On Exploring the "Dark Figure" of *

By ALBERT D. BIDERMAN and ALBERT J. REISS, JR.

ABSTRACT: The history of criminal bears testimony to a search for a measure of "criminality" present among a population, a search that led increasingly to a concern about the "dark figure" of crime—that is, about occurrences that by some criteria are called crime yet that are not registered in the statistics of whatever agency was the source of the data being used. Contending arguments arose about the dark figure between the "realists" who emphasized the virtues of com- pleteness with which data represent the "real crime" that takes place and the "institutionalists" who emphasize that crime can have valid meaning only in terms of organized, legitimate social responses to it. This paper examines these arguments in the context of police and statistics as measures of crime in a population. It concludes that in exploring the dark figure of crime, the primary question is not how much of it becomes revealed but rather what will be the selective proper- ties of any particular innovation for its illumination. Any set of crime statistics, including those of survey research, involve some evaluative, institutional processing of people’s reports. Concepts, definitions, quantitative models, and theories must be adjusted to the fact that the data are not some objectively observable universe of "criminal acts," but rather those events defined, captured, and processed as such by some institutional mechanism.

Albert D. Biderman, Ph.D., Washington, D.C., is Senior Research Associate, Bureau of Social Science Research, Inc., Washington, D.C. He explored uses of crime statistics as social indicators in a recent monograph on Social Indicators and Goals (1966). Albert J. Reiss, Jr., Ph.D., Ann Arbor, Michigan, is Professor and Chairman, Depart- ment of , and Director, Center for Research on Social Organization, University of Michigan. He is the author of some thirty articles and studies on crime, , and law enforcement.

* The support of the Russell Sage Foundation is gratefully acknowledged. 1 2

began tistics to supplement, if not supplant, STATISTICALwith the development of moral sta- them. The contending arguments were tistics.’ No subject has dominated the fundamentally between what we can field of criminal statistics more since loosely term &dquo;realist&dquo; as opposed to &dquo;in- its inception than the search for the key stitutionalist&dquo; emphases.3 The former moral -a measure of the &dquo;crim- emphasized the virtues of completeness inality&dquo; present among a population. with which data represented the &dquo;real This search led increasingly to a con- crime that takes place.&dquo; The institu- cern about the &dquo;dark figure&dquo; of crime- tionalist perspective emphasized that that is, about occurrences that by some crime could have valid meaning only in criteria are called crime yet that are terms of organized, legitimate social not registered in the statistics of what- responses to it. ever agency was the source of the data The ultimate juristic view is that a being used.22 given crime is not validly known to have The history of criminal statistics tes- taken place until a court finds someone tifies to continuing contention between guilty of that offense. Only at that those who sought to bring more of the point in the process has there been an dark figure to statistical light and those irrevocable decision as to the evidence who deplored elements of invalidity in regarding the objective facts in relation each such attempt. The major object to their legal significance. Outside the of this contention for over a century United States, there was little resistance was police statistics. Both official and to utilizing data from earlier stages in scholarly comprehensions of the inci- the adjudicatory process, such as prose- dence of crime were almost exclusively cution, indictment, arraignment, or even based on statistics of indictments or ad- investigation, particularly in legal judications. There were those who 3 William Douglas Morrison stated the dis- sought the development of police sta- tinction rather well in a paper before the Royal Statistical Society in 1897: "If ... we 1 The French are generally credited with the are anxious to know how the criminal law early development of moral statistics. Es- is being administered, we shall analyse and pecially noteworthy is the work of A. M. classify the contents of the statistics from that Guerry, Essai sur la statistique morale de la point of view. If on the other hand we desire France (Paris, 1833). Guerry calculated rates to know the movement of crime, the criminal of against persons and against property conditions of the community, and the relative for 86 departments of France and age-sex spe- value of the several methods by which these cific crime rates for seventeen crimes against methods are to, be ascertained, we shall adopt the person and seventeen against property. The a somewhat different method of classifying the rates were presented in tabular, graphic, and contents of criminal statistics. I have ven- cartographic forms. tured to classify criminal statistics into police 2 The earliest published discussion of the statistics, judicial statistics, and sta- dark-figure problem that we have been able tistics because I desire, at least in the first to find is that of Bulwer. In his two-volume place, to point out the amount of weight to treatise on France, published in 1836, Bulwer be attached to each of these methods of re- devoted an entire chapter to crime in France, cording the nature and proportions of crime." based primarily on A. M. Guerry’s major —"The Interpretation of Criminal Statistics," work. Bulwer (pp. 174-175) discusses the Journal of the Royal Statistical Society, Vol. problem of using either offenses known or of LX, Part I (March 1897), pp. 1-24, at pp. the accused as measures of crime and con- 1-2. Also: "But it would be a mistake to cludes that, despite their limitations, they are suppose that the number of crimes known to more accurate than calculations based on con- the police is a complete index of the total victions. See Henry Lytton Bulwer, France, yearly volume of crime. The actual number Social, Literary, Political, Vol. I, Book I: of offenses annually committed is always Crime (London: Richard Bentley, 1836), pp. largely in excess of the number of officially 169-210. recorded crimes" (Ibid., p. 4). 3 systems where there are police magis- they were placed after court statistics trates.44 In all countries, however, most with the statement: criminologists were less ready to credit the competence of the police to make The tables of the results of judicial pro- determinations of the objective facts ceedings, which are at once the most im- portant, the most definite, and the most and to classify them validly-police accurate of all criminal statistics, occupy in terms of competence being judged the first place. The tables as to police and the of in- legitimacy, skill, adequacy action ... are of less statistical value, formation available to the police. Pre- and follow in a subordinate position.6 sumably, the lower social status of the police than of the bench-and, correla- Not until 1923 did the argument over tively, the greater political power of the their merit abate sufficiently in England judiciary-together with the loose fash- so that they were accepted as a valid ion in which police systems were for basis for estimating crime.7 Even to- long grafted on the legal-institutional day, in England, police statistics are systems, has much to do with these considered less reliable than judicial views.5 statistics. There has been a long contest to gain An additional difficulty inhered in the institutional acceptance for police sta- localistic nature of police organization tistics over opposition from legalistic in the United States. Not only did this traditionalism. In England, a plan was make for dubiousness about the judg- worked out for the collection of police ment and record-keeping capabilities of statistics on a uniform and national police in all but the larger jurisdictions, basis in 1856, and they have been a but producing national series also posed regular part of the annual report of formidable problems of standardization criminal statistics since 1857. While and compilation of data from a multi- from the outset, police statistics were tude of jurisdictions having a myriad of logically placed prior to judicial sta- laws, definitions, and practices. The tistics in the published volumes, in 1893 present voluntary system of national crime in the United States 4 In France, for example, early statistical reporting compilations of crime provided information on owed its form and many of its limita- accusations, accusés, acquités, and condamnés. tions to the fact that the national See Recherches statistiques sur la ville de government cannot (at least not read- Paris et le department de la Seine (A Paris local to de l’Imprimerie Royale, 1821-1830). See also ily) compel governments report Guerry, op. cit. on their operations.&dquo; 5 The Report on Criminal Statistics, U.S. As police statistics were legitimated, National Commission on Law Observance and statistics on arrests generally Enforcement (Washington, D.C.: U.S. Gov- gained earlier than those based on ernment Printing Office, 1931) stated con- acceptance temporary views in the United States: "If it took the highly centralized English Govern- 6 Great Britain, Judicial Statistics, England ment 66 years to get its famous and highly and Wales, 1893, Part I: Criminal Statistics, efficient police to report correctly crimes known p. 14. 7 to the police, it is evident that it will be many Ibid., 1923, p. 5. 8 years before our decentralized and nonprofes- For a good history and discussion of the sional police forces can be induced to make problems of uniform crime reporting in the trustworthy reports of crimes known to the United States during the formative period, see police" (p. 55). After more than a third of U.S., Department of , Ten Years of a century, patience is still being counseled: Uniform Crime Reporting, 1930-1939: A Re- see Peter P. Lejins, "Uniform Crime Reports," port by the Federal Bureau of Investigation Michigan Law Review, 64 (April 1966), pp. (Washington, D.C.: U.S. Government Printing 1011-1030. Office, 1939), esp. chap. v. 4

citizen complaints or reports of offenses iceberg of crime&dquo; looks progressively known to the police. Arrests involve different from the huge submerged mass. the legal authority system, while the The classically realist view in the use status of a citizen complaint is moot. of police statistics as an index of crimi- Eventually, however, realist perspectives nality attaches greatest emphasis to prevailed, and the Uniform Crime Re- those police data which are least depen- ports (UCR’s) from the outset gathered dent on agency action. Arrests, which information on all offenses reported or vary with the extent, skill, and dis- known to the police. Nonetheless, there cretion of police activity, thus are re- is a strong disposition to count as garded as a less satisfactory basis for offenses only those that are substanti- an index of criminality than complaints, ated by police investigation-a process reports, and directly observed (&dquo;police of &dquo;unfounding&dquo; citizen complaints. on-view&dquo;) &dquo;crimes.&dquo; The realist view, Published reports of UCR count only at the same time, held that even police the number of &dquo;actual offenses&dquo; that statistics distort the &dquo;real crime prob- survive police &dquo;unfounding&dquo; procedures. lem.&dquo; An &dquo;index&dquo; of &dquo;crime,&dquo; there- To a considerable degree, precisely fore, was devised that would provide a what the institutional view regarded as measure of the &dquo;crime problem&dquo; least the vices of police statistics, the realist subject to effects of jurisdiction. The one regarded as sources of virtue. This UCR annual report states the case: was the absence of any &dquo;institutional Not all crimes come to the atten- of the data-the selecting, readily processing&dquo; tion of the police; not all crimes are of and of records of defining, winnowing sufficient importance to be significant in an events by legitimate organizations of index; and not all important crimes occur the legal system in accordance with with enough regularity to be meaningful in legally established evidentiary and eval- an index.lo uative criteria and The procedures. Among all offenses known to the classical statement for American police police, those were selected for index statistics Sellin sums by up why police for in at statistics of &dquo;offenses known&dquo; purposes which, theory least, provide the function most as the &dquo;best index&dquo; of crime: police nearly pas- sive recorders and nondiscretionary clas- In general, it may be said that the value sifiers of events that take place. Index of a crime rate for index is in purposes crimes are, in each case, offenses which inverse ratio to the distance be- procedural come to the attention of the tween the commission of the crime and the largely police by complaints from those vic- recording of it as a statistical unit. An timized the event. Violations which index based on crimes reported to or known by do not involve or which to the police is superior to others, and an specific victims, index based on statistics of penal treat- largely or wholly come to be registered ment, particularly prison statistics, is the only as a result of police action, such 9 poorest.9 as disorderly conduct, assaulting an and stolen Each remove from the crime, in terms officer, receiving property, are excluded from the measure. Of- of official procedures, leaves more of the in a commu- actual crime taking place 10 U.S., Department of Justice, Federal Bu- nity submerged in the dark figure. reau of Investigation, Crime in the United Each procedural step, furthermore, is so States: Uniform Crime Reports (Washington, selective that the &dquo;visible tip of the D.C.: U.S. Government Printing Office, 1930- [annually]). The above quotation 9 Encyclopedia of the Social Sciences, Vol. 4, is taken from the annual report for 1964, p. 565. p. 48. 5 fenses also deemed unsuitable for an pointed out that police statistics re- index are those unlikely to be reported flected only an unknown and selective to the police either because they involve portion of &dquo;all crime&dquo; and that they only persons disinclined toward police distorted in many ways the kinds of action, as is usually the case for gam- crime they did reflect. Interestingly, bling, prostitution, and other illegal defenses of police statistics have come services, or because the offenses are fre- to rest increasingly on institutionalist quently too trivial to be &dquo;worth the arguments, rather than the realist ones bother&dquo; of reporting, such as petty lar- to which they largely owe their accep- cenies and acts of malicious mischief. tance. In rebutting criticisms of UCR, An additional criterion of the realist po- for example, Lejins writes: sition was that the criminal act should The existence of serious offenses not re- be uniformly classifiable, independently ported in the police statistics should not of the varying local laws and practices. be accorded exaggerated meaning in the Miscegenation, until recently, afforded a sense of detracting from the significance clear example of an offense unsuitable of the criminal activity that is reflected in for an index. the Reports, since the latter do en- Realist views in the United States compass the bulk of the conventional, seri- became predominant, first in crimino- ous behavior to which society chooses to logical theory and then in practice, react through its public law enforcement with the establishment in 1929 of the agencies.ii compilation of a national crime sta- It is beyond the scope of this essay tistics series by the Federal Bureau of to recapitulate the many criticisms and Investigation from voluntary reports by defenses that have been made of police police agencies. The UCR index of statistics, generally, and the Crime In- crime that resulted from the application dex, in particular.l2 It is important of these &dquo;realist&dquo; criteria consists of here, however, to formulate the thrust counts of offenses known to the police of these criticisms with respect to the falling in seven predatory, common-law misleading social implications that were classifications: homicide, forcible rape, seen in police statistics. robbery, aggravated assault, burglary, Because of the partial and selective larceny ($50 and over), and automobile nature of the police data, comparisons theft. based on them of variations in &dquo;actual Although police statistics gained ac- 11 ceptance largely as a result of realists’ Lejins, op. cit., p. 1010. 12 For efforts to achieve more comprehensive recent criticisms, see Daniel Glaser, "National Goals and Indicators for the Reduc- and less selective indexes than were pro- tion of Crime and Delinquency," Social Goals vided by institutional data, the victory and Indicators for American Society, Vol. I, of police statistics had barely begun to THE ANNALS, Vol. 371 (May 1967), pp. 104- be consolidated before some realists 126 ; Stanton Wheeler, "Criminal Statistics: A Reformulation of attacked these statistics on the same the Problem," Journal of Criminal Law and Criminology, Vol. 58 Police statistics were chal- (Sep- grounds. tember 1967) ; Marvin E. Wolfgang, "Uniform lenged as not reflecting &dquo;the real crime Crime Reports: A Critical Appraisal," Uni- picture.&dquo; The criticism, as had been versity of Pennsylvania Law Review, Vol. 111 the case with older dissatisfactions with (April 1963), pp. 708-738. For a defense, court and prison statistics, concentrated see Lejins, op. cit., pp. 1011-1130. See also Albert D. Biderman, "Social Indicators on the &dquo;real crime&dquo; that the escaped and Goals," in Raymond A. Bauer (ed.), police data rather than on invalid classi- Social Indicators (Cambridge, Mass: The fication of events as crimes. Critics M.I.T. Press, 1966). 6 crime&dquo; over time, between places, and property as suggested by arrest and among components of the population, juvenile delinquency statistics. 14 are all held to be grossly invalid. Fur- Despite the great effort devoted to thermore, because of the fundamental developing and operating a uniform re- subordination of police statistics to the porting system, the use of the police particular normative perspectives and data for interarea comparisons has also workings of this institution, it is con- been subject to vigorous criticism on a tended, there are limitations and distor- variety of grounds. One form of criti- tions inherent in the significance drawn cism pointed to the many instances in from them for social policy. which abrupt and vast increases of Barely masked in these contentions crime figures for cities occurred when regarding statistics have been more fun- police reforms curtailed the practice of damental ideological cleavages.13 It is &dquo;killing crime on the books.&dquo; Police useful to make explicit that much of departments and political administra- the argument over appropriate indexes tions controlling them, it is often al- of criminality tends to array on one side leged, frequently have too great a stake those who regard a person’s social sta- in the effects of their crime figures on tus as largely a product of his own their &dquo;image&dquo; to be trusted to report vices and virtues, and on the other, fully and honestly. Beyond these those who interpret status, as well as qualms regarding &dquo;statistical conflicts vices and virtues, as largely a product of interest,&dquo; there was evidence that po- of socially conferred advantages and lice departments with effective and cen- disadvantages. With regard to meas- tralized controls over the reporting by ures for dealing with crime, the cleav- individual officers and divisions reflected ages are, for example, between deter- more of &dquo;true crime&dquo; in their commu- rence and social amelioration, or be- nities than did less tightly organized tween punishment and therapy. departments.&dquo; Ideological cleavage had clear expres- In recent years, the strongest com- sion in Sutherland’s of the plaint against police statistics has sug- failure of conventional crime statistics gested that much of the rapid and ex- to reflect &dquo;white-collar crime.&dquo; In treme reported increases in the extent prevalence and in economic and social of criminality are spurious, being but a effects, Sutherland sought to show, law surfacing of what has heretofore been violations by a person of &dquo;the upper socioeconomic class in the course of his 14 For a recent summary, see Harwin L. "Socioeconomic Status and activities&dquo; were more con- Voss, Reported occupational Delinquent Behavior," Social Problems, 13 than the lower- sequential typically (Winter 1966), pp. 314-324. See also Albert class crimes that comprised the index. J. Reiss, Jr. and Albert Lewis Rhodes, "The Something of the same thrust was in- Distribution of Juvenile Delinquency in the herent in the innovation of self-report- Social Class Structure," American Sociological Review, 26 (October 1961), pp. 720-732. studies. The of ing high proportions 15 In U.S., President’s Commission on Law middle-class persons who admit having Enforcement and Administration of Justice, committed serious delicts indicated both The Challenge of Crime in a Free Society, that the dark figure of crime must be hereinafter referred to as General Report (Washington, D.C.: U.S. Government Printing of vast at the very proportions and, Office, 1967), of the two recommendations that &dquo;criminal&dquo; behavior was not least, concerning the measurement of crime, one was nearly as exclusively a lower-class that each city adopt centralized procedures for handling crime reports from cities (pp. 27, 13 Lejins, op. cit., pp. 1029-1030. 293). 7 in the dark figure. The most general it went on to say, in this way, &dquo;have argument takes the form that crime several times produced large paper in- 19 statistics are as much (and perhaps creases in crime.&dquo; Current attempts more) a product of modern urban social at improving police-community relations organization as are the so-called urban conceivably could produce sharp &dquo;paper forms of criminal behavior For ex- increases&dquo; in some classes of crime were ample, it is suggested that the profes- they to result in a greater disposition of sionalization and bureaucratization of citizens to report offenses. 20 police forces with centralized command But no basis exists for forming pro- and control leads to improved record- portionate estimates of what kinds of keeping and greater use of formal, as criminal behavior are reported to the opposed to informal, police procedures, police. Primarily with this problem in with consequent increases in figures of mind, the National Crime Commission offenses and arrests. And it is main- undertook exploration of the use of tained that as larger proportions of the cross-sectional survey methods.21 A cen- population become integrated into the tral idea was that one could discover dominant society and come to share its crimes not known to the police by normative conceptions, more people mo- screening random samples of the popula- bilize the police to enforce middle-class tion to find the victims of these crimes. norms regarding property, violence, and SAMPLE SURVEYS AND public deportment. At the same time, THE DARK FIGURE these public agencies become less dis- Given the of what is posed toward a tolerant view and in- growth literally formal processing of a vast citizen-interviewing industry in That improvements in law enforce- the United States, it perhaps is sur- ment frequently have the effect of de- prising that the sample survey had not hitherto been to ex- creasing the dark figure, and conse- applied systematic quently inflating statistics used to judge amination of the crime problem.22 the magnitude of the crime problem, Perhaps, the very availability of the in correctional can be disconcerting for those planning captive populations insti- innovational reforms. The President’s tutions (if we include educational in- Commission on Law Enforcement and stitutions as such), and of the neatly

Administration of 19 Justice (hereinafter Ibid., p. 26. referred to as the National Crime 20 Neil Rackham, "The Crime-Cut Cam- Commission), for example, produced paign," New Society, 238 (April 1967), pp. 563-564. a table illustrating reporting-system 21 Albert D. Biderman, "Surveys of Popula- in a dozen cities that changes major tion Samples for Estimating Crime Incidence," resulted in Crime Index increases of in this issue of THE ANNALS, pp. 16-33. Sub- from 27 per cent to more than 200 sequently, a similar survey was undertaken by the Government Social of Great per cent over the immediately preceding Survey Britain. The results of this work were not The nation’s two report.18 largest cities, available at the time of writing. (Personal 16 For a discussion, see Albert D. Biderman, communication from Louis Moss, Director, "Social Indicators and Goals," in Bauer (ed.), Great Britain Government Social Survey, July op. cit., pp. 124-125. 18, 1967.) 17 John Kitsuse and Aaron Cicourel, "A 22 A search of the Roper Public Opinion Note on the Use of ," Center poll repository disclosed that, Problems, 11 (Fall 1963), pp. 131-139. until 1964, public opinion surveys had given 18 U.S., President’s Commission on Law En- little attention to crime, except for polling forcement and Administration of Justice, sentiments regarding capital punishment and General Report, op. cit., p. 25. juvenile delinquency. 8 compiled agency statistics, diverted at- some individuals to victimization, in- tention from such possibilities. cluding factors considerably more subtle Neglect of the interview survey rep- than such commonly recognized contri- resents some discontinuity in the history buting acts as negligence and provoca- of social research on crime, however. tion by which some persons precipitate In the nineteenth century, a far more criminal acts toward themselves.&dquo; prominent place was accorded surveys Attention to the victim has also been of populations for knowledge. Henry urged from a quite different evaluative Mayhew and Charles Booth, who often standpoint. In prefacing the 1963 vol- are credited with having set the path ume of Crime in the United States: for the survey movement, were, for ex- Uniform Crime Reports, the Director ample, very conscious that the signi- of the Federal Bureau of Investigation ficance of crime for the poor of the city wrote: resided as much in their being its vic- tims as in their being its contributors. 23 Statistics herein are published in terms of the number of crimes and Booth’s systematic survey sought to in- reported per- vestigate &dquo;the numerical relation which sons arrested. At the same time, they also a count of millions of victims. and bear to represent poverty, misery, depravity While some of these victims may have the regular earnings and comparative been &dquo;merely inconvenienced,&dquo; the vast comfort, and to describe the general majority suffered property losses they could conditions under which each class ill afford and many lost their physical or lives. 11 24 mental health while others lost their lives. Although there has been an occa- Nevertheless, many impassioned and articu- sional specific suggestion of using late pleas are being made today on behalf of the offender tending to ignore the victim &dquo;Gallup Poll&dquo; methods’5 as a specific and obscuring the right of a free society check on official the current statistics, to equal protection under the law.27 turning to the cross-sectional survey method has received probably greater In the 123 pages of &dquo;general United impetus from recent comments that States crime statistics&dquo; in the 1963 criminology has been neglecting the vic- Uni f orm Crime Reports, however, the tim in its concentration on the criminal. equivalent of only two pages provides These that attention in writings argue any information on the victims of crime has been misdirected criminology by the -and this only if we include categories to the victim of usual tendency regard of property as &dquo;victims.&dquo; But two crimes as a and purely passive acci- tables dealt with persons as victims: dental of the criminal act. A target one on &dquo;Murder Victims [by age]- science of &dquo;&dquo; is proposed to Weapons Used&dquo; and one on &dquo;Murder explain social, psychological, and be- Victims by Age, Sex and Race.&dquo; havioral characteristics that predispose In sponsoring cross-sectional inter- surveys, the National Crime 23 Henry Mayhew, London Labour and the viewing to be to London Poor: Cyclopedia of the Conditions Commission hoped able develop and Earnings of Those That Will Not Work (London: Charles Griffin, 1861); Charles 26 See selected bibliography in B. Mendel- Booth, Labour and Life of the People sohn, "The Origin of the Doctrine of Vic- (London: Williams and Norgate, 1891). timology," Excerpta Criminologica, Vol. 3 24 Booth, op. cit., Vol. 1, p. 6. (May-June, 1963). 25 Inkera Anttila, "The Criminological Sig- 27 U.S., Department of Justice, Federal Bu- nificance of Unregistered Criminality," Ex- reau of Investigation, Crime in the United cerpta Criminologica, Vol. 4 (1964), pp. 411- States: Uniform Crime Reports, op. cit., 1963, 414. p. vii. 9 data on the characteristics of victims poses and perspectives of a sociologistic that would go considerably beyond the institutionalist: scant information available from police Indeed in modern societies where bureau- sources. Since the same were surveys cratically organized agencies are increas- directed toward developing data on citi- ingly invested with social control func- zens’ behavior and attitudes toward the tions, the activities of such agencies are crime problem and toward law enforce- centrally important sources and contexts ment, it would also be possible to relate which generate as well as maintain defini- such attitudes to actual experience with tions of deviance which produce popula- tions of deviants. Thus, rates of deviance crime as contrasted with secondary constructed by the use of statistics rou- influences such as the mass media. tinely issued by these agencies are social facts par excellence. 28 HOW MUCH CRIME Is THERE? This quotation discriminates the use of data from the of the It should be apparent that the answer agency perspective to the question of how much crime there institutional processing of observations from that of the realist. Realists use is depends to a great extent upon as a tool for observations of whether one phrases the question from agencies realities external to them. an institutionalist or a realist position. The choice of indicators and their labels, The ideal mechanism for a realist would be a universal surveillance of time to a great extent, bears marks of these positions. The hallmarks of the realists and space by recording mechanisms sensitive to all are the prevalence of criminals and their completely pertinent Yet con- acts of crime; more recently, of victims. phenomena. any organization The hallmarks of the institutionalists fronts technical limitations to observa- tion. Inherent in observational are the prevalence of only such of these any are errors from as survive institutional validation. system prob- observation and meas- But there are no rates without some ability, faulty in organized intelligence system, whether urement, imperfections operational translation of observational that of the scientist, the police, or the categories, and to flow and feedback jurist. The sample survey, the citizen’s impedance in communication. mobilization of the police, and the pre- such as the have trial and trial proceedings are all organ- Organizations police their own surveillance For ized intelligence systems that process purposes. inside the an events and people to determine their processors organization, crime status. The criteria of knowing, ideal surveillance mechanism is not an defining, and processing lie in organiza- alien concept. However, action and ob- tion. servational mechanisms are inextricably linked one to the other. Given the diversity of sources and From a realist of this types of information on crimes, the pro- point view, makes the serve cedures that one develops for determin- linkage organization observational For ex- ing whether an event has occurred and purposes poorly. such as who was involved in it must vary. It is ample, operational organizations doubtful, therefore, whether, logically, the police or courts choose not to ob- serve more than can with any currently organized way of knowing they process and makes possible the computation of a given resources, they selectively screen observations to fit measure of crime that can serve equally organizational and tactics. Further- all purposes and perspectives. goals, strategy, Kitsuse and Cicourel state the pur- 28 Kitsuse and Cicourel, op. cit., p. 139. 10 more, organizations suffer from their it is important to observe. Nonetheless, own form of deviance, the subversion of technical limitations as well as social organizational goals by their members. choice reflect what is responded to. Three strategies are open to realists in Indeed, realism itself is a system norm overcoming these organizational barriers for members of organizations. to information. First, they can insulate From the standpoint of a scientific the surveillance apparatus from opera- criminology, a defect of the institution- tions, as, for example, through the crea- alist point of view is that it uses con- tion of central communications, intelli- cepts and data derived exclusively from gence, and records divisions.29 Second, those employed by formal organizations they can undertake independent surveil- of the law-enforcement and legal sys- lance of the operating system, by moni- tems. There is more to social life than toring through outside observers either its formally organized aspects. For the operations or records of the organi- scientific purposes, independently or- zation.3° Third, they can develop sur- ganized observations employing appro- veillance completely independent of the priate concepts and tools of measure- organization. The sample survey of the ment are necessary. public is one of a variety of such de- Thus, attacking the institutionalist vices.31 A separate intelligence organi- point of view, Glaser points out that: zation is another. The deviance of members of the sys- Variation in the public definition of most crimes is not es- tem from system norms, with respect to predatory appreciable, pecially outside of so-called &dquo;white-collar reporting as well as technical limitations crimes.&dquo; The categories of predatory to observation that are errors from a crimes most commonly distinguished in the realist point of view, negates the social- law-for example, murder, robbery, bur- choice interpretations of organizational glary, theft, fraud, and rape-have almost data made by institutionalists. From a everywhere and always been employed to radical institutionalist point of view, denote essentially the same types of behav- these errors are treated, in part, as ir- ior as criminal. In almost all societies, relevant, in that the differential sensi- they comprise the majority of acts for which severe sanctions are im- tivity of surveillance reflects, to a sub- negative stantial degree, social choices of what posed.32

29 For a discussion of this strategy by police But, clearly, a large proportion of these chiefs, see David J. Bordua and Albert J. &dquo;crimes&dquo; are &dquo;processed,&dquo; if at all, only Reiss, Jr., "Command, Control, and Charisma: by informal mechanisms. &dquo;The crimi- Reflections on Police Bureaucracy," American nal offense&dquo; itself is an social Journal of Sociology, Vol. 72 (July 1966), pp. important 68-76. transaction, quite apart from social 30 For an organized of transactions that ensue thereafter. It the police, see Donald J. Black and Albert J. should be evident that police data, Reiss, Jr., "Patterns of Behavior in Police whether on offenses or and Citizen Transactions," in Albert J. Reiss arrests, exag- (ed.), Studies in Crime and Law Enforcement gerate the incidence of those kinds of in Major Metropolitan Areas, U.S. President’s offenses for which an identifiable person Commission on Law Enforcement and Ad- is suspect, in that these are more likely ministration of Field III Justice Survey to be to the and (Washington, D.C.: U.S. Government Printing reported police proc- Office, 1967). essed by the department through in- 31 See Albert D. Biderman, "Surveys of vestigation. Population Samples for Estimating Crime The neglect of victims in processing Incidence," in this issue of THE ANNALS, pp. 16-33. 32 Glaser, op. cit., p. 107. 11 by law-enforcement, legal, and correc- Given the fact that a single event tional agencies is another case in point. may produce multiple victimization and Offense rates, today, are based on data multiple offenses and that, over time, from the police; victim rates, on data there is repeated victimization, it is from independently organized . difficult to calculate a priori the rela- tionship between offense and victimiza- of OFFENSE RATES AND VICTIMIZATION tion rates. For some types crimes, RATES the number of crime victims exceeds the number of offenses, particularly if Any simple incidence rate consists of one makes rather simple assumptions but two elements, a population that is that &dquo;collective property&dquo; ipso facto de- exposed to the occurrence of some event fines &dquo;collective victimization.&dquo; Thus, (the denominator) and a count of the if one defines all members of a house- events (the numerator). Both of these hold as victims of a burglary, a single events are measured for a given point breaking and entering of a household or period of time. An offense rate states involves all of its members as victims. the probability of occurrence of an of- Indeed, it may involve more than mem- fense for a given population while a bers of the household. A breaking and victimization rate states the probability entering, for example, that does damage of being a victim of some offense. to property, may involve a landlord as There is no simple relationship be- victim of a breaking and tenants as tween offense and victimization rates, victims of burglary. On the other hand, however. Consider an event occurring repeated victimization of a person by that is to be defined as a crime or a offenses over time and multiple offenses criminal offense. A single social en- against a victim in a single event lead to counter may involve more than one conditions where the number of offenses offense leading to multiple indictments exceeds the number of victims. of an offender or offenders in the event. While, in the aggregate of all crime This is the case, for example, when one events, it would appear that the victimi- is charged with larceny of an auto and zation rate should be higher than the larceny from an auto or when one is offense rate, assuming that the number charged with armed robbery and simple of crime victims exceeds the number of assault. A single encounter may in- offenses, it is bv no means clear what volve one or more persons as victims the magnitude of the difference is. In- or it may involve no persons as victims. deed, much depends upon how one An offense against public order or de- counts the offenses and victims in a cency may be observed only by a police situation and upon the time interval officer, while the robbery of patrons in over which one is calculating the event. an establishment may involve large The problem may not be unlike that numbers of victims. Similarly, the num- for morbidity, where, in a relatively ber of offenders may vary, and indeed short time interval, the number of visits there may be mutual victimization and to a physician exceeds the number of offending, as is the case in assaults that persons who are ill. give rise to cross-complaints. Further- for a of over more, given period time SALIENCE OF EVENTS AND which the rate is calculated, any person THEIR RECALL may be a victim of one or more crime events-one’s house may be burglarized Applying the sample-survey method on several occasions, for instance. to the realist’s objective of illuminating 12 the dark figure of crime assumes that length of time between the event and events are salient to persons as real the interview. Second, the degree of experiences and that what appear to be social threat or embarrassment is nega- socially salient events, such as crimes, tively related to rate of reporting. will be readily recalled and recounted. Third, the greater the involvement in The organized processes of the mind are institutional processing, the more likely regarded as providing more valid and it is to be recalled. Episodes that reliable information than the organized involve surgical treatment and long processes of organizations, the arma- stays are more likely to be recalled, ment of the institutionalists. for example.34 Fourth, respondents re- Yet, recent research on recall of events port their own experiences better than assumed to be salient and significant to those of others. Fifth, the more persons clearly indicates that, even in events to which one has been subject, the very short time interval, there is the more likely one is to report a selective recall of events. There is a known event. significant amount of underreporting Perhaps the crucial matter is that un- noted in studies of hospitalization and derreporting is selective among classes visits to doctors, for example.333 These of persons and events, and by time. studies and others where the sample For analysis, then, the problem of sep- survey is used to recall events that arating truth from differences in report- organizations record as having taken ing rates is confronted precisely as in place lead to several generalizations. any other organizationally processed First, underreporting increases with data. Survey interviewing, in fact, has become an institutionalized device, with 33 A study of visits to doctors for the Na- its own meanings for the population. tional Health Survey showed that 30 per cent rates of mentions of of the known visits to doctors during a two- Consequently, week period prior to the week of interviewing events can be subject to institutional were not reported in response to a standard interpretation. One such interpretation National Health Survey question; 23 per cent be the salience of a type of remained unreported after three special probe might questions had been asked. The study also experience to different classes of re- shows that underreporting was greater for less spondents. recent visits, that women reported better than The study of crime events makes men, and that persons with more serious apparent each of these conditions affect- health conditions and more visits during the recall. it is that two-week period were more likely to report. ing Indeed, likely See Charles F. Cannell and Floyd J. Fowler, institutional processing of an event is "A Study of the Reporting of Visits to an important factor in recall; yet it Doctors in the National Health Survey," clearly is not a sufficient condition, as Survey Research Center, University of Michi- events where institutional oc- gan, October 1963, p. 8. processing The study of hospitalization of persons curs-calling the police, for example- showed that hospitalized persons in the sample prove to be insufficient conditions for for themselves 7 underreported by per cent, recall. What does seem obvious is that, while the rate for both proxy adults and individuals can be to children was twice as high. The underreport- provided brought to ing rate was lowest for women reporting the report events organizations, organi- birth of a child, being but 2 per cent. See U.S., National Center for Health Statistics, 34 U.S., National Center for Health Sta- Comparison of Hospitalization Reporting in tistics, Comparison of Hospitalization Report- the Health Interview Survey, U.S. Depart- ing in Three Survey Procedures, U.S. Depart- ment of Health, Education, and Welfare Series ment of Health, Education, and Welfare Series 2, No. 6 (Washington, D.C.: U.S. Govern- 2, No. 8 (Washington, D.C.: U.S. Government ment Printing Office, July 1965), p. 8. Printing Office, July 1965), p.7 13 zational intelligence is superior to recall. be rendered comparable with that from The weight of the argument, in that survey sources. Since police data are sense, lies with the institutionalists. collected by place of occurrence rather than by place of residence of the victim, COMPARABILITY OF POLICE AND for less than national units, they must SURVEY STATISTICS be adjusted for place of residence. Furthermore, police data include of- Many of the limitations of police sta- fenses against businesses and other or- tistics, for which the survey has been ganizations ; household samples may not. claimed as a corrective, are not inherent Finally, if only the adult population is in the theoretical capabilities of law sampled and there is no reporting for enforcement as a system. Indeed, po- others in the household, offenses involv- lice agencies today collect far more in- ing persons not included in the sample formation than they process statistically must be eliminated. or publish. They collect, but rarely The fact that the two series are publish, information, for example, on not altogether comparable should make victims, multiple offenders and offenses, clear that institutionalist and realist suspects, the nature of criminal trans- perspectives are built into the data for actions, and the time and place of their reasons that derive from these very occurrence. It is primarily the failure perspectives. Consider the fact that to process information, rather than in- police statistics are for offenses by place herent limitations in collection, that of occurrence of the event. It should renders comparison between survey and be obvious that a law-enforcement sys- police data difficult. tem based on a strategy and tactics of The survey is generally designed to deployment of technology and man- gain data on victimization, while the power is interested in the location of police report data on complaints and events-events that dictate proactive observed violations, reporting them as and reactive strategies. Such an in- offenses known to the police. Even terest is not incompatible with explora- when one sets the denominator in an tion of the dark figure per se, but it is incidence rate-the exposed population incompatible with the realist ideology of -common to both, it is no simple how much crime there is. matter to render the two sets com- parable. VALIDITY OF SURVEY DATA ON CRIME To gain some comparability of vic- timization rates with police offense The crux of the traditional realist- rates, it is necessary to adjust survey versus-institutionalist controversy in- data for victimization occurring outside volves questions of validity rather than the jurisdiction sampled (a trivial prob- reliability. The cross-section sample lem for a national sample); victimiza- survey may represent an extreme pole tion of more than one person in given in the movement from &dquo;institutionalist&dquo; incidents; and &dquo;false&dquo; or &dquo;baseless&dquo; re- to &dquo;realist&dquo; approaches to crime statis- ports. Furthermore, if one is interested tics, in its complete dependence on the in comparing survey estimates of of- unsupported verbal testimony of a non- fenses with police estimates of them, official character. the survey estimates should take ac- This logical possibility should not count of whether or not the respondent obscure the fact that formal organiza- reported the event to the police. tional processing systems similarly rely At the same time, police data must primarily on unsupported oral testi- 14 mony-the complaints of citizens or ferring the motivation and competence officers as witnesses, without other evi- that make acts legally criminal. dence. most adjudicatory proc- Indeed, CONCLUSION esses, such as the pretrial hearing or the decision to prosecute, rely heavily Statistical criminology, from its out- on unsupported testimony. Nonethe- set, has searched for the key moral less, these formal systems, unlike the statistic, a measure of the &dquo;criminality&dquo; survey, rest on both the potential of present among a population. Both &dquo;in- investigation and formal sanctions to stitutionalists&dquo; and &dquo;realists&dquo; have pur- reduce fabrication. Technically, the sued this search. The foregoing discus- survey might employ many of the same sion has not made explicit our key techniques available to the police; but premise, that is, the question of whether these are alien to its basic premises, and this search has been a scientific one. the survey organization lacks formal If pragmatic objectives of criminal sta- sanctions. tistics are posed, there are no data par The survey method, rather, tries to excellence, nor is there a theory par exploit the advantage that no material excellence. consequences ensue from testimony. Although a neat polar distinction has The guarantee of anonymity, the rela- been employed that pits institutionalist tive absence of sanctions for providing against realist perspective, in practice, information, and the general absence neither camp has been comfortable in, of consequences in giving information and hence rarely consistent with, its avoid some conditions that give rise to position. The neglect of the role of nonreporting to the police and other organization in the production of knowl- formal agencies. Such an advantage is edge has led both camps astray. On of no little consequence in exploring the one hand, the realists neglect the the dark figure of crime. shaping of objective reality by what- In exploring the dark figure of crime, ever the organizational of register- the survey generally has several other ing knowledge. On the other, the in- advantages over other organizationally stitutionalists confuse the observational processed statistics. First, it provides efficacy of organizations with their a form of organization that can trans- normative functioning. Realist objec- cend local practices by providing uni- tives are best served by special organi- form operational definitions. Second, zational structures for observing and the survey taps the definitions of recording events. Institutionalist goals victims, independent of organizational would be best served by special organi- processing, and it can compare these zational structures for developing and with those of formal processing organi- scientific processing of operational or- zations. Third, the survey can identify ganizational activity. Concepts and op- and compare what is institutionally erational definitions will differ depend- labeled as crime with that consensually ing upon formally organized or informal labeled as crime. social processes, whether those of sci- Although the data cannot be adduced ence, of operations, or of social policy here, problems of evidence rather than are the primary objective. of inference probably predominate sta- In exploring the dark figure of crime, tistically in exploring the dark figure. the primary question is not how much Determining the objective character of of it becomes revealed but rather what events seems more problematic than in- will be the selective properties of any 15 particular innovation for its illumina- evaluative, institutional processing of tion. As in many other problems of people’s reports. Concepts, definitions, scientific observation, the use of ap- quantitative models, and theories must proaches and apparatuses with differ- be adjusted to the fact that the data ent properties of error has been a means are not some objectively observable uni- of approaching truer approximations of verse of &dquo;criminal acts,&dquo; but rather phenomena that are difficult to measure. those events defined, captured, and Any set of crime statistics, including processed as such by some institutional those of the survey, involves some mechanism.