Download - Measures of Effectiveness for Crime Reduction Programs

Transcript

rOPERATIONS RESEARCH, Vol. 23, No.3, May-June 1975

Measures of Effectiveness for Crime Reduction Programs

MICHAEL D. MALTZ

University of Illinois, Chicago, Illinois

(Received original June 25, 1973; final, September 23, 1974)

This paper describes some measures commonly used to evaluateanticrime programs and proposes directions for research on improvedmeasures. Since the police are usually seen as the main crimecontrol agency, the paper first discusses the differences betweenevaluating the police and evaluating crime control programs. Fivemeasures used to evaluate such programs are then analyzed: crimerate, clearance rate, arrest rate, police response time, and crimeseriousness index. The last measure suggests the direction for thedevelopment of an improved measure of crime: the developmentof a more complete taxonomy of crime and the discriminationbetween and explication of the different types of harm caused bycrime.

CRIME reduction programs have grown in number and complexity sincethe Law Enforcement Assistance Administration (LEAA) was created

in 1968. This growth has intensified the need for improved programevaluations. An important aspect of any program is the selection of ap-propriate measures of effectiveness.

This paper addresses issues concerned with the use and development ofmeasures of effectiveness for crime reduction programs. It is organized inthree sections. First, the relation between the evaluation of police and ofcrime reduction programs is discussed. The next section describes thosemeasures presently used and examines their advantages and disadvantages.The last section suggests directions for future research on output measuresfor crime reduction programs.

1. THE POLICE AND CRIME

It has long been noted that upwards of 80 percent of police activity haslittle to do with serious crime (seeNote 1), much less its reduction.[6,8,46,63jSome people, however, look upon crime reduction as the only criterionupon which to evaluate police.[2o,37jThe recently issued report of theAmerican Bar Association, Standards Relating to the Urban Police Func-tion)2j in its discussion of the evaluation of police performance, explicitly

452

Measures of Effectiveness for Crime Reduction Programs 453

avoids the traditional criteria of arrests and reduction in crime rates asoutput measures. The report suggests six measures: the extent to whichthe police

(i) safeguard freedom, preserve life and property, protect the constitutionalrights of citizens and maintain respect for the rule of law by proper enforcementthereof, and, thereby, preserve democratic processes;

(ii) develop a reputation for fairness, civility, and integrity that wins the re-spect of all citizens, including minority or disadvantaged groups;

(iii) use only that amount of force reasonably necessary in responding to anygiven situation;

(iv) conform to rules of law and administrative rules and procedures, particu-larly those which specify proper standards of behavior in dealing with citizens;

(v) resolve individual and group conflict; and

(vi) refer those in need to community resources that have the capacity to pro-vide needed assistance.

Criteria for evaluating the police in this way would include the following: Atti-tudes Toward the Police-Citizens' perceptions of the fairness, civility, and in-tegrity of the police, and the way these perceptions vary among different segmentsof the population and over time;

Police Behavior in Citizen Encounters-Victims' perceptions concerning theirtreatment by the police; arrestees' perceptions concerning their treatment; thenumber and type of citizen complaints against the police; the extent to which thepublic is made aware of them; due process in the handling of these complaints.

These criteria, however, are only supplementary in terms of crime reduc-tion. They should be looked upon as constraints on the police--a reductionin crime should be achievable without alienating major segments of thecommunity, without increasing charges of police brutality, without treat-ing the victims of crime in a perfunctory manner, without handling com-plaints against the police in secret. These are measures of police per-formance rather than crime reduction.

Yet the police are necessarily involved in any crime reduction program:They are still the ones specifically charged with reducing crime; and theycollect the information about crime (see Note 2). The wisdom of havingthe police responsible for collecting the data that are then used to judgethem has been questioned since before the Uniform Crime Reporting Pro-gram (seeNote 3) was established [31Jand is still being questioned.J-'! Al-though police-collected crime data are still the prime source of informationfor evaluating crime reduction programs, the use of population surveys todetermine crime rates has been growing in recent years.[26,33,41J

Police programs affect some crimes more than others. Intensive street

454 Michael D. Maltz

patrols, for instance, may reduce auto theft while leaving homicide or.assault untouched. Yet the police are generally assumed accountable forincreases in all types of crimes. It is questionable whether they should bethus held responsible any more than the fire department is held responsiblefor the number of fires. Although both agenciesemploy preventive strate-gies and investigate incidents after the fact, fire departments are rated forinsurance purposes primarily on their ability to minimize the damage dueto fires rather than on the number of fires. One can view the police rolesimilarly, as that of minimizing the harm caused by crime-or caused byother problems handled by the police (seeNotes 4 and 9).It is clear, then, that there is a major difference between developing

measures of police performance and developing measures to determine theeffectiveness of programs to reduce crime. The next section focusesspecifically on program measures.

2. MEASURES OF EFFECTIVENESS

This section describes problems associated with the various measures ofeffectiveness used to evaluate crime reduction programs. The ones mostcommonly used include the crime rate, clearance rate, arrest rate, policeresponse time, and crime seriousness index.

Crime Rate

The crime rate is actually the reported crime rate, the number of re-ported crimes, by type of crime (see Note 1), in a jurisdiction per year,normalized to a standard population (usually 100,000 people). [34] Thiswould seem to be the best measure to use. However, it has a number ofdefects.First, the crime rate includes only reported crimes. A crime reduction

program that reduces only the reported crime rate may appear on its faceto have done its job, but its success is illusory. For example, one policechief, disturbed by rising crime, discontinued the practice of sending patrol-men to get reports from victims of minor crimes. The victims, unless theywere insured and were required by their insurance companies to report thecrimes, did not go to the police to file reports. This reduced the reportedcrime rate. Conversely, increases in the percentage of crimes reported tothe police may make a crime reduction program (or a new police chief[12])look less effective than is actually the case.One way to surmount this difficulty is to survey the population to deter-

mine the victimization rate. [5 ,22] The LawEnforcement AssistanceAdmin-istration (LEAA) is using this method to determine the effectivenessof crimecontrol programs conducted under its Impact Cities Program. [42] Eightcities-Atlanta, Baltimore, Cleveland, Dallas, Denver, Newark, St. Louis,and Portland, Oregon-are receiving up to $20 million each over two years

f

Measures of Effectiveness for Crime Reduction Programs 455

to fund crime control programs. The National Institute of Law Enforce-ment and Criminal Justice (NILECJ), LEAA's research arm, is spendingabout $8 million over this period to evaluate the effectiveness of these pro-grams. About $5 million will be spent for victimization surveys in eachcity before and during program implementation.

This method of obtaining the crime rate is normally not used in evaluat-ing programs. Valid, reliable interviews may cost $25 to $50 or more perinterview; and because of the relative rarity of some crimes, sample sizes ofover 1000 interviews are not uncommon. Furthermore, since the surveymust be administered at least twice-before/after or experiment/control-the cost of the survey may be greater than the cost of the program.

Aside from the problems of crime reporting, there are crime recordingproblems. Police departments are often rated on their rates of "Index"crimes. [34J "Larceny $50 and over" has been the second most numerouscrime. Since "Larceny under $50" was not an Index crime, the policemay have been under pressure to undervalue the loss in larcenies and'keep the statistics down.' Because of the abuses in reporting this cate-gory, and because $50 today is worth considerably less than in 1930, theFBI has decided to change the crime category "Larcency $50 and over"to "Totallarcency" in the 1973and succeedingissues of Crime in the UnitedStates. [34J Of course, other crime categories are subject to interpretationand manipulation. [32, 64J

The police send Unified Crime Reports (UCR) data to the FBI volun-tarily. A number of states now require that all police departments sendcrime data to state crime data centers. Because they are statutorily re-quired, the statistics generated by these systems should be more accurateand reliable. Some states require external auditing of the data.

Another difficulty with the use of crime rates as output measures occurswhen they are used to determine the effectiveness of programs aimed atspecific crimes. This 'suboptimization' may be counterproductive; a pro-gram that eliminates auto theft by converting the offenders into armedrobbers should be considered a failure even if the total number of crimesdecreases slightly. This occurred to some extent in Cedar Rapids, Iowa,during a burglar alarm project.!'" This highly successful program to re-duce commercial burglaries was accompanied by a significant increase inresidential burglaries. Crime rate reductions can also signal a displace-ment of crime to other areas or to other types of crimes.P'"

The two data sources for determining crime rate, victimization surveysand police records, have different advantages and uses. Surveys of vic-timization are used only for major determinations (e.g., the national crimerate) [19J or in major programs (e.g., the LEAA Impact Cities Programr'"),and are very costly. Their cost can be reduced by adding them to a multi-purpose population survey. The advantages of a survey include a reduc-

456 Michael D. Maltz

tion in the number of unreported crimes-to the extent that victims re-member them[5,59J-and an increase in the amount of detail about a crime,especially those aspects of the harm caused by the crime that were notapparent at the time of the police report.

The crime rate determined from police reports is usually a simple countof crimes by type of crime. The data are collected in this manner by policedepartments for transmittal to the FBI under the Uniform Crime Report-ing Program. Thus, the collection of the data does not increase the re-porting burden on the police or the cost of the evaluation (especially com-pared to the cost of a victimization survey). But the rate of reporting ofcrimes to the police varies from crime to crime[26Jand from program toprogram. [38J Furthermore, the variation of crimes within crime categoriesis almost as great as the variation among categories:

Larceny is theft whether the property stolen is worth a penny, fifty dollars, or amillion. First-degree murder may mean a deliberately planned and extremelyviolent and brutal murder or a killing in self-defense by a person engaged in burg-lary. Robbery may mean the holdup of an adult by another adult at the point ofa gun or the threat of a pummeling made by one schoolboy to another one, if thelatter does not hand over his lunch money. (Reference 52, p. 42.)

Clearance Rate

The clearance rate is another frequently used output measure for crimereduction programs. It is the ratio of crimes solved (usually by the arrestof the offender) to the total number of reported crimes, by type of crime.The argument for its use is that since the police have little control over thetotal number of crimes, and since the total number is unknown, at least theclearance rate serves as a measure of how well the police operate againstthe crimes made known to them. This measure can also be misleading.

A significant change in the denominator, the number of reported crimes,without a change in the numerator, the number of cleared crimes, canchange the clearance rate considerably without any concomitant change inpolice effectiveness. The major reason for victims not reporting crimes istheir feeling that the police would be ineffective. It follows that 'the re-porting rate is higher when the victims feel the police would be effective.Thus a police department known for its ineffectivenessmight have a higherclearance rate than one known for its effectiveness, because fewer crimesconsidered hard to solve are brought to its attention.

The arrest of a person for one crime can often clear many other crimes.To clear these other crimes, which improves the police crime-fighting image,the police need the arrested person's admission of guilt. Since the policeare often more lenient if he admits to many crimes, he may exaggerate hisown lawlessness in hopes of a reduced sentence. In addition, the clearancerate does not take into consideration the number of arrests that result inprosecution, let alone conviction.

'I

Measures of Effectiveness for Crime Reduction Programs 457

If clearance rate is to be used as a measure of police proficiency, attentionshould be devoted to determining the quality of arrests made. Data canbe obtained on the number of arrests leading to indictment, to trial, and tothe various dispositions. Since the number of prosecutors and judges inthe jurisdiction, their caseloads and backlogs, and the extent of plea bar-gaining will affect the ultimate outcomes of the arrests, these factorsshould be controlled in determining arrest quality. The New York CityPolice Department recently started doing this.[lO]

Within crime categories some types are more 'clearable' than others.Stranger-to-stranger robberies are more difficult to clear up than those inwhich the offender is known to the victim; a study by the President'sCrime Commissionwl found that for all crimes the clearance rate with aknown suspect was 86 percent while the rate for crimes with no knownsuspect was 14 per cent. Yet the clearance rates are calculated withoutdistinguishing between them according to this characteristic, or accordingto other characteristics of the crime that affect clearance. However, aswith crime rates, the data can be obtained directly from police records (seeNote 10).

Arrest Rate

The arrest rate is the number of arrests, by type of crime, in a specifiedperiod, usually one year. It may be normalized even further-arrests peryear per capita, to account for population growth; or arrests per year perpolice officer,as a measure ofpoliceproductivity. This rate is distinguishedfrom the clearance rate by not being tied to the number of offenses, fre-quently because this number is not known.

For example, in the current war against narcotics there is no measure ofthe number of 'crimes' associated with breaking the narcotics laws; arrestsmay be based on quotas.t-" A violation of these laws is not a specific inci-dent, as is a robbery or a burglary, reported by the victim or a witness.In crimes such as bribery or embezzlement, discovery of the crime is tanta-mount to discovery of the offender's identity; thus, the clearance ratewould be meaningless. Increases in these arrest rates may signify increasedeffectiveness, but they may also mean increased illegal activity or increasedpolice attention to the problem.

As with clearance rates, arrests rates are poor indicators unless coupledwith some indicator of the quality of the arrest, but they can be obtaineddirectly from the police records.

Response Time

The response time is the length of time it takes a police officer to reachthe scene of an incident from the time the call is received by the police.The following times may be recorded by the police: the time of receipt ofthe call, the time the police are dispatched to the scene, the time the police

458 Michael D. Maltz

arrive at the scene, and the time the police complete handling the call. [38]The police officer radios the dispatcher to report his time of arrival (seeNote 5) and the time the call is completed. Although response time is nota direct measure of criminal activity, crime reduction, or anticrime activity,it has gained prominence as a proxy measure for police effectiveness.[4,36,54]

Using quick response to gauge police work has its basis in an analysisof police response data done for the President's Crime Commission TaskForce on Science and Technology.[45] It showed some correlation betweenfrequency of arrest and quick response, giving credence to the hypothesisthat faster response will result in more arrests. The Commission realizedthat the analysis was far from perfect and called for a controlled experimentto test the validity of this hypothesis. Since then millions of dollars havebeen spent on police radios, cars, aircraft, and computers for the purposeof reducing response time.!" ,14,60] yet the experiment called for by the Com-mission has never been performed.

Even if the relation between quick response and apprehension does hold,it does not necessarily followthat response time is a good proxy measure forpolice effectiveness. It may just show the dichotomy between two typesof policeman-the highly motivated and the poorly motivated. Theformer gets to the scene quickly and does his best to find the offender,while the latter reaches the scene later and makes a desultory search forthe offender. It is doubtful that a new computer or beat patrol algorithmwill improve the latter's motivation.

Were the proponents of this measure to promote its use for only thosecalls requiring immediate attention, response time would have greaterutility. Crime-in-progress calls and incidents involving personal injuryare examples of such calls. If the police are called fifteen minutes after thecrime is completed, a quick response will have little effect on crime.

One of the supposed advantages of using response time is that it is anunambiguous and quantitative measure, not subject to misinterpretation.This is not the case. A patrolman may not call the dispatcher upon ar-rival at the scene for a variety of reasons: the radio channels are con-gested;[45]he feels that the incident warrants his immediate response; hearrives later than he should and wants to make it appear that he forgot;or he does forget. On his way to handle a call, the police officer may getinvolved in another incident that he feels takes priority over the one as-signed him, rendering the calculation of his response time meaningless.

Using response time as a proxy measure has the same disadvantage as allproxy measures: the program focuses on achieving the proxy rather thanthe stated goal. In some ways it is equivalent to evaluating a collegepro-fessor on his ability to start his classeson time; most people 'wouldconsiderwhat he does when he starts to be more important. If program operationsconcentrate on the proxy to the exclusionof the initial objective, the proxyis no longer a good substitute for the goal of interest.

Measures of Effectiveness for Crime Reduction Programs 459

Crime Seriousness

The Crime Seriousness Index (CSI) was developed in the early 1960's bySELLINANDWOLFGANG.[52)With others, they recognized that the UCRwas an inappropriate measure of crime.[21,23.62.64)According to the UCRindex, a $5 robbery plus a $50,000 robbery equals two robberies, and one

ELEMENTS SCORED ~ X WEIGHT = TOIAL

I. NUfvBER OF VICTIMS OF BODILY HARM

A. RECEIVING MINOR INJURIESB. TREATED AND DISCHARGEDC. HOSPITALIZED AND DISCHARGEDD. KILlED

II. NUMBER OF VICTIMS OF FORCIBLE SEX INTERCOURSE

A. NUMBER OF SUCH VICTIMS INTIMIDATEDBY WEAPON

III. INTIMIDATION (EXCEPT II ABOVE)2

IV.

V.VI.

A. PHYSICAL OR VERBAL ONLYB. BY WEAPON

NUMBER OF PREMISES FORCIBLY ENTERED

NUMBER OF fIOTOR VEH I CLES STOLEN

VALUE OF PROPERTY STOLEN< DAMII.GED,OR DESTROYED (IN DOLLARS)

A. UNDER 10 DOLLARS

g; 2~i=~E. 900~3F. 3Ol)-G. OVER

ij12

TOTAL SCORE =Fig. 1. The Sellin-Wolfgangcrime seriousness index.

murder plus one larceny equals two Index crimes. The CSI is an alterna-tive scale for measuring crime. The weighting scheme is shown in Fig. 1.The weights were determined by administering questionnaires to groups

of people professionally connected with the criminal justice system (juvenilecourt judges and policemen) and to groups of nonprofessionals (collegestudents). The questionnaires were in booklet form. On each page of the

460 Michael D. Maltz

booklet a different crime was described. The respondent was asked torate each crime with respect to the standard weight given the crime de-scribed on the first page (seereference 52, p. 213). The researchers foundthat the weights were fairly consistent from group to group, irrespective ofthe group's association with criminal justice.

The word 'serious' was not defined for the raters. Thus the raters didnot have to break each crime down into its component parts and rate eachpart; they were allowed to base their weight on their overall impression ofthe incident. It is a fairly efficientway to develop a scale; the subsequentanalysis was able to relate the weight accorded each component of thecrimes. However, it tends to mask different interpretations of the word'serious.' One person might equate seriousness with the generation offear, in which case a street mugging might be close to the top of the list.Another might consider only the harm done the victim, in which case ahomicide-an intrafamily homicide or a felony homicide-might top hislist. A third might weight crimes according to the priorities he wants thepolice to use in allocating investigative resources.

The CSI has been applied in this country[30,39Jand elsewhere.J'v'v Ithas, however, become increasingly apparent that the .cSI, while a signifi-cant advance in crime measurement, has a number of shortcomings. [28 ,30,47 ,61JFor example, a study of crime and traffic accidents during an eight-weekperiod in St. Louis showed that the average 'seriousness' caused by trafficaccidents was greater than that of a crime.P" yet a crime equally as 'seri-ous' as a traffic accident would generally be considered more serious a prob-lem than the traffic accident.

The CSI treats the intimidation of victims collectively. Thus, holdingup twenty people in, say, a subway car and netting a total of $100 fromthem is considered as serious as holding up one person and relieving him of$100. Since each victim has been intimidated by the assailant(s), thereshould be some factor related to the number of victims. Similarly, moreintimidation would result from being confronted by a gang of assailantsthan by a single assailant.

Another problem relates to the basic structure of the index. In the CSIa single number represents the seriousness of an event. This number hasno intrinsic meaning; the units are units of 'seriousness' and measurepeople's perceptions of the disutility of the event. But the relative con-tributions to the total seriousness-whether due to physical injury, forciblesexual intercourse, property loss or damage, or some combination thereof-are masked in the computation of a fifteen-point robbery or an eighteen-point assault. The most recent study employing the CSI[30Jfound it desir-able to separate the total seriousness into its component parts when analyz-ing incident seriousness.

In developing the CSI, Sellin and Wolfgang rejected the legal definitions

Measures of Effectiveness for Crime Reduction Programs 461

of crime and developed a sociological classification based on the harm donethe victim. In focusing on the crime, however, the CSI neglects some of theobjective components of the crime generally used by the police, press, andpublic to determine the crime's relative importance. No distinction ismade between the physical and psychological effects of crime. The CSIdoes not distinguish between the objective components of the crime andthe subjective judgments of people about the crime and the harm. Al-though there are qualitatively different ways a person can be harmed, allare lumped together into a single dimensionless number.

Despite its shortcomings, the CSI is better than the previously discussedmeasures. It is based on a firm theoretical and empirical foundation; itcan be consistently applied by different scorers.P?' and it can be calculateddirectly from police reports. Yet the CSI has never been put into routineuse by a police department (see Note 6).

3. DIRECTIONS FOR RESEARCH ON MEASURES OF EFFECTIVENESS

This section contains a description of areas in which further research isneeded. Initially it covers some of the factors to be considered in develop-ing an improved crime classification scheme. Measuring the harm causedby crime is discussed at greater length.

The previous section examined scalar measures, either counts of events(crimes or arrests) or weighted counts of events (crime seriousness). Thissection proposes a vector measure based on different types of harm causedby crime-property loss, physical injury, and psychological injury.

The harms are discussed in the context of a particular society (the UnitedStates) at a particular time (the present). This qualification is important.The role .and definition of property, the difference between private wrongsand public wrongs, and the harm caused by injury all depend upon thesocial and legal structure. Theft of property may be meaningless in othersocieties: "Under capitalism, is crime a form of entrepreneurship; undersocialism, is entrepreneurship a form of crime?" [35]

Crime Classification

Crimes have been classified at times by the types of criminals involvedand by the types of crimes in which the criminals have been involved. Forexample, white-collar crime can be described by identifying types of crim-inals[58]or by the types of crimes. [25] These characteristics alone, however,are inadequate to distinguish among crimes for classification purposes.Few would consider the robbery of a 3D-year old man of $50 in the red-lightdistrict of a city at 2 A.M. to be more harmful than the robbery of an 80-year old woman of $10 in a residential neighborhood in the afternoon, butthis would be the case using the CSI. Crime classification using the CSIis not much different from the UCR categories. One can infer from the

462 Michael D. Maltz

CSI the categories homicide, rape, robbery, burglary, and auto theft; butthere is no more specificity than in the original legal definition.

The measures of effectiveness described in the previous section would begreatly enhanced if a more specific description of the offense were used.Such a broader taxonomy would be of great assistance in collecting andanalyzing statistics on crime. There should be distinctions made amongthe murder committed by an enraged lover, the murder committed by anervous or vicious hold-up man, and the murder committed by a gangland'enforcer.' More characteristics of the crime need to be included than arecontained in the present crime classifications or in the CSI. Some of thefactors are listed below (items with an asterisk are now part of the CSI):

1. Victim(s)-number, *age of each,sex of each.

2. Offender(s)-number,age of each (estimate),sex of each,known by victim(s)?

3. Incident-type (hold-up, break-in, etc.),weapons * or tools used,location (street, alley, park, store),time of day, week, month, year.

4. Apparent goal-money, goods, or property,harm to victim(s),sexual gratification,'kicks.'

5. Victim reaction-passivity,resistance,call for police,flight.

6. Results-property destroyed,}*property taken,extent of injury, *extent of insurance coverage.

Other factors could be added to this list, depending upon the programbeing evaluated. For example, if residential burglaries were under investi-gation it would be necessary to classify the types of housing and the methodof entry to determine their effect on the victimization rates. [7'] For policeprograms, crimes might also be categorized according to the ability of thepolice to suppress them. [48] An evaluation of the effect of street lighting oncrime would include the amount of light at the location of the crime.[65]

A crime taxonomy based on the factors described above can be used todetermine specific victimization rates for different crime types. With

Measures of Effectiveness for Crime Reduction Programs 463

additional data elements it will relate to specific programs as well. Al-though most of the factors listed above are included on police offense re-ports, the information is not used for other than investigative purposes,and rarely even for that.

Property loss

The Crir ne Seriousness Index is somewhat equivocal about the value ofproperty loss. For all thefts except automobiles the weight is approxi-

7 - ,

- .....--

..--:-

--

I I I I I

6

5

4SERIOUSNESS

3

2

1

o1 10 lrnJ io.on100

PROPERTY Loss CooLLARs)Fig. 2. Seriousness versus dollar loss for the Sellin-Wolfgang CSI.

mately proportional to the logarithm of the loss (Fig. 2) (see Note 7).For auto theft the weight is equivalent to a loss between $10 and $250.This apparent ambiguity is probably due to a number of factors:1. Many auto thefts are joyrides.2. About 74 percent of all stolen cars are recovered. [23]3. Many automobiles are insured against theft, with the insured typically

liable for the first $50 or $100.This suggests that the insurance coverage and recovery rate should be

considered in calculating the value of property loss. Since the CSI isusually calculated from data supplied in police reports, which normally donot include insurance information, this factor has not been explicitly used

464 Michael D. Maltz'

in the CSI. However, the fact that a $75,000loss is only twice as seriousas a $1000loss suggests that insurance (as well as net worth) was implicitlyconsidered by the panels whose judgments determined CSI weights. Onemight assume that a person with $75,000of stealable property is insured orcan afford the loss without undue hardship. Therefore, to find the actualvalue of the property loss, the value of the stolen or damaged goods shouldbe reduced by the amount recoverable from insurance, to obtain the netloss to the victim.

Thus far we have considered the value of property loss to be dollar value.But one man's dollar does not equal another's. One victim may be able toabsorb a $5000 (net) loss without difficulty, while a $50 loss may be aserious financial setback to another victim. Another way to calculate thevalue of property loss would be to estimate the number of hours the victimwould have to work to recoup his net loss. Although it poses some theo-retical problems, time may be a more equitable measure of the harm donethe victim as a result of property loss, for three reasons. First, it is ba-sically more democratic. The value the president of General Motors placeson his time is thus considered equal to the value the worker on the Chevroletassembly line places on his time (seeNote 8). Second, the range of valuesis more restricted. The salaries of these two men may vary by orders ofmagnitude; but their life expectancies are about the same, or at least arerelatively independent of their respective financial conditions. Third,the CSI was calculated using 1960 dollars. Using time as the measuremakes it unnecessary to account for inflation.

Physical Injury

Placing a value on human life or disability has been the subject of anumber of studies, especially in the fields of health care and trafficsafety.P' ,27 ,40 ,49] Some of the results can be applied to the assessment ofphysical injury. In doing so, however, a distinction should be made be-tween temporary and permanent injury; and time rather than money shouldbe one of the yardsticks for the measure.

The temporary disability is often expressed in monetary terms, relatingto the injured victim's loss of earnings until he is fully recovered. How-ever, this can lead to neglecting the injury suffered by housewives, retirees,and other nonwage-earners. In one victimization survey conducted by theCensus Bureau for LEAA, the questions, "Did you lose any time from workbecause of this incident? How much time?" were used to gauge the sever-ity of the injury.liS] Since housewives are not paid for their labors and"work" was interpreted to mean work for pay, their "out of commission"time was not counted.

Time appears to be the best measure of temporary injury, for the samereasons given in discussing its use in gauging property loss. The actual

Measures of Effectiveness for Crime Reduction Programs 465

amount of time needed for recovery would be one part of the total time.To this would be added the time value of the costs of treatment (hospitali-zation, treatment, physicians' fees, transportation costs, etc.), the loss ofearnings (if a wage-earner was injured), and the costs of replacement serv-ices such as child care, each to the extent they are not covered by insurance.Using time as the measure makes it relatively easy to include the replace-ment services even if, for example, the duties of an injured family memberare assumed by the rest of the family. In that case the family membersare paying with their leisure time (see Note 9).

Permanent injuries should be handled separately. Time may be onefactor in the measurement of the injury, but it should be modified by theextent to which the victim's full faculties are abridged. Insurance com-panies have attempted to develop a scale of harms in developing accidentinsurance policies, for example, by paying twice as much for the loss of twolimbs as for one, and paying the same amount for the loss of two limbs orfor total blindness as for death. Although this scale is based on loss ofearnings caused by injury rather than on the extent to which a person's lifeis affected, it can serve as a useful benchmark. The methodology used byFANSHEL AND BUSH[27] may be adaptable to this problem. They definedeleven states of health, ranging from well-being (SA) to death (SK), andweighted each state according to the utility of being in that state. Formeasuring crime it would be more useful to use states of degradation ofhealth. One can then look at the frequency distribution of victims amongthese states. Thus, a table of robbery statistics might show that 80 percentof the victims were unaffected physically (SA), 1 percent were murdered(SK), and other percentages represented the victims' states of permanentphysical injury between these two extremes.

Psychological Injury

The CSI handles psychological injury in two ways. It scores for intimi-dation of the victim and for being subjected to forcible sexual intercourse.The trauma associated with some crimes can be as debilitating as a per-manent injury and should be scored on the same terms-the extent ofabridgement of the victim's full faculties or freedom (for example, the fearof going out at night after a mugging or the inability to maintain sexualrelations after a forcible rape), and the length of time this abridgement issuffered.

The psychological effects of a crime fall not only on the victim and hisfamily, but on the community as well. This decade's preoccupation withcrime has increased the impact, with more attention being focused on crimeby the media. A rash of crimes in a previously 'safe' neighborhood cancreate a widespread fortress mentality. One common behavioral manifes-tation of this feeling is a sudden upsurge in the purchase of locks and

466 Michael D. Maltz

watchdogs, the loss of nighttime patronage of stores, and a reduction in thenumber of people outdoors at night (with the possible exception of morepeople outside walking their watchdogs).

Much of the increased attention is doubtless due to the increase in crime.This is a predictable result of the post-World War II baby boom, just as thecoming decrease in crime can be correlated with the maturation of thisgroup. But there is reason to believe that the increased attention to crimemay also be attributable to progress made on other fronts (notably medi-cine) in reducing the risk of death and injury from other causes.I'f

This psychological effect should be gauged so that we can distinguishbetween those crimes that are essentially private in nature (and thus notthreatening to the population at large) and those crimes that can affect thegeneral population. A homicide resulting from a family argument does notengender the same fear as one resulting from a store robbery. Imple-mentation of the measurement of these components of harm is discussedbelow.

Implementation

.Two of the components of the proposed vector measure of harm arereadily implemented, property loss and physical injury. The necessarydata are being collected in various victimization surveys. [22 ,26 ,33 ,41]

Property loss-Among the questions asked in the National Crime Sur-vey,[19]conducted by the Census Bureau for LEAA, are: "Altogether,what was the value of the property that was taken?" "Was all or partof the stolen money or property recovered, except for anything receivedfrom insurance?" "Was any of this loss covered through insurance?""How much would it cost to repair or replace the damaged item(s) [i.e.,damaged during the crime]?" These and other questions, combined withthe screening questions that give family income, can be used to determinethe time value of the property loss.

Physical injury-The National Crime Survey includes questions aboutthe nature and extent of the injury incurred during the crime ("What werethe injuries you suffered, if any?" "Did you receive treatment at a hos-pital? How many days?"), the cost of treatment, and the extent to whichthe victim had to pay the cost. As previously mentioned, the amount oftime lost caused by the incident is recorded only as time lost from work,so the loss of time resulting from the injuries of nonworking family mem-bers is given no value. It is hoped that this question will be changed inlater versions of the survey. No specific distinction is made betweentemporary and permanent physical injury. This distinction is anotherchange that should be considered for the National Crime Survey.

Psychological Injury-Implementing the measures of psychological in-jury is not so near at hand. Yet they are crucial parts of an index of

Measures of Effecfiveness for Crime Reducfion Programs 467

crime; it is primarily these factors that distinguish the death or injury orloss of property caused by accident from that caused by a crime. Tech-niques and scales have to be developed to determine the individual victim'spsychological injury and the psychological effect on the population atlarge.

Victimization surveys such as the National Crime Survey can be adaptedto measure these components of psychological injury. It should be pos-sible to develop a psychological 'abridgement of freedom' scale similar tothe state of health scale of Fanshel and Bush, along with an estimate of thelength of time this effect was felt.

The psychological effect on the community can be gauged by using amethodology similar to that used in developing the Crime SeriousnessIndex, but with a more representative sample of the population as re-spondents. Individuals could be asked to read descriptions of crimes andrate the perceived threat to them. Besides the variables investigated bySellin and Wolfgang, other factors should be included: the relation betweenthe 'victim' and 'offender,' the distance between the respondent's home (oroffice) and the 'crime,' the amount of newspaper coverage, the social dis-tance between the victim and respondent, and others that appear to playapart in affecting an individual's psychological response to a crime com-mitted against others.

Crime-Utility or Disutility?

The factors described above-property loss, physical injury, and psy-chological injury-are components of the harm associated with a singlecrime. They may also be components of an economist's (dis)utility func-tion but would not be the only components. Disutility is not the same asharm. One can argue that it may be better to live in a community with amoderate amount of crime than in one that is relatively crime-free. In theformer case each citizen is more likely to assume part of the burden of hisown protection and may be better prepared for the unexpected. In acrime-free community each citizen may be poorly protected and less pre-pared should an unexpected event (such as a crime) occur. Perhaps thereis an optimum level of crime-which would be different from the purelyeconomic balance point calculated by BEcKER[3]-suchthat either an in-crease or decrease from this level leads to disutility.

Economists have been taking a closer look at the whole question of thedisutility of crime.[3,29 ,50 ,56 ,57] For example, the results of an auto theft ora traffic accident may be considered a disutility, but both fuel the need forreplacement vehicles as much as a car designed to wear out after 30,000miles. Is prostitution or gambling a transfer payment or a cost to society?If a burglar is imprisoned and his family put on welfare, is there an in-creased cost to society (caused by increased welfare costs and the cost of

-

468 Michael D. Maltz

imprisonment) or a reduced cost to society (one fewer burglar on the streetswho has incidentally been the raison d'etre for theft insurance, who haskept the police occupied, and who now is helping to keep prison guardsemployed)?

These questions are not taken up in this paper. The concept of harm isoriented around the victim, his family, and other potential victims ratherthan embracing society as a whole. Although it may be considered by someto be only part of the picture, it is an important part: the harm done tovictims and to others, resulting from crime.

The Consequencesof an Improved Measure of Crime

A more complete taxonomy of crime and measure of harm caused bycrime will require data not presently collected in either victimization sur-veys or police reports. Victimization surveys can easily be adapted tocollect such data. If police reports are to be used as the source material,additional information not presently included in them needs to be obtained,e.g., insurance coverage, extent of injury.

This additional information from the police reports is fully in keepingwith the proposed reorientation of the goals of the police, from the presentprimary emphasis on the reduction of crime to an emphasis on the reductionof the impact of oictimizoiion.'» A measure of the harm caused rather thanof the number of crimes would be necessary to determine the extent towhich this goal is achieved.

Furthermore, the development of objective means of quantifying theharm will allow us to make meaningful comparisons between the harmcaused by the Index crimes and the harm caused by white-collar crimes.Since the CSI was developed for assessing crimes committed by juveniles,it did not include provision for white-color crime.

Mail fraud, one common type of white-collar crime, is often evaluated interms of the number of counts brought against the defendant, since he couldbe sentenced on each of them. However, the number of counts is notrelated to the number of victims, but rather to the number of times themail was used to further the fraud in question, be it a credit card swindle,a land fraud, or a stock manipulation. Although in many white-collarcrimes the loss is spread over many victims, the consequences can be severefor each of them. For example, in one home improvement scheme it wasfound that many victims who had previously been able to keep afloatfinancially were forced to rely on their children for financial support or goon public welfare; divorces and suicides were also attributed in part to thevictims' loss of money and self-esteem resulting from the fraud. [15 ,24] Theharm caused by these crimes should be explicitly weighed in determiningthe vigor with which they are investigated and prosecuted and in deter-mining the sentence imposed on the defendants.

Measures of Effectiveness for Crime Reduction Programs 469

Using an objective scale of harm for white-collar crime is but one furtherinstance of the usefulness of a victim-oriented measure of crime. Such ameasure should also make possible better comparisons between the harmcaused by crime and that caused by other causes, for example, punishment.This will allow us to put the whole issue of 'law and order' in its properperspective.

Crime is bad, as cancer is bad and war is bad; but even in the case of cancerone can distinguish among death, pain, anxiety, the cost of treatment, the loss ofearnings, the costs of uncertainty about life expectancy, the effects on the victimand the effects on his family. Similarly with crime. (Reference 47, page 120.)

4. CONCLUSION

This paper has analyzed the output measures currently used to deter-mine the effectiveness of crime reduction programs: crime rate, clearancerate, arrest rate, response time, and crime seriousness. The crime rate,clearance rate, and arrest rate all suffer from their being related to thelegal definition of crimes, which groups dissimilar events under the sameheading. Their chief advantages are that they are numerical measures andare easy to obtain.

One of the main virtues of response time is that it also can be measured.However, it is more a measure of police input than police output and hassome marked deficiencieswhen used as a proxy measure for crime reduction.

The crime seriousness index is a significant advance in the measurementof crime. It is based on an assumption that the true goal of crime reductionprograms is the reduction in crime seriousness rather than in the number ofcrimes. However, many factors and differences in definition are not ex-plicitly defined within the index.

We suggest a new measure of crime, along the lines of the crime serious-ness index, but based on a more complete taxonomy of crime and a fullerassessment of the harm caused by crime. Characteristics of the victim,the offender, the location, and the event would be included in the taxonomy.Four different categories of harm 'would be specified: property loss, tem-porary physical injury, permanent physical injury, and psychological in-jury. For the most part, the information already collected in victimiza-tion surveys and police reports could be used, although some additionaldata may be needed. Such a measure would improve our knowledge ofthe nature and costs of crime.

ACKNOWLEDGMENTS

This paper has benefited from the comments and suggestions of StephenM. Pollock, University of Michigan, and of the referee and editors of thisjournal.

470 Michael D. Maltz

NOTES

1. Serious crime is usually considered to be the seven felony crimes inthe FBI's crime index[34];murder and nonnegligent manslaughter, aggra-vated assault, forcible rape, robbery, burglary, larceny $50and over (before1973), or total larceny (1973 and afterwards), and auto theft. Theseseven crimes were chosen because they were felt to be the serious crimesthat are most consistently reported to the police and that occur sufficientlyoften to be useful as an index of crime.

2. The police are no longer the only ones charged with collecting crimedata. In 1971the Census Bureau, under an agreement with LEAA, begansurveying the general public concerning their victimization. This surveywill be conducted annually.P"

3. The Uniform Crime Reporting Program was developed by the Inter-national Association of Chiefs of Police. Collection of the UCR data hasbeen the responsibility of the FBI since the program's inception in 1930.[34]

4. While this paper was being reviewed, the preliminary results of astudy of police patrol effectiveness have been released.[11] The resultsindicate that preventive police patrol does not prevent crime.

5. In some departments. In most police departments there is no indi-cation of when the police officer arrives upon the scene, except when thepolice department is doing a study of response time and orders the patrol-men to report the time of arrival.

6. The Montreal, P.Q. Police Department provides for the CSI on itsoffense reports, but the police do not presently use that part of the form. [3D]

7. The relation was empirically found by Sellin and Wolfgang (reference52, p. 274) to be: Score = (Loss)o.165.

8. This is not entirely accurate, since the loss of a day's pay would affectthe assembly line worker more adversely than it would the GM president,despite the difference in dollar values. In other words, the marginal util-ity of money is greater for the assembly line worker than for the GM presi-dent. However, this estimate is more accurate than estimating the harmon the basis of dollar loss.

9. Another issue is raised by this component of harm: the relative valueof work time and leisure time. As with money, the marginal value of timevaries from individual to individual. In the case of people with 'time ontheir hands,' the marginal value of leisure time may be negative.

10. A case in point of how clearance rates are abused is found in a recentbook about the Ellsberg trial (P. SCHRAG,Test of Loyalty, New York:Simon and Schuster, 1974). The burglary of the officeof Dr. Lewis Field-ing (Dr. Ellsberg's psychiatrist) was cleared by the Beverly Hills PoliceDepartment with the arrest of "a small-time, second-story man and purse-snatcher named Elmer Davis," who, upon later investigation, turned outto have been imprisoned at the time of the burglary.

Measures of Effectiveness for Crime Reduction Programs 471

REFERENCES

1. D. D. ARMAN,A. NORMANDEAU,ANDS. TURNER, "The Measurement of De-linquency in Canada," J. Criminal Law, Criminology and Police Science58, 330-337 (1967).

2. AMERICANBARASSOCIATIONPROJECTONSTANDARDSFORCRIMINALJUSTICE,Standards Relating to the Urban Police Function, American Bar Association,Chicago, 1973.

3. G. J. BECKER,"Crime and Punishment: An Economic Approach," J. PoliticalEconomy 76,169-217 (1968).

4. W. BENNETTANDJ. R. Dulsors, The Use of Probability Theory in the Assign-ment of Police Patrol Areas, Government Printing Office, Washington, DC,1970.

5. A. D. BIDERMAN,"Surveys of Population Samples for Estimating Crime In-cidence," Annals Am. Acad. Pol. and Soc. Sci. 374, 16-33 (1967).

6. E. BITTNER,The Functions of the Police in Modern Society, Center for Studiesof Crime and Delinquency, National Institute of Mental Health, US De-partment of Health, Education and Welfare, Washington, DC, 1970.

7. A. J. BLACKBURNANDT. A. REPPETTO,"Residential Burglary as an Opera-tion," presented at the 44th National Meeting of the Operations ResearchSociety of America, San Diego, November, 1973.

8. D. J. BORDUA,The Police-Six Sociological Essays, Wiley, New York, 1967.9. W. P. BROWN,"Police Goals and Objectives: Victimization Impact Reduction,"

presented at the Conference on the Measurement of Police Performance,Academy for Contemporary Problems, Columbus, Ohio, April, 1973.

10. D. BURNHAM,"Police Agency to Monitor Courts and Prosecutors," The NewYork Times, July 9,1971.

11. ---, "A Police Study Challenges Value of Anticrime Patrol," The NewYork Times, November 11, 1973.

12. D. T. CAMPBELL,"Reforms as Experiments," American Psychologist 24., 409-429 (1969).

13. S. C. CAREY,Law and Disorder II: State Planning and Programming UnderTitle I of the Omnibus Crime Control and Safe Streets Act of 1968, NationalUrban Coalition, Washington, DC, 1970.

14. ---, Law and Disorder III: State and Federal Performance Under Title Iof the Omnibus Crime Control and Safe Streets Act of 1968. Lawyers Com-mittee for Civil Rights Under Law, Washington, DC, 1972.

15. J. CARPER,Not with a Gun, Grossman, New York, 1972.16. J. D. CASPER,American Criminal Justice: The Defendant's Perspective, Pren-

tice-Hall, Englewood Cliffs, New Jersey, 1972.17. CEDARRAPIDS,IOWA,POLICEDEPARTMENT,Installation, Test, and Evaluation

of a Large-Scale Burglar Alarm System for a Municipal Police Department,National Institute of Law Enforcement and Criminal Justice, Law Enforce-ment Assistance Administration, US Department of Justice, Washington,DC, 1971.

18. CENSUSBUREAU,Personal Crime Incident Report, Survey of Victims of Crime,Form QHS-771, Bureau of the Census, US Department of Commerce,Washington, DC, 1971.

472 Michael D. Maltz

19. ---, National Crime Survey: Basic Screen Questionnaire and Crime IncidentReport, Forms NCS-l through NCS-4, Bureau of the Census, US Depart-ment of Commerce, Washington, DC, 1974.

20. P. CHEILIKANDW. J. MERRILL, "Indices for Measurement of Law Enforce-ment Agency Performance," presented at the Annual meeting of theAmerican Association for the Advancement of Science, December, 1971.

2l. D. R. CRESSEY,"The State of Criminal Statistics," National Probation andParole Assoc. 3, 230-241 (1967).

22. R. W. DODGEANDA. G. TURNER,"Methodological Foundations for Establish-ing a National Survey of Victimization," presented at the meeting of theAmerican Statistical Association, Fort Collins, Colorado, August, 1971.

23. E. M. DOLESCHAL,Criminal Statistics, Center for Studies in Crime and De-linquency, National Institute of Mental Health, US Department of Health,Education and Welfare, Washington, DC, 1972.

24. R. DORFMAN,(ED.), Measuring Benefits oj Government Investments, BrookingsInstitution, Washington, DC, 1965.

25. H. EDELHERTZ,The Nature, Impact and Prosecution oj White-Collar Crime,National Institute of Law Enforcement and Criminal Justice, Law Enforce-ment Assistance Administration, US Department of Justice, Washington,DC, 1970.

26. P. H. ENNIS, Criminal Victimization in the United States: A Report oj a Na-tional Survey, Field Surveys II, President's Commission on Law Enforce-ment and Administration of Justice, Government Printing Office, Washing-ton, DC, 1967.

27. S. FANSHELANDJ. W. BUSH, "A Health-Status Index and Its Application toHealth-Services Outcomes," Opns. Res. 18,1021-1066 (1970).

28. R. J. GIBBS,Crime Seriousness: A Review oj the Literature in Relation to Possi-ble Police Use, Police Scientific Development Branch, Home Office, London,England, 1970.

29. J. R. HARRIS, "On the Economics of Law and Order," J. Political Economy78,165-174 (1970).

30. N. B. HELLERANDJ. T. McEWEN, The Use of an Incident Seriousness Indexin the Deployment oj Police Patrol Manpower, Board of Police Commis-sioners, St. Louis, Missouri, Police Department, 1972.

3l. ILLINOISASSOCIATIONFORCRIMINALJUSTICE,Illinois Crime Survey, Summaryand Recommendations, (1929). Reprinted as Appendix B of J. LANDESCO,Organized Crime in Chicago,University of Chicago Press, Chicago, 1968.

32. INSTITUTEOF PUBLICADMINISTRATION,"Crime Records in Police Manage-ment," in M. E. WOLFGANG,L. SAVITZ,ANDN. JOHNSON,The Sociology ojCrime and Delinquency, Wiley, New York, 1970.

33. C. B. KALISH, Crimes and Victims: A Report on the Dayton-San Jose PilotSurvey oj Victimization, National Criminal Justice Information and Sta-tistics Service, Law Enforcement Assistance Administration, US Depart-ment of Justice, Washington, DC, June, 1974.

34. C. M. KELLEY,Crime in the United States-1972, Government Printing Office,Washington, DC, 1973.

Measures of Effectiveness for Crime Reduction Programs 473

35. M. C. KENNEDY, "Beyond Incrimination," Catalyst (1970); Reprint 212,Warner Modular Publications, Andover, Massachusetts, 1973.

36. R. C. LARSON,Urban Police Patrol Analysis, MIT Press, Cambridge, Massa-chusetts, 1972.

37. R. C. LINDANDJ. P. LIPSKY,"The Measurement of Police Output," Law andContemporary Problems 36, 566-588 (1971).

38. M. D. MALTZ,Evaluation of Crime Control Programs, National Institute ofLaw Enforcement and Criminal Justice, Law Enforcement Assistance Ad-ministration, US Department of Justice, Washington, DC, 1972.

39. W. J. MERRILL,M. N. MILKS,ANDM. SENDROW,Case Screening and SelectedCase Processing in Prosecutors' Offices, National Institute of Law Enforce-ment and Criminal Justice, Law Enforcement Assistance Administration,US Department of Justice, Washington, DC, 1973.

40. E. J. MISHAN, "Evaluation of Life and Limb: A Theoretical Approach," J.Political Economy 79, 687-705 (1971).

41. NATIONALCRIMINALJUSTICEINFORMATIONANDSTATISTICSSERVICE, Crimein the Nation's Five Largest Cities: National Crime Panel Surveys of Chicago,Detroit, Los Angeles, New York, and Philadelphia, Law Enforcement Assist-ance Administration, US Department of Justice, Washington, DC, April,1974.

42. NATIONALINSTITUTEOFLAWENFORCEMENTANDCRIMINALJUSTICE,PlanningGuidelines and Programs to Reduce Crime, National Institute of Law En-forcement and Criminal Justice, Law Enforcement Assistance Administra-tion, US Department of Justice, Washington, DC, 1972.

43. D. J. NEWMAN,Conviction: The Determination of Guilt or Innocence withoutTrial, Little, Brown, Boston, 1966.

44. A. NORMANDEAU,"The Measurement of Delinquency in Montreal," J.Criminal Law, Criminology and Police Science 57,172-177 (1966).

45. PRESIDENT'SCOMMISSIONON LAW ENFORCEMENTAND ADMINISTRATIONOFJUSTICE,Task Force Report: Science and Technology, Government PrintingOffice, Washington, DC, 1967.

46. A. J. REISS, JR., The Police and the Public, Yale University Press, New Haven,1971.

47. G. N. G. ROSE, "Concerning the Measurement of Delinquency," Brit. J.Criminology 6, 414-421 (1966).

48. ST. LOUISPOLICEDEPARTMENT,Allocation of Patrol Manpower Resources inthe St. Louis Police Department, St. Louis, Missouri, Police Department,V.1, p. 81, 1966.

49. T. C. SCHELLING,"The Life You Save May Be Your Own," in S. B. CHASE,JR. (Ed.), Problems in Public Expenditure Analysis, Brookings Institution,Washington, DC, 1966.

50. ---, "The Economics of Organized Crime," in President's Commission onLaw Enforcement and Administration of Justice, Task Force Report: Orga-nized Crime, Government Printing Office, Washington, DC, 1967.

51. D. SEIDMANANDM. COUZENS,"Getting the Crime Rate Down: PoliticalPressure and Crime Reporting," Law and Society 8, 457-493 (1974).

474 Michael D. Maltz

52. T. SELLINANDM. E. WOLFGANG,The Measurement of Delinquency, Wiley,New York, 1964.

53. J. H. SKOLNICK,Justice Without Trial, Wiley, New York, 1966.54. SMALLBUSINESSADMINISTRATION,Crime Against Small Business, Govern-

ment Printing Office, Washington, DC, 1969.55. C. STARR,"Social Benefit versus Technological Risk," Science 165, 1232-1238

(1969) .56. G. J. STIGLER,"The Optimum Enforcement of Laws," J. Political Economy

78, 526-536 (1970).57. R. F. SULLIVAN,"The Economics of Crime: An Introduction to the Litera-

ture," Crime and Delinquency 19,138-149 (1973).58. E. H. SUTHERLAND,White Collar Crime, Holt, Rinehart and Winston, New

York, 1961.59. A. G. TURNER,San Jose Methods Test of Known Crime Victims, National In-

stitute of Law Enforcement and Criminal Justice, Law Enforcement As-sistance Administration, US Department of Justice, Washington, DC, June,1972.

60. US HOUSE OF REPRESENTATIVES,Committee on Government Operations,Legal and Monetary Affairs Subcommittee, The Block Grant Programs of theLaw Enforcement Assistance Administration, House Report No. 92-1072,Government Printing Office, Washington, DC, 1972.

61. UNIVERSITYOF LANCASTER,Department of Operational Research, PatrolEffectiveness and Patrol Deployment, University of Lancaster, Lancaster,England, 1968.

62. S. WHEELER, "Criminal Statistics: A Reformulation of the Problem," J.Criminal Law, Criminology and Police Sci. 58, 317-321 (1967).

63. J. Q. WILSON,Varieties of Police Behavior, Harvard University Press, Cam-bridge, Massachusetts, 1968.

64. M. E. WOLFGANG,"Uniform Crime Reports: A Critical Appraisal," U. Pa.Law Review Ill, 708-738 (1963).

65. R. WRIGHT,D. THOMAS,P. PELLETIER,ANDK. DICKINSON,Study to Deter-mine the Impact of Street Lighting on Crime, Institute for Survey Research,University of Michigan, Ann Arbor, Michigan, 1972.