and limits the ability of our findings to inform actual legal practitioners. There is very clearly a need to expand our knowledge beyond the controlled circumstances of the laboratory and extend them to the more muddled circumstances of the courtroom.
It is worth characterizing our present knowledge so that we can more fully understand what it speaks to and where its limits are. First, virtually every study to date, when operationalizing its "high publicity" condition, has exposed every jury member to a relatively large amount of publicity or grouped together all respondents to be exposed to the publicity. In general, the findings of these studies tend to suggest that jurors will discuss events that they heard in news reports during deliberation (e.g., Kerr et al., 1991) even if the information did not come up in the trial. The instance where every juror is exposed to the same publicity, however, may be fairly rare in actual trials. If only half the jury has been exposed to facts that are inadmissible at trial, non-exposed jurors may successfully point out which evidence was not presented at trial if suppressed facts come up during deliberation. At any rate, virtually nothing is known about how jury deliberations will be conducted when jurors are exposed to different pretrial information. Most simply put, publicity effects are likely to be strongest where every jury member is exposed to identical publicity and are likely to be more modest where some jurors have experienced publicity but others have not. The research to date has studied only the former condition.
Second, while it may be expected that pretrial publicity is generally pro-prosecution (see Imrich, Mullin, & Linz, 1995), the operationalizations of publicity have been especially damning to the defendant. Typical "high publicity" conditions include information about a prior criminal record of the defendant (e.g., Dexter et al., 1992), retracted confessions (e.g., Dexter et al., 1992; Kerr et al., 1991), or the implication that the defendant, in a separate crime than the one at trial, ran over a small child with a car (e.g., Kerr et al., 1991). In even the most sensationalized trials of the past decade the public has not been exposed to information so conclusively anti-defendant. Imagine the difference in public reaction, for example, if it were widely publicized that a man matching OJ Simpson's description in a white Bronco had ruthlessly run over a 5-year old orphan on his famous freeway chase. While it will not be the purpose of this study to compare the content of actual press coverage to the coverage as operationalized in the studies to date, it does seem reasonable to question whether it can be reliably concluded that pretrial publicity has a strong influence on trial outcomes or whether such findings are only present under very extreme operational definitions of "coverage."
Third, no study to date has included the effect of evidence as a possible counter-weight to juror bias. In fact, those studies that have included some estimate of the prosecution's case strength have gone to some lengths to ensure that the prosecution's case is weak, or that the evidence is inconclusive without the inadmissible facts (e.g., Padawer-Singer et al., 1974). Such an exclusion can have a profound influence on results, as Visher (1987), in a study of actual trials rather than simulations, found that evidence is three times more important in determining jury verdicts than any juror or defendant characteristic. Otto et al. (1994), in a much less realistic study, produced nearly identical results. While they did find an effect for pretrial publicity, trial factors were roughly three times more determinative of the final outcomes. Tanford & Penrod (1982) have produced similar support for the primacy of case strength in determining case outcomes. Thus, studies to date have examined publicity only in the cases where the evidence is most inconclusive and thus publicity is likely to have the greatest effect. The significance of publicity is likely to fade in cases where the evidence is more compelling, as may be more typical. At a minimum, the present state is one where only cases with inconclusive evidence have been considered. This situation again calls into question the generalizability of our current conclusions. While we know that pretrial publicity has an influence in situations where pretrial publicity is especially damning and prosecution evidence is weak, it is less clear that a similar influence is present when the prosecution has a strong case or the defense can present strong evidence in its own behalf.
All this is not to say that the factors identified above are the truly determinant variables that dictate court outcomes. They are simply offered as concrete examples of the more general concern that the studies that presently exist may not greatly inform interested parties about the influence of pretrial publicity on actual trials. Simply put, we know a great deal about how mock jurors will react to laboratory simulations of court trials, but what research has been done on actual trials suggests that these operations will be profoundly different in actual courtroom settings. The need to improve over current measurements has been recognized by a growing number of scholars. To improve our current understandings, as Carroll et al. (1986) summarize, "An excellent starting point would be research employing realistic cases, publicity, and involved parties" (p. 196). As Carroll et al. point out earlier in their article, those studies that examine real cases have not measured case outcomes, and those studies that have measured case outcomes have not studied real cases. More realism may result in more modest estimates of any pretrial publicity effect -- the only study to date that has measured trial format found that as the trial becomes more realistic conviction rates drop (Kramer & Kerr, 1989).
The present study proposes to analyze actual court trials and compare murder cases with large amounts of pretrial publicity against those that have not been covered widely. We do not propose to content-analyze the news coverage each case receives, but instead to measure the gross amount of coverage. Because research previously conducted on actual trial coverage seems to have demonstrated that the vast majority of pretrial news coverage is harmful to the defendant (Imrich, Mullin & Linz, 1995; Tankard, Middleton, & Rimmer, 1979), and that in fact much pretrial coverage actually contains information that violates ABA rules on pretrial information release designed to protect defendants, it can be assumed that more coverage means more anti-defendant coverage. Conventional news practices can explain this, since reporters tend to develop enduring relations of mutual dependency with law enforcement officials and find them to be steady sources of crime news (Stephens, 1993; Bennett, 1996). Particular defense attorneys are not as constantly helpful or necessary in the day-to-day practice of gathering news. Robert Shapiro, a member of OJ Simpsons "Dream Team," cautioned an audience of defense attorneys to develop good relations with reporters in order to guard their clients interests (Shapiro, 1994). He noted that this is necessary because reporters will reflexively publish information they obtain from police and prosecutors because these people have credibility with the public and are known to the reporters.
It is certainly possible that some news coverage of a murder would favor a defendant, but even if the suspects claims of innocence were given full and loud coverage it is far from clear that such attention constitutes "positive coverage." Denying a felony is hardly good for ones reputation. As a first step in examining the overall effects of publicity on real trial outcomes, we feel that it is reasonable to use a simple measure of overall publicity without attempting to analyze the content too closely. Future research may benefit from a more careful distinction between prejudicial and favorable coverage, but at present the separation seems theoretically tenuous and the occurrence of positive coverage seems rare enough to conclude that gross measures of trial coverage represent the best research choice at present.
Further, we believe that research conducted at this point will be most fruitful if it is exploratory. While certainly there are theoretical assumptions that will guide our research choices (most of the tests we intend to conduct will presume that more coverage works contrary to defendant interests), and we are extremely sensitive to the dangers involved in "data snooping" to find significance at all costs. We intend to make informed choices about patterns that are suggested in the data rather than test firm hypotheses. Because we expect that media effects will be much different in actual trials than in courtroom trials we are not suggesting firm hypotheses and are instead content with a preliminary analysis that should suggest more concrete hypotheses for more controlled testing in the future. The data analyses reported below were guided by two main research questions:
RQ1: What is the relationship between the amount of pre-trial publicity and the probability of conviction in federal first degree murder trials?
RQ2: What is the relationship between the amount of pre-trial publicity and the length of sentences handed down in federal first degree murder trials?
All federal first degree murder cases that were terminated at trial from 1993 to 1995 were considered for the sample. The Federal Judicial Center (FJC) compiles information on many aspects of each federal case, including the trial disposition and the sentence length, if any, of the criminal defendants. This information is made publicly available through the Inter-University Consortium for Political and Social Research (ICPSR) at Ann Arbor, Michigan; all first degree murder cases were identified in that data set (Federal Judicial Center, 1997). However, while the FJC includes the docket numbers for each case they do not included defendant names. In order to obtain the names of defendants, the docket numbers were cross-referenced against the Public Access to Court Electronic Records (PACER) databases maintained by the federal courts. The PACER system is a network of electronic databases of the federal court system; each district maintains its own database. While the vast majority of federal districts are incorporated in the PACER system, some are not. In addition, different districts started maintaining their databases at different times and thus while some districts include case filings as early as 1990 some districts have very sketchy records before 1995.
In all, between 1993 and 1995 there were 165 first degree murder defendants that faced trial and obtained a verdict. If multiple defendants were tried together each defendant was treated as a separate unit of analysis. Defendants who faced trial in front of either a jury or a judge were included in the sampling frame (7 cases were tried in front of judges; the rest were tried by juries). Fifteen defendants were excluded because the ICPSR docket number was not found in the PACER system (probably because the case was initiated before the local court signed on to the PACER system), 2 defendants were excluded because a defendant involved in a multiple-defendant case was identified in the ICPSR data set did not exist in the PACER docket report, 10 defendants were excluded because they were tried in districts that did not yet grant PACER access, 1 defendant was identified only as a "juvenile male," and 3 were excluded because repeated attempts to log into the local PACER system were unsuccessful. In all, 134 of 165 (81.2%) defendant names were obtained.
The names of all murder defendants were entered into the Lexis/Nexis database, a full-text compilation of thousands of newspapers around the world that is updated daily. The database is searchable with individual words and will locate the occurrence of any search term in any article in the database; the result is that typing in a given name should produce a listing of all articles in which that name appeared. The full text of those articles can then be analyzed. Although limiting a search to the Lexis/Nexis database does present limitations television coverage is not measured and not all newspapers are included in the database Lexis/Nexis does represent a state-of-the-art system and is the most complete national collection of newspaper coverage available.
A number of research decisions had to be made to insure the uniformity of database searches and article counting. In relation to what counted as "pretrial publicity," we decided to err on the side of inclusion, based on the broad definitions of pretrial publicity utilized in prior research and the research findings cited above that even tangential material can damage a defendants chances. Thus, any article that contained the defendants name in relation to the crime was included, as was any mention of the defendants name in relation to a prior crime. Further, any coverage of the crime that identified the victims name but not the defendants name was included. If the crime was gang-related and it was feasible to conduct a search by the name of the gang, all articles implicating the gang in criminal behavior were counted. Finally, any coverage on co-defendants relating to the crime was counted. In sum, an attempt was made to identify all stories related to the defendant, the defendants criminal history, the criminal conduct of the gang or group that committed the crime, and any story about the victim after the crime occurred. Any article preceding the verdict was counted, including articles produced during the trial.
The only articles that were excluded were obituaries. One search attempting to identify coverage of a murder defendant in Virginia produced a story on a robbery defendant in California who was beaten by police; that article was excluded. Some papers produced multiple versions of the same edition; coverage of the same defendant on the same day of the same paper with the same headline was counted as only one article, even if they appeared differently in the database as a "metro" and a "downtown" edition.
In relation to search term selection, one difficulty was that the defendants name may have appeared differently in press coverage than it did in court records. The database was searched with the full first, last, and middle name of a defendant (when available), followed by a separate search with the first and last name and middle initial, followed by a separate search with the first and last name only. Nicknames, pseudonyms, and shorter versions of first names (i.e., "Jon" from "Jonathan") were not searched. When victim names could be identified from the coverage generated by the search of defendant name, the victim name was searched in the same fashion (first with a full middle name, then with a middle initial, and finally with no middle initial). Where it was easily identifiable and relevant, the name of the gang the defendant belonged to was searched.
Where a search turned up an impossibly large number of articles (a search of criminal defendant "James Smith," for example, turned up more than 1,000 articles) the search term "murder" was added to the search. Thus, only the articles with the defendants name and the word murder were examined. All articles were screened to insure that they referred to the criminal defendant in question.
Special Limitations of the Database
Initially, not all papers are included in the Lexis/Nexis database; for example, no Utah paper is included. Thus, a criminal defendant tried in Utah could conceivably be identified as having no press coverage at all when in fact the Salt Lake Tribune may have covered the defendant every day before the trial on its front page. This is troubling for obvious reasons. However, we believe the information remains useful for three reasons. First, any attempt to "fill in the holes" left by Lexis/Nexis would have produced uneven success; some areas would have easily accessible materials and others would not. Second, even with the excluded papers Lexis/Nexis remains the most comprehensive database at present. Third, virtually all major media markets are represented in the database to some extent; areas with no representation in Lexis/Nexis tend to be less populated areas. Thus, counting only the coverage that is reproduced in Lexis/Nexis may still serve as a useful, if rough, index of the total media coverage of the case. Simply put, a trial held in New York City is likely to produce more media coverage than one held in Ogden, Utah simply due to the larger number of people and newspapers located in New York. Ultimately, it seemed more useful to have all defendants researched in an identical manner than it did to try to expand the search beyond the Lexis/Nexis database.
From a practical standpoint, and based mostly on our feel of what the data were representing, it was our impression that huge oversights were unlikely. We simply are convinced that very few of the "no publicity" defendants actually had several articles written about them that were missed; one intriguing feature of publicity is that the more of it a given defendant receives the more easily the newspapers can be located. If the defendant is famous and the crime is well known, a large number of articles are readily identifiable. When the crime and the defendant is more obscure articles are more difficult to locate. These limitations may have resulted in the under-counting of coverage for some defendants, which is most troubling for defendants in the low-to-moderate coverage categories, but the "no publicity" and "high publicity" categories are very likely to be highly accurate. Simply put, missing an article or two doesnt matter at all if the defendant has already been located as a "high publicity" defendant, and it seems inconceivable that the defendants in the "no publicity" category actually had a large number of articles written about them, all of which were missed with the current search methods. The hypothetical defendant in Salt Lake, if sufficiently notorious (as were Ted Bundy and Gary Gilmore), would likely have some coverage turn up somewhere that was accessed by the data base.
Nonetheless, the final numbers of articles counted do appear to be only a rough index of the amount of coverage each trial received; more refinement of these methods might produce slightly more precise categories of publicity and we encourage such efforts in future research. At present we believe that the measures of publicity for each defendant are, at least, in their correct ordinal positions.
The frequency of the number of articles for the 134 defendants is reported in Table 1. A normal median split would have classified the defendants into those receiving no publicity and those receiving any publicity, and we suspected that there was a vast difference between a case that received coverage in 79 different articles and one that received coverage in only 1. We therefore examined the frequencies in Table 1 and looked for what appeared to be natural breaks; the articles seemed to cluster around those cases that had 5 or fewer articles, 6 to 10 articles, and more than 11 articles. For descriptive purposes, these categories were labeled "no publicity," "low publicity," "moderate publicity," and "high publicity" conditions.
The different conviction rates are reported in Table 2. The overall pattern does not produce significance (X2 = 4.45, df = 3, p = .207), and as the Table reveals the high and no publicity conditions produce virtually identical conviction rates while the low and moderate rates deviate from the overall conviction rate of 79.9% most sharply. Of those two, low publicity cases produced markedly higher conviction rates (over 90%) while defendants in the moderate publicity condition were the most successful, with conviction rates at a relatively low level of 68%. Obviously, the non-significant finding is a product of how closely the high and no publicity conditions match the grand conviction rate; to test whether the low and moderate publicity conditions differed from one another a second test was conducted including only the low and moderate publicity conditions. The results were significant (X2=6.91, df =1, p < .01), confirming that conviction rates varied significantly between low and moderate publicity conditions.
The length of sentence (in months) for convicted defendants is reported in histogram form in Table 3 (capital sentences, n = 5, were excluded). As can be observed in the Table, the data do not conform to a normal curve but take on a rather bimodal nature. As a result, rather than treating the data as continuous interval measures and subjecting them to an ANOVA procedure (which would produce very high standard deviations), the sentence length was dichotomized into those defendants receiving sentences of fewer than 500 months and those defendants receiving sentences of more than 500 months.
A cross-tabulation of the publicity conditions with sentence length is reported in Table 4. In general, the more publicity a defendant received the more likely s/he was to fall into the longer sentence category. The overall pattern of the data approached but did not obtain significance for the standard Chi-Square measure (X2 = 6.78, df = 3, p =. 08) but did produce significance for the linear-by-linear association measure (X2 = 5.29, df = 3, p< .05). An examination of the Table makes it clear that the biggest jump is between the no publicity condition and any of the three publicity conditions; long sentences are given to about 29% of defendants receiving no publicity and range between roughly 52 and 56% for all other conditions. To contrast the no publicity condition with any publicity condition, all publicity categories were collapsed and the data subjected to another analysis. Not surprisingly, the results were significant for both the traditional Chi-Square (X2 = 6.71, df = 1, p < .01) and the linear-by-linear association measure (X2 = 6.65, df = 1, p < .01). Long sentence rates in the no publicity condition remained 28.6% and were 53.6% in the collapsed publicity category.
At least three substantive findings are noteworthy. First, conviction rates between high and no publicity conditions were virtually identical. The overall conviction rate for all murder defendants was about 80%, for defendants without any publicity it was about 79%, and for defendants with high publicity it was about 82%. This finding is in sharp contrast to the results of the laboratory studies. Given that the research to date suggests that pretrial publicity is highly prejudicial, voir dire examinations are an ineffective counterbalance, and jurors will ignore instructions to disregard prior information, one would expect that the conviction rates of actual trials would be at least somewhat higher for defendants subjected to high degrees of publicity as opposed to defendants with no publicity whatsoever. This is not the case. It is fascinating to speculate as to why this may be the case either the effect of pretrial publicity is lower than may have been originally thought, or the protections offered by the legal system may function more effectively than has thus far been supposed but at any rate it does not appear that highly publicized defendants are treated much differently in terms of ultimate conviction rates than are defendants that receive no publicity at all.
A second and equally provocative pattern is that defendants in low as opposed to moderate publicity conditions have such sharply higher conviction rates. Defendants who are the most likely to be convicted (92%) have had some publicity but fewer than five articles written about their cases; defendants who are most likely to be acquitted (68% conviction rate) have had more than six but fewer than ten articles written about their cases. This suggests a fascinating new possibility pretrial publicity may indeed have a prejudicial influence on trial outcomes, but that influence occurs only in relation to low as opposed to moderate conditions of publicity and does not exert an influence under extreme conditions of publicity. Research to date is virtually unanimous in contrasting high and no publicity conditions. Perhaps we have been looking for differences in the wrong places.
Third, pretrial publicity does have an effect at the sentencing phase, but at that point the amount of pretrial publicity seems unimportant. While defendants with no pretrial publicity at all receive much lower sentences than defendants receiving publicity, there is virtually no difference in the sentence length of defendants who have been covered in one as opposed to 149 articles.
These empirical findings evoke a series of weighty theoretical questions. Many of the issues are not resolved by the data presented here, but the findings serve as a starting point for the following theoretical discussions and suggest areas for further research. First, we can begin to translate our thinking from either/or questions to inquiries of degree; rather than asking whether or not pretrial publicity has an effect, and devising studies to prove that it does or does not, we can ask at what degree pretrial publicity begins to influence trials. The results here suggest that the answer may not be a simple linear association; when asking at what level pretrial publicity become prejudicial it is important to allow the possibility that there is not some magic point at which the publicity becomes too much, but rather be sensitive to the potentiality that low levels of publicity might be the most prejudicial. Our research here has in a largely intuitive way divided publicity into four divisions based on simple article counts; future research should seek to refine more carefully what the important classifications are and at what level publicity becomes prejudicial.
Second, we can start to ponder why it is that publicity has its most damaging effect at low levels. One possibility is that high levels of publicity stimulate a more aggressive defense. The defense burden is simply to raise a reasonable doubt, and while this is no simple task one would expect that as public attention to the case raises the efforts of both the defense and the prosecution, and thus they tend to cancel each other out at high publicity levels. With both lawyers trying equally hard the litigant with the task of creating doubt ought to prevail over the litigant with the task of trying to eliminate doubt. At low publicity levels in a context where criminal defendants are convicted the vast majority of the time, however, the case may seem an easy prosecution win; an attorney in a case that has been covered in a single article outlining the heinousness of the crime might chalk up the case as a defeat. This suggests a model that includes another variable virtually unexamined to date -- publicity might influence attorney effort which in turn influences trial outcomes.
The cause of defense success rates at moderate levels of publicity is more difficult to construe, but it could be that something about the content of the coverage is more favorable to the defendant. It could be, for example, that cases that receive moderate coverage are cases where the media is reporting that the state may have the wrong suspect. This possibility contrasts prior research, identified at the outset, which has found that coverage is almost universally anti-defendant. At any rate, these are mere possibilities but fertile ground for further research.
Third, the research here suggests that legal reforms might be more usefully aimed at the sentencing phase of the trial rather than the guilt or innocence phase. While pretrial effects may be present at the guilt phase of the trial, they are subtle, elusive, and intricate at best. At the sentencing phase, however, they appear to be strong and relatively easy to detect. This is not to suggest that guilt-phase protections are unimportant or no longer necessary, but simply to point out that the battle for legal fairness does not end before the sentence is rendered and in fact may become more important.
Fourth, future research could expand the measures included here to provide a more complete picture of the legal process. While we have studied only newspaper coverage, research that studied the influence of other media could add an additional element of realism. In fact, extant research gives cause to believe that publicity effects may be more strong when television and print media cover a case (Ogloff & Vidmar, 1994). Virtually all studies to date have examined murder cases only; it would be interesting to know if publicity effects extend to other charges as well. The dependent measure could be expanded beyond dichotomous measures of guilt and innocence as well. Although both OJ Simpson and Don Edward Cash (a defendant identified in our research whose first public mention is this one) were both acquitted it is difficult to maintain that they were therefore treated equally. Some measure of defendant treatment other than guilt or innocence and sentence length may be warranted. One very useful direction may be to consider plea bargains. Of the 315 murder defendants in 1995, only 84 went to trial. In order to understand how publicity influences all criminal defendants it is necessary to examine how publicity influences those defendants who choose not to come to trial (a much larger group) than only those defendants who do go to trial. At present, it seems to safe to say that scholarly theorizing thus far (including this work) has included only considerations of guilt-phase trial outcomes, probably because they are so easily treatable with our extant research designs. The actual legal process contains many more intricacies that have not yet been widely explored but may be more important and are certainly more frequent.
A related issue is how large the size of the effect of publicity might be. Murder defendants in the federal system start with an 80% chance of conviction (the overall conviction rate for all crimes is a very comparable 77%). To say that publicity prejudices the system against certain defendants means that their conviction rates go up to 90%, and to say that some defendants are better situated still means that they are convicted 7 out of 10 times. Pretrial publicity is an issue that has captured media attention, but it is important to keep the issue in context. The total context of the criminal justice system is one where many crimes are never solved, most cases are not federal, most cases are not publicized, most cases do not come to trial, and most cases that do come to trial result in conviction. Efforts aimed at dispensing justice in a more fair manner might reasonably focus on issues other than pretrial publicity, such as how to prevent more murders, how to track down murder perpetrators more regularly, or what happens in the completely media-free plea bargaining process. At the very least, the very high actual murder conviction rate stands in stark contrast to the very balanced conviction rates obtained in many laboratory studies. Even if there is a pretrial publicity effect on trials, this happens in the context of very high conviction rates across the board. The magnitude as well as the direct of the effect is worth keeping in mind.
A final question this research raises is the circular nature of explanations for publicity effects. As we were trying to reason through and make sense of the pattern of results presented here, there was a constant tendency to try to imagine what sort of a case would stimulate coverage in only a single article. This line of thinking, of course, suggests that there is one variable that has something to do with the case content that drives both the media coverage and the conviction rate. Theoretically, this suggests that there is one class of variables that measure the case content, a separate set of variables that measure media coverage, and the pressing question is whether or not those distinct classes of variables make a unique contribution to case outcomes. This presents an urgent need to be able to measure in more detail the content of the case. It is worth noting that the legal system treats all cases as similar; a murder one charge is a murder one charge, and the only difference in those cases examined in this paper, so far as legal classifications are concerned, is the amount of media coverage.
Of course, in laboratory studies this is not an issue; the reasons behind press coverage are unimportant and coverage levels are artificially manipulated. In actual trials, however, media editors base their decisions on whether to cover a case or not on some criteria that is at this point largely unexamined by scholars. Simply put, laboratory researchers can concoct a case that would not normally receive press coverage and artificially generate coverage around it, or conversely generate a case that would generate vast coverage and then expose simulated jurors to none of it. These situations are of clear academic interest but may not exist during actual trials. A better approach would be to try to identify those elements of trials that result in conviction and then see if they can also explain levels of press coverage. If they can, the question of media effects on trial outcomes could evaporate. If they cannot, media influences may have a separate effect. At any rate, the route that may best inform us about the conduct of actual trials is one that begins with measuring case elements rather than creating cases in our laboratories. Such is the daunting task for future research.
One final limitation that is especially troublesome is that while this study has measured the amount of press coverage that surrounds a given case, there is no guarantee that the particular jurors that hear the case have been exposed to it. In some instances, such as Branch Davidian Clive Doyles case covered in 179 articles, it is difficult to imagine that no juror heard anything about the case. In other cases, where only a single article was written about the trial, it is possible to imagine that the jury was exposed to case information for the first time at trial. Future research should certainly attempt to address whether media coverage results in jury exposure before the trial, or compare trials in which jury members report media exposure before the trial. Still, the results contained here make the issue poignant: If jurors in the no publicity and low publicity condition are both equally unlikely to be exposed to any information about the case before the trial, why do the conviction rates differ? As noted above, there are at least two ways to explain the different conviction rates that do not rely on jury exposure. One concerns the nature of the case (some sorts of cases may contain factors that lend themselves both to conviction and publicity) and the other concerns the possible influence of publicity on the attorney (defense attorneys may exert less effort in low as opposed to no publicity conditions). At this point, either explanation serves simply as intriguing speculation, but this is exactly the sort of speculation that research of this sort is able to generate that further laboratory investigation likely would not.
In conclusion, this article began by arguing that the examination of actual murder trials could produce insights that are not found in laboratory studies. The substantive findings of this research project for example, that low rather than high levels of publicity seem most damaging, or that the influence of publicity seems greatest at the sentencing phase suggest that our suspicions were correct. The results here seem to make a unique and important contribution to the ongoing debate about pretrial publicity. Still, they suggest that the search has just begun and more intricate and subtle explanations await us.
Bennett, W.L. (1996). News: The Politics of Illusion, 3rd edition. White Plains, NY: Longman.
Carroll, J. S., Kerr, N. L., Alfini, J. J., Weaver, F. M., MacCoun, R. J., & Feldman, V. (1986). Free press and fair trial: The role of behavioral research. Law and Human Behavior, 10, 187-201.
Constantini, E., & King, J. (1980-81). The partial juror: correlates and causes of prejudgment. Law and Society Review, 15, 9-40.
Dexter, H. R., Cutler, B. L., & Moran, G. (1992). A test of voir dire as a remedy for the prejudicial effects of pretrial publicity. Journal of Applied Social Psychology, 22, 819-832.
Federal Judicial Center (1997). Federal Court Cases: Integrated Database, 1970-1994 [Parts 1-14, 29-33, 58-60, 67-69: Criminal Data, 1970-1991] [Computer file]. 3rd ICPSR version. Washington, DC: Federal Judicial Center [producer], 1992. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor].
Fulero, S. M. (1987). The role of behavioral research in the free press/fair trial controversy: Another view. Law and Human Behavior, 11, 259-264.
Greene, E. L., & Loftus, E. F. (1984). What's new in the news? The influence of well-publicized news events on psychological research and courtroom trials. Basic and Applied Social Psychology, 5, 211-221.
Greene, E. L., & Wade, R. (1988). Of private talk and public print: General pretrial publicity and juror decision-making. Applied Cognitive Psychology, 1, 1-13.
Hans, V. P., & Doob, A. N. (1976). Section 12 of the Canada Evidence Act and the deliberations of simulated juries. Criminal Law Quarterly, 18, 235-253.
Hoiberg, B. C., & Stires, L. K. (1973). The effect of several types of pretrial publicity on the guilt attributions of simulated jurors. Journal of Applied Social Psychology, 3, 267-275.
Horowitz, I., & Willing, T. (1984). The psychology of law: Integrations and applications. Boston: Little, Brown, & Co.
Hough, G. A. (1970). . In C. R. Bush (Ed.), Free press and fair trial: Some dimensions of the problem. Atlanta: University of Georgia Press.
Hvistendahl, J. K. (1979). The effect of placement of biasing information. Journalism Quarterly, 56, 863-865.
Imrich, D. I., Mullin, C. & Linz, D. (1995). Measuring the extent of prejudicial pretrial publicity in major American newspapers: A content analysis. Journal of Communication, 45, 94-117.
Kaplan, M. F., & Miller, L. E. (1978). Reducing the effects of juror bias. Journal of Personality and Social Psychology, 36, 1443-1455.
Kerr, N. L., Kramer, G. P., Carroll, J. S., & Alfini, J. J. (1991). On the effectiveness of voir dire in criminal cases with prejudicial pretrial publicity: An empirical study. The American University Law Review, 40, 665-693.
Kline, G. F., & Jess, P. H. (1966). Pretrial publicity: Its effects on law school mock juries." Journalism Quarterly, 43, 113-116.
Kramer, G. P., & Kerr, N. L. (1989). Laboratory simulation and bias in the study of juror behavior: A methodological note. Law and Human Behavior, 13, 89-99.
Kramer, G. P., Kerr, N. L., & Carroll, J. S. (1990). Pretrial publicity, judicial remedies, and jury bias. Law and Human Behavior, 14, 409-438).
McConahay, J. B., Mullin, C. J., & Frederick, J. (1977). The uses of social science in trials with political and racial overtones: The trial of Joan Little. Law and Contemporary Problems, 41, 205-229.
Moran, G., & Cutler, B. L. (1991). The prejudicial impact of pretrial publicity. Journal of Applied Social Psychology, 21, 345-367.
Nietzel, M. T., & Dillehay, R. C. (1983). Psychologists as consultants for changes of venue: The use of public opinion surveys. Law and Human Behavior, 7, 309-335.
Ogloff, J. R. P., Vidmar, N. (1994). The impact of pretrial publicity on jurors: A study to compare the relative effects of television and print media in a child sex abuse case. Law and Human Behavior, 5, 507-525.
Otto, A. L., Penrod, S. D., & Dexter, H. R. (1994). The biasing impact of pretrial publicity on juror judgments. Law and Human Behavior, 18, 453-469.
Padawer-Singer, A., & Barton, A. H. (1975). In R. J. Simon (Ed.), The impact of pretrial publicity on juror's verdicts (pp. 125-139). Beverly Hills, CA: Sage Publications.
Padawer-Singer, A. M., Singer, A., & Singer, R. (1974). Voir Dire by two lawyers: An essential safeguard. Judicature, 9, 386-391.
Padawer-Singer, A., Singer, A., & Singer, R. (1977). Legal and social-psychological research in the effects of pretrial publicity on juries, numerical makeup of juries, nonunanimous verdict requirements. Law and Psychology Review, 3, 71-79.
Riley, S. G. (1973). Pretrial publicity: A field study. Journalism Quarterly, 50, 17-23.
Rollings, H. E., & Blascovich, J. (1977), The case of Patricia Hearst: Pretrial publicity and opinion. Journal of Communication, 58-65.
Shapiro, R. (1994). Secrets of a celebrity lawyer. Columbia Journalism Review, 33(3), 25-29.
Simon, R. J., & Eimermann, T. (1971). The jury finds not guilty: Another look at media influence on the jury. Journalism Quarterly, 48, 343-344.
Simon, R. J. (1966). Murder, juries, and the press. Trans-Action, May-June, 40-42.
Sohn, A. (1976, Spring). Determining guilt or innocence of accused pretrial news stories. Journalism Quarterly, 53, 100-105.
Stephens, M. (1993). Broadcast News, 3rd edition. Fort Worth, TX: Harcourt Brace Jovanovich College Publishers.
Sue, S., Smith, R. E., & Gilbert, R. (1974). Biasing effects of pretrial publicity on judicial decisions. Journal of Criminal Justice, 2, 163-172.
Sue, S., Smith, R. E., & Pedroza, G. (1975). Authoritarianism, pretrial publicity, and awareness of bias in simulated jurors. Psychological Reports, 37, 1299-1302.
Tanford, S., & Penrod, S. (1982). Biases in trials involving defendants charged with multiple offenses. Journal of Applied Social Psychology, 12, 453-480.
Tankard, J. W., Middleton, K., & Rimmer, T. (1979). Compliance with American Bar Association fair trial-free press guidelines. Journalism Quarterly, 56, 464-468.
Thompson, W. C., Fong, G. T., & Rosenhan, D. C. (1981). Inadmissible evidence and juror verdicts. Journal of Personality and Social Psychology, 40, 453-463.
VanDyke, J. M. (1977). Jury selection procedures. Cambridge, MA: Ballinger.
Vidmar, N., & Judson, J. (1981). The use of social sciences in a change of venue application. Canadian Bar Review, 59, 76-102.
Vidmar, N., & Melnitzer, J. (1984). Juror prejudice: An empirical study of a challenge for cause. Osgoode Hall Law Journal, 22, 487-511.
Visher, C. A. (1987). Juror decision making: The importance of evidence. Law and Human Behavior, 11, 1-17.
Wilcox, W. (1970). The press, the jury, and the behavioral sciences. In C. R. Bush (Ed.), Free press and fair trial: Some dimensions of the problem (pp. 49-106). Atlanta: University of Georgia Press.
Zeisel, H., & Diamond, S. S. (1978). The effect of peremptory challenges on jury and verdict: An experiment in a federal district court. Stanford Law Review, 30, 491-531.
Frequency of Article Counts for All Cases
Number of articles Frequency
Conviction Rates by Different Publicity Levels
Publicity Level Acquitted Convicted
None 13 (21.0%) 49 (79.0%)
Low 2 (8.0%) 23 (92.0%)
Moderate 8 (32.0%) 17 (68.0%)
High 4 (18.2%) 18 (81.8%)
Total 27 (20%) 107 (80%) n=134
Histogram of prison sentence lengths
Prison terms by Publicity Levels
Prison Term (in months)
Publicity Level Less than 500 Greater than 500
None 35 (71.4%) 14 (28.6%)
Low 11 (47.8%) 12 (52.2%)
Moderate 8 (47.1%) 9 (52.9%)
High 7 (43.8%) 9 (56.3%)
Total 61 (58%) 44 (42%) n=105