Correctional Service Canada
www.csc-scc.gc.ca
Common menu bar links
FORUM on Corrections Research
Public Support for Criminal Justice Policies: Some Specific Findings
|
One of the things that can be most frustrating for correctional workers is the perceived lack of
public support for the things they do. This can be felt in many ways, from the graphs in magazine
articles charting public disapproval of how well - or how poorly - the criminal justice system is
functioning, to the casual remarks overheard from strangers who place the blame for the outrages
committed by criminals as much on workers in the criminal justice system as on the offenders themselves.
In general, there is a feeling that the public is punitive and opposed to innovative or rehabilitative
correctional programs; this not only discourages the morale of correctional workers, but it may even
deter them from attempting new programs. These perceptions are largely based on unsystematic impressions, or the disproportionate weight that is given to a few loud voices, but they are also said to be supported by "scientific" opinion polls. Indeed, in the past decade a variety of such polls have shown that Canadians are generally poorly informed on criminal justice issues, but that they also seem to favour more punitive and restrictive policies. What do Public Opinion Surveys Tell Us? Some of the surveys that have found these results have been carefully conducted. They include several studies done in this decade for the Ministry of the Solicitor General and the Department of Justice,(1) or for other agencies such as the Canadian Sentencing Commission.(2) For example, a widely cited survey directed by Doob and Roberts(3) documented that Canadians are very poorly informed about crime rates and the detailed workings of the criminal justice system. However, in addition to this abysmal level of ignorance about the system, answers to other surveys have appeared to show a great amount of general dissatisfaction with the system. Gallup polls have regularly asked questions such as: "In general, how would you rate the treatment of criminal offenders?" Whether the question refers to criminal offenders in general, or in prison, or the frequency of parole, or almost any other aspect of the criminal justice system, the answers are almost always the same: People respond that the system is not working well, and that offenders are treated too leniently. Is this really correct? Do Canadians want tougher treatment of offenders? Do they really not support rehabilitative efforts? Or have their thoughts and desires been somehow misrepresented? For the past several years, my students and I have been looking at some of these questions, and by now it seems to us very clear that the previous impression has been much mistaken. We can confirm that most Canadians have large areas of ignorance about the criminal justice system. However, our data say that they are not basically punitive, nor do they really want a generally harsher system. To anticipate the conclusion, we have found that the problem lies in the questions one asks. There's an adage that originated among computer programmers that says "GIGO" or "Garbage in, garbage out". In the present context, it says that the answers one gets are only as meaningful as the questions one starts with. It does not matter how sophisticated one's analyses are, or how careful are the polling techniques. Any question determines the range and types of possible replies, and poor questions may yield unwanted, or even misleading, information. Every question may get replies, but only some give answers. We learned some of this the hard way when we first tentatively started our survey work a few years ago. When we asked subjects some of the very general questions that had been used in other polls, many of them asked for details and further information. They wanted to know what sort of cases we were asking about, and who the offenders were, and lots of other specific case circumstances. Although they were able to answer the global question, many indicated that they felt uncomfortable in doing so. In teaching students, I sometimes discuss how easy it is to see only information that confirms one's expectations instead of objectively observing behaviour. I have sometimes used the aphorism "The subject is always right" to make the point that observations should take precedence over expectations. While my slogan is not to be taken literally, it emphasizes that subjects were giving us a lesson in those first attempts at measuring opinions. The way that they thought about justice decisions was much more complex than was allowed for by the questions we were asking, because those questions requested them to give global answers in situations where they wanted to say "Well, it depends . . This set me to considering just how much the use of global attitudinal questions really does distort the picture of citizens' preferences on criminal justice matters. The result has been a set of studies done over the past four years in which we have looked at judgments that people make in specific cases. Sometimes we compare the results directly with those from the general or global questions. The results we have gotten are quite instructive. A Study of Sentencing One of the first studies in this line looked at attitudes regarding sentencing.(4) Subjects were recruited in a large, busy shopping centre that is centrally located and a major node for public transportation, so that it is frequented by a wide cross-section of the local population. Most people who were approached and told about the purpose of the study were quite agreeable to filling out a questionnaire. The largest part of the study presented a set of scenarios, each describing a criminal case. The instructions clearly stated that the (hypothetical) offenders had already been tried and convicted, and that the subject's task was to decide what would be a proper sentence in each case. The scenarios each contained some details of a crime and an offender, such as the following: A white male, 35 years of age, enters a pharmacy while the cashier is at the rear of the store. He goes around the counter so that when the cashier returns the intruder is behind him. The intruder then tells the surprised cashier not to turn around but to empty the till into a bag. He takes the bag containing $580 and leaves. The offender has been convicted of several previous offences and has served a prison sentence before. What the subjects were not told was that the scenarios differed systematically on several dimensions. In this study, we varied the offense from minor theft to robbery with aggravated assault. Some of the offenders were described as older (as in the case above) and some of them were just old enough to be dealt with in adult court. Some of them had extensive previous records, and some had none. As one might expect, each of these factors made a significant difference in the sentences assigned, using any of several measures of severity. For example, the proportion of subjects who would have imposed a prison sentence (disregarding the length) varied from 8% for a first-time young offender convicted of theft, to a high of 100% for the older habitual offender guilty of robbery and serious bodily injury. From this, we can conclude that subjects were well aware of the differences among cases and that they considered those differences important for sentencing. However, the most interesting thing about the results of this first study is not in the validation of the individual case methodology, but rather in the information it provides about the severity of judgments by the public. Aside from the variations in severity across cases, how punitive were the sentences assigned? Fortunately, at about the time the results were being prepared for publication, the reports from the Canadian Sentencing Commission appeared, and these included some compilations of actual sentences for various offences.(5) This allowed comparison with the results we had obtained from our subjects. Probably to some people's surprise, there was little difference between the average or typical sentences that were thought appropriate by the person on the street and actual court decisions. For example, 82% of subjects would have sentenced those guilty of robbery to prison, as compared to the 84% of actual offenders who are sentenced to imprisonment. This may not be a fair comparison because the two percentages are derived differently, but the same similarity appears if one examines the length of the term assigned when the sentence is imprisonment. For example, if we compare sentences for robbers without previous convictions, the median sentence assigned by subjects was in the category of 7-12 months, as compared to 6-12 months for sentences assigned by the courts. (The categories we used to classify term lengths were a bit different from those of the Sentencing Commission because their reports appeared only after we had collected and analysed our data.) The only real differences in sentencing were for the crime of breaking and entering, where subjects wanted sentences of up to twice as long as those generally assigned by courts. Thus, the results from our study contradict the impression of a generally severe and punitive public, as inferred from the results of surveys with global questions. Of course, we could have had a very atypical sample. While our demographic measures showed a fairly typical representation for adult Canadians, corrections issues do have a high profile in the local area. However, we had anticipated such issues in the design of the study, for in addition to the sentencing task we also asked subjects a set of very general or global questions like those used previously, e.g., "In general, the sentencing of convicted offenders by the courts is ...(much too lenient, too lenient, about right, too harsh, much too harsh)." The answers to these questions were exactly what previous studies had reported: The great majority of subjects said that they were dissatisfied with the criminal justice system and that it was too lenient in all respects. At the same time that they were assigning sentences in hypothetical cases that were entirely typical of those assigned by real courts, 88% said that court sentences were too lenient. There is a profound contradiction here. The two types of questions yielded very different sorts of answers, and there was very little correlation between them - actually, the correlation was statistically significant but trivial. We can say that the answer one gets is determined by the type of questions one asks. While the public disapproves of what they think the system does, if they were given the choices they would probably do just about what the system actually does now. Attitudes toward Parole This message is not a fluke result of one study. We have replicated the basic effect in a very similar project that looked at attitudes toward earned conditional release.(6) In this study, we gave subjects a set of descriptions summarizing the cases of hypothetical inmates who were in prison and applying for parole. We varied the offence as before, although with fewer levels, and we also included the factor of previous record. However, this time we varied a dimension of how much the inmate had been participating in institutional programs while in prison. (In practice, three dimensions is about as many as we can include at the same time without getting into very complicated statistical procedures or becoming sloppy.) Here's an example: The offender was convicted after he broke into a residence and stole a stereo system, TV, and VCR. Before this offence, he had no previous criminal record. During his term in prison, he has upgraded five high school courses taught in the prison and also works in the welding shop. The offender has completed a program dealing with drug problems run by a prison psychologist. While in prison, he has had satisfactory conduct reports and has not had any discipline problems. The results in this study replicate the pattern of what we found for sentencing. As before, responses to the global questions seemed to show punitive attitudes: 82% said that overall the parole system was too lenient. It was also clear that subjects were particularly concerned about violent offenders released on parole, for 54% said that the number of non-violent criminals released on parole was about right, but 87% stated that the number of violent offenders released was too high. At the same time, the answers to the specific cases showed widespread support for the use of conditional release options. The great majority (86%) endorsed some form of release (at least temporary absences, with more liberal release terms promised after successful completion) for the "best risk" case described above. Despite the particular concern with violent crimes indicated in the global results, 58% would have allowed release for a prisoner serving a term for a brutal he had no previous record and had aggravated assault and robbery, if participated in several institutional programs. Even our most serious case, a habitual offender guilty of a serious violent crime and with no participation in educational, vocational, or psychological treatment programs, would have been approved for a release program by 28%. While we do not have access to figures for parole decisions comparable to our own breakdowns, the two seem generally very similar. Even if we did, the comparison would be difficult, because there are no records of how individual members of the Parole Board vote. Still, if the percentages in each of our cases are converted to majority decisions, release programs would have been approved for property offenders in all cases regardless of previous record or program participation, and also for the one violent case cited above. This pattern of response is very much compatible with current policy statements. It certainly is no harsher than actual decisions, and it may even be more lenient. Thus, again we have evidence that citizens choose decisions that are in line with current practice, despite the results of the general attitudinal questions. In addition, the data show very good support for parole and early release programs. Some Extensions Given the functionality of the scenario method, we can use it to answer questions about what people really think the criminal justice system should be doing. By varying the dimensions, we can use it to investigate the model that citizens use in making their decisions, and test whether they follow such logico-legal constructs as the just deserts philosophy. For example, given the significant impact of the institutional behaviour variable, the second study shows that citizens' decisions are at least in part informed by the elements of utilitarian theories of corrections, rather than being entirely based on a just deserts model. We have seen support as well for other current policies that do not conform with the just deserts philosophy. In this respect, we can cite the results of another study of sentencing,(7) in which it was shown that sentences assigned by subjects were less severe in cases where the legal principle of diminished responsibility might apply. In fact, citizens not only assigned significantly shorter sentences to offenders with diagnosed mental illness, but they did the same for cases of chronic emotional disorder without psychosis. This latter effect might indicate positions that are more liberal than current practice, and there is more evidence of such liberality in the finding that the greatest reduction in sentences was for intellectually subnormal offenders. Thus, when we ask for decisions from the public about specifically defined situations, the data lead to conclusions that are quite different than those from global questions. Specific questioning allows one to see what things matter to citizens, and how much. It also provides answers that are not contaminated or dominated by great generalizations or global emotional responses. There are other advantages of the scenario method that should be discussed here. Not only does it provide a more accurate picture of public preferences overall, but it yields much richer information than other methods, both to guide policy making and also to lead to some understanding of the processes and values that determine people's opinions. It would be useful to have such detail routinely, especially if lawmakers wish to establish procedures that are consonant with the values of the public. The level of public support for justice policies clearly depends on situational factors, rather than being derived from absolute and invariable rules. In addition to the value of detailed and specific information, another one of our studies of attitudes toward capital punishment(8) showed another, unexpected advantage of the specific case method over the usual type of question. Along with a set of scenarios of different types of homicide, we asked subjects a global question on capital punishment based on the question used most commonly in the Gallup polls.(9) This question asks respondents whether or not they favour use of the death penalty, but it also allows a "don't know" answer. When we compared results to those from the scenarios, we found that a much higher proportion of subjects would have chosen the death penalty in at least one specific case compared to the proportion that endorsed it on the Gallup question. However, further analyses showed that the great majority of the people who answered "don't know" actually chose the death penalty on at least one scenario, so they should functionally be classed as advocates. When we accounted for these, the results of the two methods became about equivalent and much of the apparent difference disappeared. It is likely that in other circumstances the noncommittal votes might mask opinions in the other direction, for there is no reason to expect that they would always be in the direction we found. What is certain is that the assumption that they will split in proportions like those of committed opinions is demonstrably false. To have made that assumption would have led to a result that seriously misrepresented the split of public opinion. If one wants to use very general questions to measure opinion, there is a conundrum here. Many people will object to answering a question that gives them only "yes" or "no" options, ignoring the complexity that they perceive in an issue. The "don't know" option is often interpreted by respondents as "it depends", so it might seem a good alternative, except that, as we have seen, this can lead to sizable errors. In contrast, one does not have this problem with specific case decisions, for we have found that the better the cases are described the less trouble people report in making decisions. Conclusions Conventional opinion polling has become widely accepted as a way of measuring public opinion in the past few decades, largely because of its success in predicting voting patterns in elections. The sampling rules are well known and carefully observed by consulting firms specializing in enumerating the public mood on political questions. The accuracy of polls in predicting the rise and fall of political fortunes has brought them much credibility, especially among politicians and other policy makers. As a result, the techniques used to measure preferences among candidates have also been used (to measure attitudes on complex issues, such as the treatment of criminal offenders. However, there are some profound differences between the two situations. In an election the choices (candidates or parties) are a small set of completely specified options, which are usually well known to respondents even before the question is asked. Moreover, all the specific factors that might affect the choice are determined: the date of the election, the issues, the claims, the charges, the context of other events, are all established in particulars. It is very different if one wants to measure something like attitudes toward the criminal justice system. Cases vary widely, as do the dispositions that people consider just and proper. Even relatively uninformed citizens are aware of a great many cases, and they are likely to have some implicit dimensions of variation that will allow the production of an indefinitely large number of variations. Asking a person to give an overall judgment requires some way of averaging all the cases available to memory or imagination. Not only does this averaging lose much information, but we do not know how it is done. The required averaging can occur in a variety of ways, many of which will lead to serious distortion. For example, we can hypothesize that the wording or context of global questions will sometimes cause people to think of the worst case they are aware of, and to base their judgment on this case: the result will be that judgments will express severe disapproval of the entire system. Although we do not have direct evidence of exactly what happens when people answer global questions, our studies confirm the arguments above and show that overly simple questions can lead to misleading answers about public opinion on criminal justice questions. Criminal justice decisions are complexly determined, and the specifics of each case are critical in making decisions, even if general policies for making decisions are clearly articulated. To reduce the choices to the generalizations of global questions is to make the answers vacuous and not to be trusted. GIGO. On the other hand, methods that gather detailed and specific information, such as our scenario technique, are more informative and are also more likely to yield an accurate picture overall. Using such methods, it may be possible to construct a workable model of the cognitive processes that people use in evaluating the system, and to see the values and perspectives they use to make decisions. We have actually stated some hypotheses along these lines, and we hope to provide some experimental tests of them in the near future. In the meantime, what we can say is that the Canadian public is not so punitive as they have sometimes been depicted in the past, or as they might seem from the results of global polling techniques. There may be some very audible individuals of the "hang 'em high" mentality, but in general the Canadians we have seen in our research are pragmatic in their approach to decisions on criminal justice matters. The research we have done gives the overall impression that Canadians base their decisions on the desire to control crime and minimize its harmful effects; punishment for its own sake does not seem to be among the bases of their decisions. We can also say that our subjects functionally endorsed most current principles in the criminal justice system, because their own decisions generally coincide with actual current practice. There may certainly be cases where public opinion will massively disagree with decisions in actual cases. Sometimes this may occur because of misrepresentation or poor dissemination of relevant information. At other times, it probably shows the difficulties of making decisions when so many factors are relevant, for even people with very similar values may differ in cases where decisions pivot like angels on the proverbial pin. In any case, it should not be forgotten that the system can make mistakes, and given the argument here that the public generally agrees with the bases of decisions by the system, serious public outcry about specific cases should be taken very seriously as indications of possible errors. If our general conclusion is warranted by the results of our data, it still appears to contradict the consistent findings that people evaluate the system negatively and say that it is too liberal. If this is so, then what is needed is not sterner policies but better public information. Citizens answer the global questions on the basis of some generalization from their impressions of the system. If at the same time their choices in actual cases would be similar to those made by the system they condemn, it must be those impressions that are at fault. What is needed are changes in public information about criminal justice. In the short term, a substantial public information campaign would be very useful. On the basis of some anecdotal findings in our data, we believe that statements of concrete facts from politicians can have effects in this area. When we first started in the area several years ago, we found, in agreement with the data of other researchers,(10) that very few subjects realized the size of the difference between US and Canadian rates for violent crime. In our most recent study,(11) with people recruited from the same population in the same ways, we found quite a different result, with close to two thirds hitting the correct range. All we can see in the interim that might account for the change are repeated statements about the relative rates by officials at various levels. Self-serving politicking it may have been, but it was also factually correct and educationally effective. If this can be widened, and accompanied by substantial reform in the educational system to inform our children about how the criminal justice system works, then we would expect that expressed citizen approval of the system would markedly increase. Only a few years ago, the federal correctional system in Canada seemed hesitant to disclose information about its operation, and this has now changed so that openness and accessibility are actively promoted. It is time to move further, for there is a need for an active effort to disseminate information to the public about actual practices in the entire criminal justice system. Edward Zamble, a professor of psychology at Queen's University, Kingston, has been involved in correctional research in the area of criminal behaviour for over 10 years. The Correctional Service of Canada and specific federal institutions have called on Dr. Zamble's expertise on numerous occasions. He is presently acting as a consultant to the Ontario Regional Treatment Centre. Dr. Zamble earned his Ph.D. at Yale University. (1)Doob, A.N., and Roberts, J.V. (1983). Sentencing: An Analysis of the Public's View. Ottawa: Department of Justice. (2)Roberts, J.V. (1988). Empirical Research on Sentencing: Research Reports of the Canadian Sentencing Commission, #1. Ottawa: Department of Justice. Several others in the series are also relevant. They are summarized in Canadian Sentencing Commission (1987). Sentencing Reform: A Canadian Approach. Ottawa: The Commission. (3)Op. cit. (4)Zamble, E., and Kaim, K.L. (in press). "General and Specific Measures of Public Attitudes toward Sentencing." Canadian Journal of Behaviourial Science. (5)Hann, R.G., and Kopelman, F. (1987). Break and Enter, Robbery, and Extortion Offenses: Custodial and Probation Sentences: 1984/85. Ottawa: Department of Justice; Hann, R.G., and Kopelman, F. (1987). Assault: Custodial and Probation Sentences: 1984/85. Ottawa: Department of Justice. (6)Cumberland, J., and Zamble, F. (1989). "General and Specific Measures of Public Attitudes: Release Decisions." Unpublished manuscript, Queen's University; submitted for publication. (7)Banks, D. (1988). "The Effects of Mental Impairment on Sentencing." B.A. (Honours) thesis, Queen's University. (8)Mills, J., and Zamble, E. (1989). "General Question, Specific Questions: Canadian Public Attitudes toward the Death Penalty." Unpublished manuscript, Queen's University; submitted for publication. (9)Gallup Poll of Canada (24 November 1979). Gallup Poll Report. Canadian Institute of Public Opinion. (10)Doob and Roberts, op. cit. (11)Cumberland and Zamble, op. cit. |
