Levitt Sample.qxd - SAGE Publications Ltd

4CONCEPTUALIZATION AND

MEASUREMENT

89

In Chapter 1, it was stated that conceptsare human-created ideas that representreality. Concepts allow researchers to

gain a better understanding of the social worldwe live in. For example, intelligence is aconcept—a human-created idea to measurehow smart an individual is. Researchers inter-ested in intelligence and its relationship toother variables (crime, for example) could notstudy varying levels of intelligence in individ-uals and whether it influences behaviors with-out first conceptualizing what intelligence isand constructing methods for measuring it.Intelligence has been measured in the past byan intelligence quotient (IQ), in order that personscan be compared on this concept. Incidentally,IQ has been used both as a dependent variable(which individual or societal factors influencesomeone’s IQ) and an independent variable(IQ’s influence of other behaviors such ascrime). Psychologists continue to debatewhether current methods for measuring IQ areaccurate and whether IQ should be measureduniformly across all groups and cultures insociety (Weinberg, 1989). This debate notwith-standing, intelligence is a concept that has beendefined and measured to make it possible forresearchers to study it.

Conceptualization is a very important stepin the research process. Researchers cannotstudy particular phenomena unless they firstdefine what it is exactly that they are studying.Conceptualization involves identifying impor-tant attributes and describing and definingthem to make them amenable for study.Blalock (1982) states that “conceptualizationinvolves a series of processes by which theo-retical constructs, ideas, and concepts are clar-ified, distinguished, and given definitions”(p. 11). A primary reason why researchers oftencome to contradictory conclusions in studiesof the same topic is that individual researchershave conceptualized the same phenomenon inslightly different ways. Crime, for instance, isa very broad concept, and depending on how aresearcher conceptualizes crime, differentvariables may be shown to be influential in theoccurrence of it.

For example, if a researcher wants to studythe influence of gender on crime, the types ofcriminal behaviors included in the study willgreatly influence whether gender has aneffect on crime. If the researcher only studiesviolent crime, he or she may find a large gen-der effect (males have a greater likelihood tocommit violent crime); however, if the same

researcher includes only property and publicorder crimes in his or her conceptualizationof crime, a gender effect may be less pro-nounced. Beyond conceptualization, a researchermust also construct ways to operationalizeconcepts. The way in which concepts aremeasured may also influence the signifi-cance of the relationship between indepen-dent and dependent variables.

CONCEPTUAL VERSUS

OPERATIONAL DEFINITIONS

Researchers almost always report the ways inwhich their independent and dependent vari-ables were conceptualized and operationalizedin their research. This helps the consumersof this research understand exactly what theresearchers studied and may also help interpretsome of the findings from the study.Researchers usually distinguish between con-ceptual and operational definitions of concepts.A conceptual definition might be found in a dic-tionary, whereas an operational definition ismore empirical and outlines how a concept willbe measured or observed for study (Ellis,Hartley, & Walsh, 2010).

For example, say a researcher believesthat religiosity is related to crime; thehypothesis is that those who are more religiousare less likely to be criminal. Religiosity is acommon concept; less obvious is how reli-giosity can be measured in an individual.Perhaps church attendance would be a goodmeasure of religiosity. The researcher maywant to collect information on how manytimes per year or month an individual attendschurch services. Maybe, however, there aresome persons who are very religious but donot attend church. The same researcher mayalso want to ask individuals about theirbelief in a higher power, or whether or notthey pray. These preceding measures mightall be ways to operationalize religiosity sothat the absence or presence or variability of

religiosity can be observed for selected indi-viduals in order to assess religiosity’s rela-tionship to criminal offending.

Issues in conceptualization and operational-ization in criminology and criminal justiceresearch are ever present as there is alwaysdebate about the most appropriate ways tomeasure certain phenomena and becauseresearchers across the globe adopt differentmethods for measuring the same concepts ofinterest. From a research standpoint, conceptu-alization and operationalization are importantsteps in transforming abstract concepts aboutthe social world into more concrete, observ-able ideas. There is also a great deal of debateabout how accurate and exact measures ofcrime and justice variables should be.

VALIDITY, RELIABILITY,AND PRECISE MEASUREMENT

Although variables can be measured at differ-ent levels (Stevens, 1946; see also Bachmanand Schutt, 2008 for a more in-depth discus-sion of levels of measurement), accuracy, con-sistency, and precision in measurement arealso important steps in the research process. Inorder to advance knowledge in the discipline,researchers have to make their best attempts toensure that their measures are accurate. Inother words, are researchers observing whatthey intended to observe? Accurate observa-tions are critical to getting accurate results andmaking conclusions that reflect the true realityof phenomena. Some concepts are easier toaccurately measure than others; a respondent’sage, sex, and education level can be measuredmore accurately than their attitudes or opin-ions of a particular subject. Measurement erroris one of the most persistent problems threat-ening social and behavioral science researchers’ability to get accurate results. Accuracy inmeasurement is comprised of three elements:validity, reliability, and precision (Ellis et al.,2010).

90 • SNAPSHOTS OF RESEARCH

Validity

Validity refers to the extent to whichresearchers measure what they planned tomeasure. In other words, are the variablesused in the study good indicators of what theresearchers are trying to study? There are dif-ferent ways to assess the validity of measures.Some simply involve researchers asking them-selves if their measures seem like logical andcommon sense ways to measure concepts(face validity). Others seek outside criteria toassess the validity of measures; researchersasking a respondent to self-report drug usecould validate the response by also asking theparticipant to provide a urine sample (criterionvalidity). Further still, researchers may want toensure that they are examining the full rangeor meaning of the concept (content validity) orthat their measures fit with philosophical ortheoretical ideas of the concept they areattempting to measure (construct validity).

Even though researchers may use all of thepreceding ways in assessing the validity oftheir measures, problems with validity maystill persist. Constructing valid measures are acontinual problem in criminology and criminaljustice because researchers in these disciplinesstudy very complex and difficult concepts (Elliset al., 2010). Hagan (2005) notes that “validityis never entirely demonstrated or proven;rather, invalidity is lessened, or researchers areable to express greater degrees of confidencein their data” (p. 280).

Reliability

Reliability, on the other hand, has to dowith the consistency and stability of measures.Will the measures, if repeated using the samesubjects, yield the same results? For example,a scale, if calibrated correctly, is a very reli-able measure of weight—much more reliablethan having someone guess a person’s weightbased on his or her height and body type. Theradar gun that police officers use to measurethe speed of passing cars is another example of

a reliable way to measure a concept—in thiscase, the velocity of automobiles. Just as withvalidity, there are some common ways forresearchers to assess the reliability of theirmeasures. They could administer the same sur-vey to the same respondents twice to assessconsistency (test-retest methods), or have twodifferent persons gather observations andassess the consistency between the two (inter-rater reliability). Often, researchers assess reli-ability of their measures using what is knownas the split-half technique; the instrument orquestionnaire is divided into two halves, bothof which measure the same concept; however,the scores from the two halves are examinedseparately and compared to assess the reliabil-ity of the measures.

Although both validity and reliability areconcerned with measurement accuracy, theyare not the same thing. You probably knowpeople who, to make themselves feel better,set their scale back 5 pounds. Each day theyweigh themselves, provided their weight doesnot change much from day to day, it displaysthe same weight. Thus, this scale is reliablymeasuring their weight (consistent from day today) but it is not valid because they set it sothey weigh 5 pounds less than their trueweight, and thus the scale is not accurate.Researchers would be wise to assess the valid-ity and reliability of their measures beforegathering data or making observations withthe sample they intend to study. There are noremedies for invalid or unreliable measures ifthey are found after the data has been col-lected. Most researchers conduct pilot tests(tests on a small group of people) to ensurevalidity and reliability of their measures.

Precise Measurement

Accuracy of measurement also refers to thefact that researchers can make observationsor measure variables with varying degrees ofpreciseness. For example, using a scale is avery precise way to measure weight; having

Chapter 4 Conceptualization and Measurement • 91

students guess weight, however, is a very crudeway to measure weight. As a general rule, pre-cise measurements are superior to impreciseones; however, a high degree of precision maynot be necessary for every project, nor desir-able for that matter. Ultimately, it will dependon the purposes of the research. For example,measuring someone’s annual income to theexact dollar, or age to the exact year, is moreprecise than using categories of income or age(for example, <$10,000, $10,001–$29,999,$30,000–$49,999, $50,000+; <19, 20–29,30–39, 40–49, 50+). Similarly, measuring theimplementation of a new policy at its exactdate is more precise than measuring it by thedecade it was implemented (summer of 1996versus 1990s). However, these more precisemeasures may not always be necessary.

Obviously, operationalization of conceptswill depend a great deal on the amount of pre-cision that is required in the study. If knowingthat someone’s income is in between$30,000–$49,000 or that someone is in their40s satisfies the purposes of a specificresearch study, then the additional effort tomore precisely measure concepts may not beworth it; in other words, this may be timewasted for both the researcher and the respon-dent. Further, some respondents may notknow their exact annual income because itvaries from year to year. Researchers alsowant to avoid asking questions that requirerespondents to spend a great deal of time try-ing to recall answers. Measures that elicit eas-ily recalled responses are ultimately likely tobe more accurate.

92 • SNAPSHOTS OF RESEARCH

Source: Gibbs, J. P. (1989). Conceptualization of terrorism. American Sociological Review, 54(3), 329–340.

RESEARCH READING

Conceptualization is an important step in the research process. If researchers want to exam-ine social phenomena, they must first define what it is exactly that they intend to study aswell as what the parameters of their research subject are. In other words, researchers needto construct both conceptual and operational definitions of their topics in order to ensure thatthey are observing and measuring exactly what they wanted to. The article by Gibbs makesit clear that the things researchers are interested in studying are often difficult to conceptu-alize and measure. Although terrorism has probably been discussed and redefined numer-ous times since this article was first published, it is nonetheless a good example of whyconceptualization is a very important step in the research process.

Gibbs here reveals that some phenomena are not always as easy to define and measure asthey may seem. Issues with conceptualization and operationalization are ever present insocial science research and are often the reason for conflicting findings in research of thesame topic. In considering terrorism, Gibbs provides a good example of some of the impli-cations of definitions that are too broad or too narrow. As you will see, a definition of ter-rorism can be quite complex, and complexity is not necessarily what researchers wantwhen studying a topic; however, simplifying terrorism can make researching it problem-atic as well.

Reading 6 Conceptualization of Terrorism • 93

INTRODUCTION

Definitions of terrorism are controversial forreasons other than conceptual issues and prob-lems. Because labeling actions as “terrorism”promotes condemnation of the actors, a defini-tion may reflect ideological or political bias(for lengthy elaboration, see Rubenstein 1987).Given such considerations, all of which dis-courage attempts to define terrorism, it is notsurprising that Laqueur (1977, p. 5) argued that

a comprehensive definition of terrorism does notexist, nor will it be found in the foreseeablefuture. To argue that terrorism cannot be studiedwithout such a definition is manifestly absurd.

Even granting what Laqueur implies—that terrorism is somehow out there awaitingdefinition—it is no less “manifestly absurd” topretend to study terrorism without at least somekind of definition of it; leaving the definitionimplicit is the road to obscurantism.

Even if sociologists should overcome theirostensible reluctance to study terrorism (for arare exception, see Lee 1983), they are unlikelyto contribute to its conceptualization. The situ-ation has been described succinctly by Tallman(1984, p. 1121): “Efforts to explicate key con-cepts in sociology have been met with stiflingindifference by members of our discipline.”

There are at least two reasons why sociolo-gists commonly appear indifferent to conceptu-alizations. First, Weber and Parsons gave thework a bad name in the eyes of those sociolo-gists who insist (rightly) on a distinctionbetween substantive theory and conceptualanalysis. Second, conclusive resolutions ofconceptual issues are improbable because theultimate justification of any definition is animpressive theory that incorporates the defini-tion. Nonetheless, it is crippling to assume thatproductive research and impressive theories arepossible without confronting conceptual issuesand problems. The argument is not just thattheorizing without a definition is sterile, normerely recognition that theory construction andconceptualization should go hand in hand.Additionally, one can assess definitions with-out descending to purely personal opinion,even when not guided by a theory.

Systematic tests of a theory require defini-tions of at least some of the theory’s constituentterms; but test findings, even those based onthe same units of comparison, will diverge ifeach definition’s empirical applicability is neg-ligible, meaning if independent observers dis-agree when applying the definition to identifyevents or things. To illustrate, contemplate aquestion about any definition of terrorism:How much do independent observers agree injudging whether or not President Kennedy’s

CONCEPTUALIZATION OF TERRORISM

Jack P. Gibbs

Abstract:Many issues and problems surround the conceptualization of terrorism. Most definitions of theterm are indefensible if only because they do not speak to those issues and problems. An assessment ofcontending definitions can transcend purely personal opinions; and an assessment can be undertakenwithout a theory, even though an impressive theory is the ultimate justification of its constituent defini-tions. The present conceptualization goes beyond a definition of terrorism by emphasizing the defini-tion’s bearing on five major conceptual questions, each of which introduces a major issue and/orproblem. Then it is argued that thinking of terrorism and other sociological phenomena in terms of con-trol promotes recognition of logical connections and/or empirical associations, each of which couldbecome a component of a theory.

assassination was terrorism in light of the def-inition? As subsequent illustrations show, sim-ple definitions may promote agreement inanswers to the Kennedy question and yet beobjectionable for theoretical reasons; but theimmediate point is that an empirically applic-able definition does not require a theory. Bycontrast, given evidence that a definitionpromises negligible empirical applicability, notheory can justify that definition.

Still another “atheoretical” criterion is thedefinition’s consistency with convention. Thatcriterion cannot be decisive, because it wouldpreclude novel definition, but it is importantwhen the field’s professionals must rely on“outsiders” for data and, hence, presumeappreciable congruence between their defini-tions and those of the outsiders. That consider-ation is particularly relevant here, because inanalyzing terrorism social scientists often relyon reports of government officials, journalists,and historians.

Conceptual issues and problems haunt vir-tually all major terms in the social and behav-ioral sciences, and any definition is ambiguousif it does not answer questions bearing onthose issues and problems. There are at leastfive such questions about terrorism. First, isterrorism necessarily illegal (a crime)?Second, is terrorism necessarily undertaken torealize some particular type of goal and, if so,what is it? Third, how does terrorism neces-sarily differ from conventional military opera-tions in a war, a civil war, or so-calledguerrilla warfare? Fourth, is it necessarily thecase that only opponents of the governmentengage in terrorism? Fifth, is terrorism neces-sarily a distinctive strategy in the use of vio-lence and, if so, what is that strategy?

The questions are answered in light of a sub-sequent definition of terrorism, but more than adefinition is needed. The pursuit of a theoryabout terrorism will be furthered by describingand thinking about terrorism and all other soci-ological phenomena in terms of one particularnotion, thereby promoting the recognition oflogical and empirical associations. The most

appropriate notion is identified subsequentlyas “control,” but a defense of that identifica-tion requires a definition of terrorism (not of“terror”).

A DEFINITION OF TERRORISM

Terrorism is illegal violence or threatened vio-lence directed against human or nonhumanobjects, provided that it:

(1) was undertaken or ordered with a view toaltering or maintaining at least one putativenorm in at least one particular territorialunit or population;

(2) had secretive, furtive, and/or clandestinefeatures that were expected by the partici-pants to conceal their personal identityand/or their future location;

(3) was not undertaken or ordered to further thepermanent defense of some area;

(4) was not conventional warfare and because oftheir concealed personal identity, conceal-ment of their future location, their threats,and/or their spatial mobility, the participantsperceived themselves as less vulnerable toconventional military action; and

(5) was perceived by the participants as con-tributing to the normative goal previouslydescribed (supra) by inculcating fear of vio-lence in persons (perhaps an indefinite cat-egory of them) other than the immediatetarget of the actual or threatened violenceand/or by publicizing some cause.

Clarification, Issues, and Problems

In keeping with a social science tradition,most definitions of terrorism are set forth in afairly brief sentence (see, e.g., surveys by Oots1986, pp. 5-8, and Schmid and Jongman I986,pp. 32-38). Such definitions do not tax thereader’s intellect or patience, but it is inconsis-tent to grant that human behavior is complexand then demand simple definitions of behav-ioral types.

94 • CHAPTER 4 CONCEPTUALIZATION AND MEASUREMENT

The Illegality of Terrorism. Rubenstein’s defi-nition (1987 p. 31) is noteworthy if onlybecause it makes no reference to crime orillegality: “I use the term ‘terrorism’ . . . todenote acts of small-group violence for whicharguable claims of mass representation can bemade.” However, even granting that terrorismis an illegal action, there are two contendingconceptions of crime, one emphasizing thereactions of officials as the criterion and theother emphasizing normative considerations(e.g., statutory law). Because of space limita-tions, it is not feasible to go much beyond rec-ognizing the two contending conceptions. Itmust suffice to point out that an action may beillegal or criminal (in light of statutes and/orreactions by state officials) because of (1)where it was planned; (2) where it com-menced; and/or (3) where it continued, espe-cially in connection with crossing a politicalboundary. Such distinctions are relevant evenwhen contemplating the incidence of terrorism.

One likely reaction: But why is terrorismnecessarily a crime? The question suggeststhat classes of events or things exist indepen-dently of definitions. Thus, it may appear that“stones” and “humans” denote on to logicallygiven classes, but in the context of gravita-tional theory stones and humans are not dif-ferent. However, to insist that all definitionsare nominal is not to imply that conventionalusage should be ignored; and, again, the pointtakes on special significance when definingterrorism. The initial (unnumbered) part of thepresent definition is consistent, with mostother definitions and also with this claim:most journalists, officials, and historians wholabel an action as “terrorism” evidently regardthe action as illegal or criminal. However, it isnot denied that two populations may differsharply as to whether or not a particular actionwas a crime. As a necessary condition for anaction to be terrorism, only the statutes and/orreactions of officials in the political unitwhere the action was planned or took place (inwhole or in part) need identify the action ascriminal or illegal.

Violence and Terrorism. Something like thephrase “Violence or threatened violence”appears in most definitions of terrorism (seeSchmid and Jongman 1988, p. 5). As in thosedefinitions, the phrase’s key terms are here leftas primitives; and whether they must bedefined to realize sufficient empirical applica-bility can be determined only by actualattempts to apply the definition.

Despite consensus about violence as a nec-essary feature of terrorism, there is a relatedissue. Writers often suggest that only humanscan be targets of violence, but many journal-ists, officials, and historians have identifiedinstances of destruction or damage of nonhu-man objects (e.g., buildings, domesticated ani-mals, crops) as terrorism. Moreover, terroristspursue their ultimate goal through inculcationof fear and humans do fear damage or destruc-tion of particular nonhuman objects.

The Ultimate Goal of Terrorists. The presentdefinition indicates that terrorists necessarilyhave a goal. Even though it is difficult to thinkof a human action that is not goal oriented, theconsideration is controversial for two reasons.One reason is the allegation that terrorists areirrational or mentally ill (see, e.g., Livingston1978, pp. 224-39; and Livingstone’s commen-tary, 1982, p. 31 on Parry), which raisesdoubts as to whether terrorists have identifi-able goals. The second reason why part 1 ofthe definition is controversial: many sociolo-gists, especially Durkheimians, do not empha-size the purposive quality of human behavior,perhaps because they view the emphasis asreductionism. In any case, a defensible defini-tion of virtually any term in sociology’svocabulary requires recognition of the rele-vance of internal behavior (e.g., perception,beliefs, purpose). Thus, without part 1 of thepresent definition, the distinction between ter-rorism and the typical robbery becomesobscure. The typical robber does not threatenviolence to maintain or alter a putative norm;he or she is concerned only with behavioralcontrol in a particular situation.


A defensible definition of a norm is not pre-sumed (see Gibbs 1981, pp. 9-18, for a litanyof difficulties). Rather, it is necessary only thatat least one of the participants (those whoundertake the violent action or order it) viewthe action as contributing to the maintenanceor alteration of some law, policy, arrangement,practice, institution, or shared belief.

Part l of the definition is unconventional onlyin that goals of terrorists are not necessarilypolitical. Many definitions create the impressionthat all terrorism is political (for a contrary view,see Wilkinson 1986, p. 51), but the very term“political terrorism” suggests at least two types.The concern of social scientists with terrorismtypologies is premature (see, e.g., the commen-tary by Oots [1986, pp. 11, 30] on Mickolus’snotions of international, transnational, domes-tic, and interstate terrorism). No terrorism typol-ogy amounts to a generic definition (see thesurvey in Schmid and Jongman 1988, pp. 39-59),and without the latter the former is bound to beunsatisfactory.

Military Operations and Terrorism. To repeat aprevious question; How does terrorism neces-sarily differ, if at all, from conventional militaryoperations in a war, civil war, or so-called guer-rilla warfare? The question cannot be answeredreadily because there are no clearly accepteddefinitions of conventional military operation,war, civil war, and guerrilla warfare. “Guerrilla”is especially troublesome because journalists areprone to use the word without defining it butsuch as to suggest that it is synonymous with ter-rorism (a usage emphatically rejected byLaqueur 1987 and Wilkinson 1986).

Conventional military operations differfrom terrorism along the lines indicated byparts 2, 3, and 4 of the definition. However,the definition does not preclude the possibilityof a transition from terrorism to civil war. Onetragic instance was the Easter Rising in Ireland(1916), when rather than perpetuate the terror-ism tradition, a small group of Irish seized andattempted a permanent defense of governmentbuildings in Dublin, vainly hoping that thepopulace would join them in open warfare.

Today, it is terrorism rather than civil war thathaunts Northern Ireland, and the term, “guer-rilla warfare” has no descriptive utility in thatcontext.

Terrorism as a Special Strategy. One featureof terrorism makes it a distinctive (though notunique) strategy in violence. That feature isdescribed in part 5 of the definition.

Part 5 is controversial primarily because itwould exclude actions such as this threat:“Senator, if you vote for that bill, it will beyour death warrant.” Why would such a threatnot be terrorism? A more theoretically signifi-cant answer is given subsequently. Here itmust suffice to point out that scores of writershave emphasized “third party” or “general”intimidation as an essential feature of terror-ism; and journalists, officials, or historiansonly rarely identify “dyadic intimidation” (Xacts violently toward Y but not to control Z’sbehavior) as terrorism.

“State Terrorism” as a Special Issue. Zinam’sdefinition (1978, pp. 244-45) illustrates one ofmany reasons why definitions of terrorism areso disputable: “[Terrorism is] the use or threatof violence by individuals or organized groupsto evoke fear and submission to obtain someeconomic, political, sociopsychological, ideo-logical, or other objective, “ Because the defi-nition would extend to the imposition of legalpunishments by government officials to pre-vent crimes through general deterrence, invirtually all jurisdictions (see Morris 1966,p. 631), some aspects of criminal justice wouldqualify as terrorism; and Zinam’s definitionprovides no basis for denying that it would be“state terrorism.” Even granting that a stateagent or employee acts for the state only whenacting at the direction or with the consent of asuperordinate, there is still no ostensible dif-ference between the use or threat of violencein law enforcement and Zinam’s terrorism.

Had Zinam defined terrorism as beingnecessarily illegal or criminal, then manyinstances of violence by a state agent oremployee at the direction or with the consent of


a superordinate would not be terrorism.However, think of the numerous killings inNazi Germany (Erust Roehm, the Storm Troophead being a well-known victim) during theNight of the Long Knives (June 30, 1934). Hitlerordered the slaughter, and at the time thekillings were illegal in light of German statutes;but Hitler publicly acknowledged responsibil-ity, and the only concealment was that per-ceived as necessary to surprise the victims.Surely there is a significant difference betweensuch open, blatant use of coercion by a stateofficial (dictator or not) and the situation whereregime opponents are assassinated but officialsdisavow responsibility and the murders are sosecretive that official complicity is difficult toprove. The “rule of terror” of Shaka, thefamous Zulu chief, is also relevant. Shaka fre-quently ordered the execution of tribalmembers on a seemingly whimsical basis, butthe orders were glaringly public (see Walter1969). Shaka’s regime illustrates another point:in some social units there may be no obvious“law” other than the will of a despot, in whichcase there is no basis to describe the despot’sviolence as illegal. The general point: becausevarious aspects of government may be publicviolence, to label all of these aspects “terror-ism” is to deny that terrorism has any secretive,furtive, or clandestine features.

Given the conceptual issues and problemsthat haunt the notion of state terrorism, it ishardly surprising that some writers attributegreat significance to the notion, while others(e.g., Laqueur 1987, pp. 145-46) seem to rejectit. The notion is not rejected here, and the fol-lowing definition does not make it anextremely rare phenomenon. State terrorismoccurs when and only when a government offi-cial (or agent or employee) engages in terror-ism, as previously defined, at the direction orwith the consent of a superordinate, but onewho does not publicly acknowledge suchdirection or consent.

The foregoing notwithstanding, for theoret-ical reasons it may prove desirable to limit theproposed definition of terrorism (supra) tononstate terrorism and to seek a quite different

definition of state terrorism. Even so, it willnot do to presume that all violence by stateagents is terrorism. The immediate reason isthat the presumption blurs the distinctionbetween terrorism and various kinds or aspectsof law enforcement. Moreover, it is grosslyunrealistic to assume that all instances ofgenocide or persecution along racial, ethnic,religious, or class lines by state agents (includ-ing the military) are terrorism regardless of themeans, goals, or circumstances. Nor is itdefensible to speak of particular regimes (e.g.,Stalin’s, Hitler’s, Pol Pot’s) as though all ofthe related violence must have been state ter-rorism. For that matter, granted that theregimes were monstrous bloodbaths, it doesnot follow that the state agents in questionmade no effort whatever to conceal any oftheir activities and/or their identity. Readerswho reject the argument should confer withAmerican journalists who attempted to coverStalin’s Soviet Union, Hitler’s Germany, orPol Pot’s Cambodia. Similarly, it is pointlessto deny that secretive, clandestine, or furtiveactions have been characteristic of “deathsquads” (many allegedly “state”) in numerousLatin American countries over recent decades.It is commonly very difficult to prove thatsuch groups murder with the knowledgeand/or consent of state officials; but the diffi-culty is one justification for identifying themurders as terrorism, even though the state-nonstate distinction may be debatable in par-ticular instances.

Difficulties in Empirical Application

One likely objection to the present defini-tion of terrorism is its complexity; but, again,demands for simplicity are inconsistent withhuman behavior’s complexity. Nonetheless,application of the definition does call for kindsof information that may not be readily avail-able. Reconsider a previous question: WasPresident Kennedy’s assassination terrorism?The present definition does not permit anunequivocal answer, largely because there aredoubts about the goals of the assassination and


whether or not it was intimidation. If terrorismwere defined as simply “the illegal use orthreat of violence,” an affirmative answer tothe Kennedy question could be given; but thedefinition would also admit (inter alia) all rob-beries and many child abuses. Similarly, thephrase “for political purposes” would justifyan affirmative answer to the Kennedy ques-tion; but the implication would be a tacit denialof apolitical terrorism, and divergent interpre-tations of “political” are legion. Finally,although a definition that specifically includes“murder of a state official” would maximizeconfidence in an affirmative answer to theKennedy question, there must be doubts aboutthe feasibility of such an “enumerative” defin-ition of terrorism. And what would one makeof the murder of a sheriff by his or her spouse?

The general point is that a simple definitionof terrorism tends to delimit a class of eventsso broad as to defy valid generalizations aboutit (reconsider mixing presidential assassina-tions, robberies, and child abuses) or so vaguethat its empirical applicability is negligible. Inthe latter connection, the Kennedy illustrationindicates the need to grant this methodologicalprinciple: the congruence dimension (but notthe feasibility dimension) of a definition’sempirical applicability is enhanced when inde-pendent observers agree that the definitioncannot be applied in a particular instancebecause requisite information is not available.If that principle is not granted, sociologistswill try to make do with simple definitions andwhatever data are readily available.

Presumptive and Possible Terrorism.Comparativeresearch on terrorism commonly is based onthe use of the term “terrorism” by journalists orofficials. Hence, insofar as the use of data onpresumptive terrorism can be justified, a defin-ition’s utility is enhanced by its correspon-dence with the use of the term “terrorism” byjournalists and officials. Although only poten-tially demonstrable, my claim is that the pre-sent definition corresponds more with such useof the term than does any simpler definition,such as: terrorism is illegal violence.

Even when terrorism research is based ondescriptions of violent events, as in newspaperstories, there may be cases that can be desig-nated as possible terrorism even though theinformation is not complete; and a definition’sempirical applicability can be assessed in termsof agreement among independent observers insuch designations. In that connection, the pre-sent definition points to the kind of informationneeded for truly defensible research on terror-ism, which is not the case when investigatorstry to make do with a much simpler definition,or no definition at all.

TOWARD A THEORY OF TERRORISM

The presort definition of terrorism does notanswer any of a multitude of questions, suchas: Why does the incidence of terrorism varyamong political units and over time? Althoughit is an illusion to suppose that any definitionanswers empirical questions, a definition maybe much more conducive than are alternativesto thinking about phenomena; if so, the defin-ition furthers the pursuit of a theory.

SUMMARY AND CONCLUSION

An impressive theory of terrorism requiresmore than a conceptualization that confrontsissues and problems. A definition of terrorismmust promise empirical applicability and facil-itate recognition of logical connections andpossible empirical associations. Such recogni-tion requires a notion that facilitates describ-ing and thinking about terrorism; and thenotion must be compatible with each of threepossible explanatory mechanisms: strict cau-sation, selective survival, and purposiveness.

The notion of control is the most promisingcandidate. Although that notion has no equalwhen it comes to underscoring human behav-ior’s purposive quality, it is not alien to anyparticular explanatory mechanism. All ofsociology’s subject matter can be described


and thought of in terms of control (at least asit has been conceptualized here), and thenotion is particularly relevant in the study ofterrorism. That phenomenon and attempts toprevent it are nothing less than one vastattempt at control.

REFERENCES

Becker, Jillian. 1988. Terrorism in West Germany.London: Institute for the Study of Terrorism.

Catton, William R. 1966. From Animistic toNaturalistic Sociology.NewYork:McGraw-Hill.

Clark, Robert P. 1986. “Patterns in the Lives ofETA members.” Pp. 283-309 in PoliticalViolence and Terror, edited by Peter H. Merkl.Berkeley: University of California Press.

Davis, Kingsley. 1959. “The Myth of FunctionalAnalysts as a Special Method in Sociology andAnthropology.” American SociologicalReview 24:757-72.

Durkheim, Emile. 1949. The Division of Labor inSociety. New York: Free Press.

Gibbs, Jack P. 1981. Norms, Deviance, and SocialControl. New York: Elsevier.

Harris, Marvin. 1979. Cultural Materialism.New York: Random House.

Laqueur, Walter. 1977. Terrorism. London:Weidenfeld and Nicolson.

________. 1987. The Age of Terrorism. London:Weidenfeld and Nicolson.

Lee, Alfred M. 1983. Terrorism in NorthernIreland. Bayside, NY: General Hall.

Livingston, Maritis H., ed. 1978. InternationalTerrorism in the Contemporary World.Westport, CT: Greenwood.

Livingstone, Neil C. 1982. The War AgainstTerrorism. Lexington, MA: Heath.

Morris, Norval. 1966. “Impediments of PenalReform.” University of Chicago Law Review,33:627-56.

Noakes, Jeremy. 1986. “The Origins, Structureand Function of Nazi Terror.” Pp. 67-87 inTerrorism, Ideology, and Revolution, editedby Noel O’Sullivan. Brighton, England:Harvester.

Oots, Kent I. 1986. A Political OrganizationApproach to Transnational Terrorism.Westport. CT: Greenwood.

Parsons, Talcott. 1951. The Social System. NewYork: Free Press.

Rubenstein, Richard E. 1987. Alchemists ofRevolution. London: L.B. Tauris.

Schmid, Alex P. and Albert I. Jongman. 1988.Political Terrorism. Rev. ed. Amsterdam:North-Holland.

Skocpol, Theda. 1979. States and SocialRevolution. London: Cambridge UniversityPress.

Taillman, Irving. 1984. Book Review. SocialForces 62:1121-22.

Walter, Eugene V. 1969. Terror and Resistance.New York: Oxford University Press.

Weber, Max. 1978. Economy and Society. 2 vols.Berkeley: University of California Press.

Wilkinson, Paul. 1986. Terrorism and the LiberalState. 2nd ed. New York: New YorkUniversity Press.

Zinam, Oleg. 1978. “Terrorism and Violence inLight of a Theory of Discontent andFrustration.” Pp. 240-63 in InternationalTerrorism in the Contemporary World, editedby Marks H. Livingston. Westport, CT:Greenwood.


DISCUSSION QUESTIONS

1. Why does Laqueur argue that a definition of terrorism does not exist? Why does Gibbsargue that social scientists are often indifferent to conceptualization?

2. One of the first issues Gibbs discusses is the illegality of terrorism. What are some ofthe issues that Gibbs raises regarding defining terrorism as a crime?

(Continued)

RESEARCH READING

This research article examines whether the ways in which questions about sexual and physicalvictimization are asked to inmates affects the reporting of these acts. Wolff and colleaguesargue that not much is known regarding the prevalence of violence inside correctional settingsbecause researchers have not been consistent with their definitions of victimization. Their pur-pose is to explore the difference between using questions that are very broad or general versusthose that are related to specific acts. The results of this study reveal that depending on howquestions are worded, the prevalence rates of sexual and physical violence vary. Their researchdemonstrates the importance of asking the right questions and in the correct manner in orderthat researchers obtain accurate measures. Accurate measures of the amount of victimizationthat occurs inside prison are important to developing programs to reduce and control it.

MEASURING VICTIMIZATION INSIDE PRISONS

Questioning the Questions

Nancy Wolff, Jing Shi, and Ronet Bachman

Abstract: Violence and victimization inside the prison setting are accepted as facts, although the factsabout their prevalence remain uncertain. Variation in the methods used to estimate rates of sexual and


Authors’ Note: This study was supported by the Office of Justice Programs (Grant #OJP-2004-RP-BX-0012) and the NationalInstitute of Mental Health (Grant #P20 MH66170). The authors acknowledge with gratitude the generous and helpful commentsthey received from Dr. Richard Tewksbury in the development of the manuscript.

Source: Wolff, H., Jing, S., & Bachman, R. (2008). Measuring victimization inside prisons: Questioning the questions. Journalof Interpersonal Violence, 23(10), 1343–1362. Copyright 2008 Sage Publications.

(Continued)

3. Under certain definitions of terrorism, why might the military operations of somecountries be considered as terrorism, in essence making their governments terrorists?

4. According to this article, do terrorists have goals? What makes terrorists any differentthan other criminals (such as murderers)?

5. What are some of the difficulties with empirical application of terrorism? In otherwords, why is it difficult to measure?

6. What does this author conclude about whether the assassination of President Kennedywas an act of terrorism?

7. What does Gibbs conclude regarding what is necessary to the formulation of a theoryof terrorism?

Reading 7 Measuring Victimization Inside Prisons • 101

INTRODUCTION

Some people housed in America’s prisons arevictimized both sexually and physically byother inmates and staff (Bowker, 1980; Gaes &Goldberg, 2004; Lockwood, 1980; Toch,1985). This statement once considered a mythis now readily accepted as fact. What remainsless evidence based, however, concerns howmany people inside prison are victimized inparticular ways and by whom. These issueshave been the source of considerable investiga-tion over the past two decades, especially withrespect to sexual victimization. Representativeprevalence rates remain elusive, with rangesvarying from less than 1% to 40% for sexualvictimization (Gaes & Goldberg, 2004) and10% to 25% for physical victimization(Wooldredge, 1998). Extant studies are charac-terized by probability and convenience samplesof inmates with (uncontrolled) nonresponserates of 50% or higher drawn from a smallnumber of prison facilities, by survey questionsthat require varying degrees of subjective inter-pretation, by varying recall or reflection peri-ods, and by survey administration strategiesthat vary in their ability to minimize stigma andshame, as well as danger to the inmate.

In an effort to focus the measurement ofvictimization inside prisons, this study con-trols for certain types of methodological bias(sample size, time period, and survey adminis-tration) to investigate the framing of questionsthat are used to elicit information about vic-timization inside prison. Although framing ofquestions is a central focus of survey and vio-lence research, it has not received carefulinvestigation in the measurement of sexual or

physical victimization inside prisons. The pur-pose of this study is to explore the use of gen-eral questions about victimization framed interms of “assault” (a broadly and commonlyused term in the victimization literature; cf.Kilpatrick, 2000) versus specific questionsabout particular acts or behaviors on the rate ofphysical and sexual victimization inside pris-ons during a 6-month time period.

LITERATURE REVIEW

One of the basic components of surveyresearch is the phrasing of questions to measurewhat is intended (Presser et al., 2004).Questions cannot be reasonably phrased, how-ever, until there is clarity about what is to bemeasured. In the prison victimization literature,what is to be measured is interpersonal vio-lence resulting from one person (the perpetra-tor or aggressor) harming another person (thevictim) in particular ways: physically, sexually,psychologically, or emotionally. Violencemight be manifested as an actual (i.e., hitting,coercing, or forcing sexual contact), attempted(i.e., striking out but missing the person,attempting forced sexual contact), or a threat-ened act that is harmful. The harmful act mighthave occurred at any point, because the personwas under the authority of the correctionalentity; the act occurred at one or more prisonsoperated by that entity; or the act occurredonce or multiple times and been perpetratedby inmates, staff, or both. Furthermore, theharmful act might have been reported toauthorities and have led to further or moresevere mistreatment because it was reported;

physical victimization contribute to the wide range in estimates appearing in the prison literature. Thisarticle focuses on the questions used in the prison victimization literature to elicit information on vic-timization from inmates, compared to questions used in the general victimization literature. The ques-tions used in the National Violence Against Women and Men Surveys are used to estimate sexual andphysical victimization rates for an entire prison system. Rates of victimization were found to vary sig-nificantly by specificity of the question, definition of perpetrator, and clustering of behaviors. Facts aboutvictimization inside prison will become more certain when the methodology becomes more standardizedand consistent with definitions of victimization.

likewise, it might have resulted in medicaland/or mental health treatment.

Given the multidimensionality of violence,being clear about what is to be measured meanslimiting the definition first in terms of (a) whetherthe goal is to measure harmful acts that arecommitted, attempted, or threatened; and (b) thetypes of acts that are considered harmful.Sexual violence, as defined by the NationalCenter for Injury Prevention and Control(NCIPC), is divided into nonconsensual sexualacts, consisting of forced or threatened sex acts,including vaginal, oral, and anal sex; and abu-sive sexual contacts, including intentionaltouching of specified areas of the body (Basile& Saltzman, 2002). Nonconsensual sexual actsare defined to be consistent with notions ofrape, attempted rape, or sexual assault. Physicalabuse is typically defined as a threat or anattempt to do bodily harm and may or may notinvolve a weapon. As such, definitions of sex-ual and physical violence share three common-alities in that they do the following: (a) produceinterpersonal harm (which may be manifestedphysically, emotionally, and/or psychologi-cally); (b) may be completed, attempted, orthreatened; and (c) include specific types ofbehavior. These commonalities define thedomains of what is to be measured by questionsin surveys about victimization inside prisons.These domains can be described in terms ofplace, time, and perpetrator.

Victimization questions used in the prisonvictimization literature have been idiosyn-cratic and, as such, have generated rates of vic-timization that cannot be meaningfullycompared across studies based on correctionalpopulations or the general population. Moremeaningful and useful prevalence rates wouldbe available if prison-based studies used mod-ified questions based on the NVAW survey.Modification is needed because the weapons(and language for these weapons) and the typeof perpetrators are different in prison settings,although the harmful behaviors or acts are thesame as are their types of commission (i.e.,

completed, attempted, or threatened). Thestudy described next used both general andspecific questions to measure sexual andphysical victimization inside male and femaleprisons located in a single mid-Atlantic state.The specific questions were modified from theNVAW. The questions were framed to mea-sure bid time and 6-month prevalence rates ofboth sexual and physical victimization withina state prison system.

METHOD

Sampling

A sample of 7,443 inmates was drawnfrom a single state (N = 22,231), inclusive of12 adult male facilities and 1 adult femalefacility. Inmates were excluded from the sam-ple if they were younger than 18 or in admin-istrative segregation custody, detention, deathrow, a sex offender treatment facility, or oth-erwise too sick to participate in the survey.Also excluded were inmates residing inhalfway houses or off-site at the time of thesurvey. In total, 85% (n = 18,956) of theinmate population was eligible. Data were col-lected from June through August 2005.Inmates housed in the general population (n =18,956) were invited by researchers to partici-pate in the survey. Enough time inside thefacility was requested to collect a 40% proba-bility sample from the general population, typ-ically requiring between 2 and 5 days. A 66%random sample of inmates was selected from apopulation frame in advance with the expecta-tion of a 60% response rate among those ran-domly selected, yielding the expected 40%sample target within the allotted time at thefacility. Target samples across facilitiesranged from 26% to 53%, with a mean targetsample of 40% (SD = 0.061). Data collectionat the facility with a target sample of 26% wasprematurely terminated because of a lockdownsituation (unrelated to the study) at the facility.


Procedures

The survey was administered using anaudio-computer-administrated survey instru-ment (audio-CASI) available in English andSpanish. Respondents sat at a laptop computerin a private location and were read the ques-tions via headset and responded via mouse.Thirty computer stations were available andresearchers were available for assistance.Completing the English version of the instru-ment by audio-CASI took approximately 60minutes, whereas the Spanish version tookapproximately 90 minutes. Participants wereprovided the option of a Spanish version(68 respondents [0.9%] completed the Spanishversion). The recruitment protocol, the con-sent procedures, and the consent form wereapproved by the appropriate university andDepartment of Corrections (DOC) institu-tional review boards. Participants wereinformed about the survey through inmate liai-son representatives and an informational videoabout the survey (played on the inmate televi-sion channel). DOC staff was not involved inthe recruitment or marketing of the survey.Participants were not compensated for theirparticipation to reduce any overt coercion.

Participants

Of the 18,956 inmates from general popula-tion eligible to participate, approximately13,000 were briefed on the survey and 7,443(57% response rate) were recruited and com-pleted the survey. A total of 6,879 men (M age= 33.9, SD = 6.5) and 564 women aged 18 orolder participated in the study (M age = 35.5,SD = 6.4). Over two thirds (68.4%) of thefemale inmate respondents were non-Whiteand 83.1% of the male respondents were non-White. These demographic statistics are equiv-alent to the general prison population (67.3%of females are non-White, with a mean age of35.4 years; 79.8% of the males are non-White,with a mean age of 34.4). The survey sample

overrepresented the Hispanic (inclusive ofWhite and non-White; 14.5% female; 19.9%male) population within the prison (10.1%female; 15.0% male).

Instruments

The questions regarding sexual victimiza-tion were adapted from the NVAW survey(Tjaden & Thoennes, 2000). Violence wasmeasured using four general questions foreach type of perpetrator (inmate or staffmember). The general questions were as fol-lows: Have you been sexually assaulted (phys-ically assaulted) by (inmate/staff member)within the past 6 months? Have you ever beensexually assaulted (physically assaulted) by(inmate/staff member) on this bid? In contrastto community-based studies of victimization,a 6-month reflection period, instead of a 12-month period, was selected in the phrasing ofquestions for three reasons. First, one of ourstudy objectives was to measure interfacilityvariation in victimization rates. Given the ten-dency of inmates to move between and amongfacilities, a 6-month reflection period providedgreater stability within the denominator of theprevalence rates for each facility. Second, theliterature suggests that inmates are at greatestrisk of sexual victimization in the first 6 monthsat a facility. Limiting the reflection period to 6months allowed us to test (analysis to be con-ducted in the future) whether, indeed, victim-ization rates are highest when inmates firstarrive at a facility. Third, given the frequencyof violence expected inside prison, especiallyin terms of physical victimization, more reli-able reporting is expected from shorter reflec-tion periods. Longer reflection periods, say 12months, are most appropriate for low probabil-ity events. Prevalence rates for 6-month peri-ods, in combination with the bid timeprevalence rates, can be linearized to approxi-mate 12-month rates under alternative growthrate assumptions. Only findings based on the6-month data are presented herein.


Ten additional questions about specifictypes of sexual victimization were used (e.g.,“During the past 6 months, has (anotherinmate or staff member) ever . . . touched you,felt you or grabbed you in a way that you feltwas sexually threatening or made you havesex by using force or threatening to harm youor someone close to you?”). The specific sex-ual assault questions were clustered to reflectdefinitions of sexual violence developed bythe NCIPC.

Behavior-specific questions about physicalvictimization were framed in terms of whetheranother inmate (or staff member) has done anyof the five specific behaviors “during the past6 months.” The five specific questions relating tophysical victimization were collapsed into twocategories differentiated by whether a weaponwas involved (i.e., with or without weapon).The specific questions were collapsed into cat-egories of physical victimization involving andnot involving a weapon.

Statistical Analysis

Unweighted data are presented herebecause the emphasis is on question phrasing.It is noteworthy, however, that nearly identi-cal results are generated with weighted data.Unless otherwise indicated, the significancelevel used to assess the validity of the nullhypotheses is p < .05. Prevalence of sexualand physical victimization measures thenumber of people in the population experienc-ing sexual and physical victimization within a6-month period.

RESULTS

Rates of inmate victimization varied signifi-cantly by phrasing of the victimization ques-tion. As shown in Table 4.1, specificquestions, phrased in terms of specific types ofbehavior, produced rates of victimization thatwere 1.1 to 9 times larger than rates based on

the general question, phrased in terms of sex-ual or physical assault for both inmate andstaff perpetrators. Rates of inmate-on-inmatevictimization reported by women based on thespecific questions dramatically increased rela-tive to the general question and relative to theirmale counterparts. Less dramatic yet still sig-nificant (except for female inmate rates bystaff) increases occurred in the estimated ratesfor physical victimization within gender oracross gender groups.

The rates of sexual victimization, accordingto Table 4.1, could be reported as low as 1.7%of male inmates to as high as 10.7%, depend-ing on whether respondents were asked ifthey had been sexually assaulted in the past6 months by another inmate or whether they hadexperienced various types of sexually explicit(and abusive) behavior by another inmate or astaff person during that same time period. Therange of rates for female inmates was evenlarger, ranging from 2.5% to 26.4%. If experi-ences of physical and sexual victimization arecombined and inclusive of both inmate andstaff perpetrators, roughly 40% of female andmale inmates experienced some form of vic-timization within a 6-month period in this stateprison system.

Using questions that focus on inmate-on-inmate rape understates the rape experience ofmale (1.6% vs. 3.0%) and female (3.7% vs.5.0%) inmates. Similarly, using questionsphrased in terms of rape or sexual assault, asmeasured by nonconsensual sexual acts,understates exposure to sexual victimizationwithin a prison setting (males: 3.0% vs. 9.4%;females: 5.0% vs. 25.0%). Likewise, impor-tant gender and perpetrator patterns aremasked when physical victimization questionsdo not distinguish the use of weapons in thecommission of physically assaultive acts.

Table 4.2 explores the response patterns tothe general and specific questions. Nearly allof those who reported “no” to all the specificquestions about sexual victimization alsoreported “no” to the general questions. Yet ofthose who said “yes” to at least one of the


105

Tab

le4.

1Six-M

onthPrevalenceRates(Per1,000)forVictimizationby

Gender,TypeofPerpetrator,and

Phrasingofthe

SurveyQuestion

Phr

asingof

the

Survey

Que

stion

MaleInmates

(N=

6,89

7)Fem

aleInmates

(N=

564)

Inmate

95%

Con

fide

nce

Interval

Staff

95%

Con

fide

nce

Interval

Eithe

ror Both

95%

Con

fide

nce

Interval

Inmate

95%

Con

fide

nce

Interval

Staff

95%

Con

fide

nce

Interval

Eithe

ror Both

95%

Con

fide

nce

Interval

Sexualvictimizationby

type

ofperpetrator

General

1714

to20

2723

to31

3732

to41

2512

to39

2713

to40

4528

to62

Specific

3935

to44

7367

to79

9992

to106

226

192to261

8461

to107

264

227to300

Com

bined

4440

to49

7973

to86

107

99to114

230

195to265

8662

to109

264

227to300

Nonconsensual

1613

to19

2118

to25

3026

to34

3722

to53

208to31

5032

to68

Abusive,

sexual

3632

to41

6963

to75

9487

to101

216

182to252

7452

to96

250

214to285

Physicalvictimizationby

type

ofperpetrator

General

7367

to79

164

155to172

198

188to207

102

77to127

5939

to78

138

109to166

Specific

194

185to203

220

210to230

326

315to337

203

170to237

6544

to85

228

193to263

Com

bined

207

198to217

261

251to272

360

315to371

224

190to259

9167

to115

260

223to296

Without

weapon

119

111to126

148

140to157

218

208to228

164

133to195

5233

to70

187

155to220

Withweapon

145

136to153

158

149to167

251

241to261

100

75to125

2512

to38

114

88to140

Com

binedsexualandphysicalvictimizationby

type

ofperpetrator

General

8074

to87

173

164to182

210

200to220

117

914to144

7150

to92

158

128to188

Specific

208

198to217

249

239to259

357

345to368

336

297to375

125

98to152

383

342to423

Com

bined

223

213to233

289

278to299

390

378to401

350

310to389

142

113to171

400

359to440

specific questions, roughly 7 in 10 reported“no” to the general question, except for theinmate-on-inmate responses by the femalerespondents in which 9 in 10 reported “no” tothe general question while responding “yes” toat least one of the specific questions.

The general physical victimization, espe-cially for staff-on-inmate form of victimiza-tion, yields responses more consistent with thespecific questions (yes-yes combinations). Yetthe slightly lower proportion reporting “no” tothe specific question and “yes” to the generalquestion regarding physical victimization(ranging from 1.7% to 5.3%, compared to the

0.2% to 0.6% for the sexual victimizationquestions) suggests that types of physicallyabusive behavior for the specific behaviorquestion are not fully representative of thebehavior that respondents consider physicallyassaultive.

DISCUSSION

The results of this research illustrate theimportance of question wording whenattempting to measure sexual and physical vic-timization. More specific behavior questions


Table 4.2 Percentage of Inmates Responding to Specific and General Phrasing ofVictimization Question by Gender and Type of Perpetrator

Response to SpecificQuestions

Male Inmates Female Inmates

General, % Yes General, % No General, % Yes General, % No

Sexual Victimization, Inmate-on-Inmate by Response to General Question

Specific, Yes 30.0 70.0 9.7 90.3

Specific, No 0.6 99.4 0.5 99.5

Sexual Victimization, Staff-on-Inmate by Response to General Question

Specific, Yes 28.9 71.1 31.1 68.9

Specific, No 0.6 99.4 0.2 99.8

Physical Victimization, Inmate-on-Inmate by Response to General Question

Specific, Yes 30.6 69.4 39.5 60.5

Specific, No 1.7 98.3 2.7 97.3

Physical Victimization, Staff-on-Inmate by Response to General Question

Specific, Yes 55.2 44.8 50.0 50.0

Specific, No 5.3 94.7 2.7 97.3

elicit information on violent or abusivebehaviors that are threatened or attempted;behaviors, like actual completed behaviors,create a sense of psychological and emotionaldistress and fear within the victim and theenvironment. In general, specific behaviorquestions increase the percentage of peoplereporting sexual and physical victimizationinside prisons within a specific referenceperiod. Questions focusing on specific typesof behavior also provide more detail aboutactual victimization, allowing it to be delin-eated into types, such as nonconsensual sex-ual acts or abusive sexual conduct (Fisher,Cullen, & Turner, 2000). More specificity onthe types of sexual victimization is criticallyimportant for prevention strategies becauseabusive sexual conduct is often considered aprecursor to nonconsensual sexual acts(Lockwood, 1980). Defining the source ofvictimization to include both inmates andstaff also increases the prevalence of victim-ization, in some cases doubling the estimatedrate. The effect of including staff as perpetra-tors appears to depend on the type of victim-ization and gender.

Although it is clear that the form of thequestion and its scope affects estimates ofvictimization within a prison, it also affectsthe ability to compare violence in prisons tothat in the general population residing in thecommunity. Using the same questions, theNVAW survey found that 0.2% of womenreported being raped (attempted or com-pleted) during a 12-month period (Tjaden &Thoennes, 2000), compared to 4.6% ofwomen and 2.7% of men during a 6-monthperiod in prison. (Note: In the NVAW sur-vey, fewer than 5 male victims in a sample of8,000 reported being raped and, as such, noestimates were calculated.) Rates of physicalassault for a 12-month period were 1.3% forfemales and 0.9% for males in the generalpopulation, compared to 22.8% of femalesand 32.6% of males in prison, again during a6-month period.

Comparing victimization rates inside prisonto those in the general population is problem-atic, because people inside prison are not rep-resentative of the general population. Given thestrong correlation between poverty and crimi-nalization (Jargowsky, 1997; Wilson, 1987)and between poverty and victimization(Rennison & Rand, 2003), a more appropriatecomparison group may be inner-city areas—areas more representative of the places wherepeople residing in prison once lived and wherethey will eventually return after completingtheir sentences (Travis, 2005). Teplin,McClelland, Abram, and Weiner (2005) usedthe NCVS to estimate 12-month prevalencerates of crime victimization for “central cities.”For central cities over a 12-month period, ratesof rape or sexual assault were estimated at0.23% for women, 0.025% for men. Rates ofphysical assault for men and women were esti-mated at, respectively, 1.6% and 1.1%, in innercities based on the NCVS questions on sexualand physical assault. Although the screeningquestions for the NCVS are less specific thanthose used in the NVAW, they still generaterates, adjusted to central city areas, which aresubstantially smaller than rates of victimizationestimated for male and female prisons over atime period half in length.

One of the central advantages of usingvictimization questions that are used innational surveys is that it allows meaningfulcomparisons across groups in different set-tings and geographical areas. As such, itbecomes clear that prisons are more violentplaces than the general population and cen-tral city environments. It remains unclearwhether this is because the prison environ-ment congregates people prone toward vio-lence, because the secluded nature andpower structure of the prison environmentfacilitates violence, or because the deprivationwithin the environment foments the stressesthat cause violence. Also, it is unclear towhat extent group-to-group violence moti-vates or directs interpersonal violence. These


data simply suggest that prison environmentsare more violent and that the people residingthere live with the consequences and fear ofvictimization during their incarceration, andthe data also suggest areas where moreresearch is needed.

Putting prison victimization methods on parwith those in the general victimization literatureis long overdue. What is being measured—victimization—is independent of setting andneeds only to be translated to a prison context interms of type of perpetrator and names ofweapons (e.g., “shanks”). Standardization ofmethods will provide more useful informationfor identifying the sources and context of vic-timization, which can then be used to developpolicies to prevent violence inside prisons andto treat and protect those who have been vic-timized. The Commission on Safety and Abusein America’s Prisons (2006) is calling foractions that will make correctional settingssafer. Collecting reliable and valid data on vic-timization inside prisons is a necessary first steptoward this goal in terms of identifying howmuch victimization is occurring and what typesof violence occur and by whom, as well as pro-viding guidance on how best to prevent vio-lence and how much treatment is needed torespond to the trauma and other consequencesassociated with violence.

It is important to keep in mind that ourfindings are based on a single state prisonsystem. Whether these rates generalize toother prisons systems is an empirical issuethat requires future research. Yet for this sin-gle state system, our findings indicate thatpeople inside prisons are at greater risk ofsexual and physical victimization than theaverage citizen in the community, even cen-tral cities characterized by poverty. For thisreason, prisons can be considered a naturallaboratory for systematically studying vio-lence in terms of its causes, its consequences,and ways to prevent and treat it. In particular,researchers studying violence in prison havean opportunity to develop and inform the

literature on the measurement of victimizationand its prevention. Interpersonal violence,whether sexual or physical, has many dimen-sions and manifestations, with actual,attempted, and threatened violence havingconsequences that affect physical health (suchas HIV infection) as well as psychologicaland emotional well-being. The challenge is tomeasure the dimensionality of different formsof violence, to explore the interdependencebetween and among these dimensions, and todevelop interventions to prevent its onset andto treat its occurrence in ways that addressesthe ensuing layers of harm. Yet in advancingthe literature, it is critical that prison-basedresearch incorporates standard measures ofvictimization and methods of collection toensure valid estimates and comparisonsacross settings.

REFERENCES

Bachman, R. (2000). A comparison of annual inci-dence rates and contextual characteristics ofintimate-partner violence against womenfrom the National Crime VictimizationSurvey (NVAWS). Violence Against Women,6, 839-867.

Basile, K. D., & Saltzman, L. E. (2002). Sexual vio-lence surveillance: Uniform definitions andrecommended data elements. Atlanta, GA:National Center for Injury Prevention andControl, Centers for Disease Control andPrevention.

Beck, A. J., & Hughes, T. A. (2005). Sexual vio-lence reported by correctional authorities,2004 (National Criminal Justice Report210333). Washington, DC: U.S. Departmentof Justice.

Bowker, L. (1980). Prison victimization.NewYork:Elsevier North Holland.

Bureau of Justice Statistics. (2004). Data collec-tions for the Prison Rape Elimination Act of2003. Retrieved July 9, 2006, from http://www.ojp.usdoj.gov/bjs/pub/pdf/dcprea03.pdf

Butler, T., Donovan, B., Levy, M., & Kaldor, J.(2002). Sex behind the prison walls. Australian


and New Zealand Journal of Public Health,26(4), 390-391.

Commission on Safety and Abuse in America’sPrisons. (2006). Confronting confinement.Retrieved June 19, 2006, from http://www.prisoncommission.org/report.asp

Fisher, B. S., Cullen, F. T., & Turner, M. G.(2000). The sexual victimization of collegewomen (National Criminal Justice Report182369). Washington, DC: U.S. Departmentof Justice.

Fuller, D. A., & Orsagh, T. (1977). Violence andvictimization within a state prison system.Criminal Justice Review, 2, 35-55.

Gaes, G. G., &Goldberg, A. L. (2004). Prison rape: Acritical review of the literature (Working paper).Washington, DC: National Institute of Justice.

Hensley, C, Tewksbury, R., & Castle, T. (2003).Characteristics of prison sexual assault tar-gets in male Oklahoma correctional facilities.Journal of Interpersonal Violence, 18(6),595-607.

Jargowsky, P. A. (1997). Poverty and place: Ghettos,barrios, and the American city. New York:Russell Sage.

Kilpatrick, D. G. (2000). Rape and sexual assault.Retrieved June 7, 2007, from http://www.musc.edu/vawprevention/research/sa.shtml

Lockwood, D. (1980). Prison sexual violence.New York: Elsevier.

Maitland, A. S., & Sluder, R. D. (1998).Victimization and youthful prison inmates:An empirical analysis. The Prison Journal,78, 55-73.

Presser, S., Rothger, J. M., Couper, M. P., Lessler,J. T., Martin, E. L., Martin, J., & Singer, E.(2004). Methods for testing and evaluatingsurvey questionnaires. Hoboken, NJ: JohnWiley.

Rennsion, C. M., & Rand, M. R. (2003). Criminalvictimization, 2002 (National Criminal JusticeReport 199994). Washington, DC: Bureau ofJustice Statistics.

Straus, M. A. (1979). Measuring intrafamily con-flict and violence: The conflict tactics (CT)scales. Journal of Marriage and the Family,41, 75-88.

Straus, M. A., & Gelles, R. J. (1999). Physical vio-lence in American families. New Brunswick,NJ: Transaction Publishing.

Struckman-Johnson, C, & Struckman-Johnson, D.(2000). Sexual coercion rates in seven mid-western prison facilities. The Prison Journal,80, 379-390.

Struckman-Johnson, C, & Struckman-Johnson, D.(2002). Sexual coercion reported by women inthree midwestern prisons. Journal of SexResearch, 23, 217-227.

Struckman-Johnson, C, & Struckman-Johnson, D.(2006). A comparison of sexual coercion expe-riences of men and women in prison. Journalof Interpersonal violence, 21, 1591-1615.

Struckman-Johnson, C, Struckman-Johnson, D.,Rucker, L., Bumby, K., & Donaldson, S. (1996).Sexual coercion reported by men and women inprison. Journal of Sex Research, 33, 67-76.

Teplin, L. A., McClelland, G. M., Abram, K. M., &Weiner, D. A. (2005). Crime victimization inadults with severe mental illness: Comparisonwith the National Crime Victimization Survey.Archives of General Psychiatry, 62, 911-921.

Tewksbury R. (1989). Fear of sexual assault inprison inmates. The Prison Journal, 69, 62-71.

Tjaden, P., & Thoennes, N. (2000). Full report ofthe prevalence, incidence, and consequencesof violence against women: Findings from thenational violence against women survey(National Criminal Justice Report 183781).Washington, DC: National Institute of Justiceand the Centers for Disease Control andPrevention.

Toch, H. (1985). Warehouses of people. Annals ofAmerican Academy of Political and SocialScience, 478, 58-72.

Travis, J. (2005). But they all come back: Facingthe challenges of prisoner reentry.Washington,DC: Urban Institute.

Wilson, W. J. (1987). The truly disadvantaged: Theinner-city, the underclass and public policy.Chicago: University of Chicago Press.

Wooden, W. S., & Parker, J. (1982). Men behindbars: Sexual exploitation in prison.New York:Plenum.

Wooldredge, J. D. (1994). Inmate crime and vic-timization in a southwestern correctional facil-ity. Journal of Criminal Justice, 22, 367-381.

Wooldredge, J. D. (1998). Inmate lifestyles andopportunities for victimization. Journal ofResearch in Crime and Delinquency,35,480-502.




1. How much have prevalence rates of sexual and physical victimization varied in pastresearch?

2. Why do Wolff and colleagues believe it is important for researchers to clarify whatthey want to measure, especially regarding violence?

3. How did these researchers solicit information from prison inmates? Regarding ethicsof subjects, do you think respondents would have been less comfortable answeringquestions about victimization through face-to-face interviews with the researchers?

4. How much difference in victimization rates did these researchers find for general ver-sus specific questioning?

5. In Table 4.2, did some inmates who answered yes to a specific question confirmingthey had been victimized, also report no, that they had not to a more general ques-tion about victimization? What does this reveal?

6. In what ways are specific questions about violent victimizations better for research?

7. What do these researchers say are some of the problems with comparing victimiza-tions rates inside prison versus rates of the general population not in prison?

8. What are some of the limitations of this research study?

Source: Batton, C., & Kadleck, C. (2004). Theoretical and methodological issues in racial profiling research. Police Quarterly,7(1), 30–64. Copyright © 2004 Sage Publications.

RESEARCH READING

Racial profiling has become a well-recognized and popular buzz word to both researchersand the general public alike. The authors of the current article, however, point out that rel-atively little is known about the cause of racial profiling. They contend that a lack of con-ceptualization of racial profiling is the reason for this void in knowledge. In this article,Batton and Kadleck attempt to conceptualize racial profiling in order to make it moreamenable for future researchers to assess causality. Specifically, they outline four concep-tual issues and the implications each has for data collection and analysis of racial profiling.Pay particular attention to the separate discussions of conceptualization and operationaliza-tion issues regarding racial profiling and how the two are related but different. The appendixprovides information on the various different types of definitions that have been used todefine racial profiling. This appendix is a great example of the difficulty in coming up withuniversal conceptualizations of phenomena that are of interest to social science researchers.

Reading 8 Theoretical and Methodological Issues in Racial Profiling Research • 111

INTRODUCTION

Racial disparities in the justice system havelong been of interest to researchers who havedocumented disproportionate numbers of racialand ethnic minorities at virtually every stage ofjustice processing (Walker, Spohn, & DeLone,2000). The latest issue to garner scholarlyattention with respect to race is racial profiling,a term generally used to describe situations inwhich race or ethnicity functions as an indica-tor of criminal propensity, typically by lawenforcement officers in the context of a trafficstop (Harris, 1997,1999a; Maclin, 1998; U.S.General Accounting Office, 2000). Althoughracial profiling is a relatively new concept thathas only recently generated broad social con-cern, discrimination on the basis of race haslong plagued the U.S. criminal justice system(Kennedy, 1999; Walker, 2001).

The broad, sweeping consequences of racialprofiling for minorities, police, and society ingeneral make it an issue worthy of scholarlyattention. The necessity of research is furtherunderscored by the flurry of legislative actionand reforms to law enforcement policies thathave occurred due to heightened concern overracial profiling and the potential civil libertiesviolation it constitutes (Commission onAccreditation for Law Enforcement Agencies,2002; Harris, 1999b; Strom, Brien, & Smith,

2001; Walker, 2001). Although there havebeen recent efforts to measure the extent ofracial profiling, much of the existing literaturefails to adequately conceptualize the problemin that there is no consensus on the meaning ofracial profiling or what it entails. We use theterm racial profiling to refer to the use of dis-cretionary authority by law enforcement offi-cers in encounters with minority motorists,typically within the context of a traffic stop,that result in the disparate treatment of minori-ties. The defining characteristics of racial pro-filing incidents have yet to be identified, but inthis article, we suggest several factors thatshould be considered.

CONCEPTUAL FRAMEWORK

Social and CulturalFoundation of Racial Profiling

With the “sociological imagination,” C.Wright Mills (1961) stressed the importance ofsituating phenomena in their broader social andcultural context to more fully understand them.Toward that end, we begin by defining racialprofiling and discussing its historical roots. Theterm racial profiling has been used to looselyrefer to a variety of instances in which minori-ties, African Americans in particular, are treated

THEORETICAL AND METHODOLOGICALISSUES IN RACIAL PROFILING RESEARCH

Candice Batton and Colleen Kadleck

Abstract: Racial profiling has generated a significant amount of social concern in American society. Yetlittle is known about the etiology of this phenomenon, in part because its defining characteristics haveyet to be identified and conceptualized. In this article, the authors summarize and organize the existingracial profiling literature and establish a strong conceptual foundation for future racial profiling research.Four issues that must be explicitly addressed in the conceptualization process of research on racial pro-filing are identified and described; realm of activity, level of aggregation and unit of analysis, popula-tion of interest, and characteristics of the incident. These factors have methodological implications interms of data collection control variables, and methods of analysis.

distinctly because of their skin color by personsoccupying various social roles (e.g., store own-ers, taxi drivers, police) (Meeks, 2000).Although a broad conception is not incorrect,we use racial profiling to refer to a narrowerrange of social behavior-specifically, the use ofdiscretionary authority by law enforcementofficers in encounters with minority motorists,typically within the context of a traffic stop.

The roots of racial profiling run deep.Throughout the course of American history,minorities, and Blacks in particular, have beensubjected to differential treatment by the crim-inal justice system (Donziger, 1996; Kennedy,1999; Meeks, 2000; Tonry, 1995; Williams &Murphy, 1990). In the mid-1800s, slave patrolswere implemented as an arm of the law tosearch for disobedient and runaway slaves(Reichel, 1988; Williams & Murphy, 1990).Following the abolition of slavery, “Blackcodes” were enacted in many states in attemptsto legally limit the rights and liberties ofBlacks. Although the federal governmentbanned “Black codes” in 1868, many Southernstates still attempted to use the legal system tomaintain White supremacy with Jim Crowlaws, which required separate facilities forBlacks (e.g., schools, rail cars, bathrooms)(Free, 1996; Williams &Murphy, 1990). It wasnot until the Civil Rights Act of 1964 that itbecame illegal to discriminate on the basis of“race, color, . . . or national origin” (LegalInformation Institute, 2002). Although dis-crimination is prohibited, it is well documentedthat racial disparities exist at virtually everystage of U.S. criminal justice system process-ing (Donziger, 1996; Free, 1996; Mauer, 1994;Tonry, 1995; Walker, 1999; Walker et al.,2000). It is within this social and historical con-text that racial profiling emerged.

Several facets of police training and experi-ences are conducive to racial profiling. First,profiling is generally viewed as an efficient lawenforcement technique for getting contrabandoff the street and “netting” persons who arewanted (Cleary, 2000: Harris, 1999b, 2002;Kennedy, 1999; Milazzo & Hansen, 1999).Unless it is the sole factor prompting a stop, the

use of race/ethnicity in profiling is generally notbelieved to constitute discrimination. In consid-eration of this, racial profiling may occur inconjunction with anti-crime crackdowns. Forexample, in an attempt to get guns off KansasCity streets, a high crime rate area was assignedextra patrols with the task of establishing “rea-sonable suspicion” to perform searches, manyof which occurred in the context of traffic stops(Sherman, Shaw, & Rogan, 1995). Althoughminorities were not explicitly targeted, 92% ofresidents in the targeted area were non-White,Uniform Crime Report (UCR) statistics indicat-ing disproportionate minority involvement incrime also support the use of race in profiling(Taylor & Whitney, 1999). Although they havebeen criticized by academics for not accuratelyreflecting the nature and extent of crime or thecharacteristics of criminals, UCR statistics con-stitute police reality in that they reflect arrestsand offenses known to the police (Robinson,2000). Scholars may argue that self-report andvictimization surveys paint a different image ofracial differences in criminal behavior, but it isunknown to what extent police are aware of thisinformation. On a related note, officers aretrained to look for persons who do not belong orappear suspicious (Johnson, 1983; Maclin,1998). As Skolnick (1994) noted, citizen reportsof “suspicious” persons more often involveminorities than Whites, which reinforces link-ages between criminality and minorities.

Drug interdiction training also facilitatesracial profiling as drug courier profiles aremore likely to target minorities, AfricanAmericans and Hispanics in particular (Cole,1999; Donziger, 1996; Harris, 2002; Leitzel,2001; Maclin, 1998; Tonry, 1995; Walker et al.,2000). Harris (1999b) argued that racial profil-ing was institutionalized in law enforcement inthe mid-1980s through Operation Pipeline, aDrug Enforcement Agency (DEA) trainingprogram on using traffic violations as pretextsfor stopping persons who fit drug courier pro-files, especially on roads known as drug“pipelines.” Once a stop is made on the pretextof a traffic violation, the officer has a betteropportunity to determine whether a search is


warranted (Cole, 1999). Even with no evi-dence, consent for a search may be requested.Motorists can refuse, but they may believe thatconsent is not required or that cooperationdecreases the risk of a citation (Cole, 1999;Harris, 1999a, 2002). If consent is not given, acanine unit may be summoned in an attempt toestablish probable cause for a full search(Harris, 1999a, 2002; Meeks, 2000). The ideathat training includes race/ethnicity as a riskfactor is important because training influencesofficer behavior (McNamara, 1967;Niederhoffer, 1967; Walker, 1999).

All of these factors contribute to the perpet-uation of stereotypes that link criminality withracial and ethnic minorities (Harris, 1999b;Kennedy, 1997). These stereotypes permeatepolice work and are manifested through theuse of discretionary authority contributing tothe belief that searches of minorities and theirvehicles are likely to be “productive” in termsof netting contraband and criminals (Cole,1999; Johnson, 1983; Milazzo & Hansen,1999; Ramirez, McDevitt, & Farrell, 2000;Robinson, 2000). The result is a self-fulfillingprophecy as officers target vehicles operatedby minorities and simultaneously pay lessattention to others (Harris, 1999b; Ramirez et al.,2000). Interestingly, “hit rate” research indi-cates that drugs are found in a smaller percent-age of vehicles operated by minorities than byWhites (Harris, 1999a; Ramirez et al., 2000).Officers vary in the extent to which they arecognizant of how race affects their use of dis-cretion, with many denying that it has anyeffect at all (Johnson, 1983; Ramirez et al.,2000; Robinson, 2000).

Finally, it should be noted that U.S.Supreme Court decisions on the constitution-ality of pretext stops also facilitate racial pro-filing. In 1996, the use of drug courier profilesand pretext stops was upheld by the Court inWhren v. United States as long as race is notthe only motivating factor for the stop orsearch (Cole, 1999; Harris, 1999b; Kennedy,1999). Whether this constitutes racial profilingis the subject of debate. Although some defineracial profiling as occurring when race is the

only factor motivating a stop or search, othersargue that it encompasses situations when raceis used as one of a constellation of factors(Cleary, 2000; Fridell et al., 2001; Kennedy,1999; Walker, 2001). Harris (1999b) arguedthat any use of race is problematic because“police are therefore free to use Blackness as asurrogate indicator or proxy for criminalpropensity” (p. 291). In other words, racebecomes a proxy for risk of criminal behavior(Gaynes, 1993; Hemmens & Levin, 2000;Johnson, 1983; Kennedy, 1999; Mann, 1993).The U.S. Supreme Court handed down a simi-lar decision in 1976 supporting the actions ofBorder Patrol agents who used drivers’ ethnicheritage as a factor in the decision to stop andsearch vehicles on the southern California bor-der (Meeks, 2000).

Explicitly identify factors that should betaken into account in the conceptualizationand operationalization of racial profiling.

CONCEPTUALIZATION

OF RACIAL PROFILING

It is difficult to overstate the importance of clearconceptualization in the research process. Afirm understanding of the meaning and parame-ters of key concepts is necessary for identifyingvalid and reliable measures and adequatelyoperationalizing variables. A strong conceptualframework is also important for model devel-opment in terms of specifying causal mecha-nisms and relevant control variables. Thefollowing discussion focuses on four factorscentral to the conceptualization of racial profil-ing; realm of activity, level of aggregation, pop-ulation of interest, and incident characteristics.

Realm of Activity

It has been well documented that profiling ina broad sense occurs in many realms of activityand by persons occupying a multitude of socialroles. Studies indicate that race and sex arevery powerful determinants of the manner in


which individuals interact with others. They areparticularly important in interactions betweenstrangers who do not have any other priorknowledge of each other and therefore mustrely on outward appearances to “size up” andmake assessments about the other person(Anderson, 1990, 1999; Leitzel, 2001). It hasbeen argued that African Americans are dis-criminated against by retailers who assume thatthey either do not have any money or are morelikely to shoplift, by taxi cab drivers who fearcriminal victimization, and by Customs agentssearching for contraband on the basis of skincolor (Cole, 1999; Meeks, 2000).

Although this type of “profiling” undoubt-edly occurs and is likely based at least in part onrace, of particular interest here is racial profilingthat occurs in the context of police interactionswith minority motorists. First, racial profiling bypolice is likely to be distinct from that done byothers because of the discretionary authority ofthe police and the power differential that existsbetween police and those they stop. Second,traffic law and vehicle code enforcement is anarea in which officers have broad discretion, inpart because of the sheer number of vehiclesoperated in violation of the law (Ramirez et al.,2000; Walker, 2001). Vehicle stops, most ofwhich involve suspected traffic or vehicle codeviolations, are also discretionary as they arelargely proactive, or officer-initiated, as opposedto other types of encounters, which are morelikely to be reactive, or driven by service callsand citizen complaints (Mastrofski et al., 1998;Walker, 2001). Vehicle stops are also distinctfrom many other types of police-citizen encoun-ters in terms of their situational characteristics,the absence of a “victim” demanding that justicebe served, and the potential outcomes (Smith &Visher, 1981; Worden, 1989). Therefore, a dis-tinction should be made between racial profilingin the context of interactions between police andminority motorists and other social settings.

Unit of Analysis and Aggregation

In the process of conceptualizing racial pro-filing, it is important to consider the unit of

analysis and the possibility of effects at differ-ent levels of aggregation. The most basic unitof analysis is the incident, or the individualvehicle stop. For this unit of analysis, datamust be collected on the characteristics of allvehicle stops that occur within a specific geo-graphical area during a set period of time,Incident data are important because officerdecision making is affected by the characteris-tics of incidents. Officers tend to have a lim-ited amount of time and information to use indeciding whether to stop a vehicle, thus theyrely on various aspects of the situation (e.g.,driver behavior, seriousness of the violation,location of the stop, other traffic conditions,and vehicle characteristics) (Worden, 1989).As Walker (2001) noted, the context in whichthe stop occurs should be considered.

Although data should be incident level, it isimportant to consider the possibility of effectsacross officers, districts, and/or agencies. At theofficer level, the risk of racial profiling mayvary because of differences in attitudes andbeliefs about the relationship between race/ethnicity and criminality. Officer-level differ-ences may also stem from differences in train-ing and guidelines from supervisors (Verniero& Zoubek, 1999), which could be detected bycontrolling for shift or beat. District effects arepossible as several factors related to the distrib-ution of crime vary across police districtsincluding the (a) demographic characteristics ofresidents, (b) extent of social disorganization(e.g., population density, mixed use areas,rental properties), and (c) number of officersassigned to an area (Bursik & Webb, 1982;Klinger, 1997; Reiss, 1986; Schuerman &Kobrin, 1986; Shaw & McKay, 1942). Klinger(1997) also argued that the structure of the divi-sion of police labor in conjunction with theautonomous nature of patrol work results in dis-trict variation in policing norms. Finally,agency-level effects are possible as a result ofvariation in goals, operating philosophies, andorganizational culture (Engel, Calnon, &Bernard, 2002; Fridell et al., 2001). This is con-sistent with research by Mastrofski, Ritti,and Hoffmaster (1987) on drunk driving


enforcement and Fyfe (1982) on police use oflethal force; both studies found that agencyphilosophies affect officer behavior.

Population of Interest

Identifying and defining the population atrisk is central to the conceptualization of racialprofiling. Are all persons “at risk” or are par-ticular racial/ethnic groups more likely thanothers to be targeted? The popular phrase “dri-ving while Black” implies that AfricanAmericans are the targets of racial profiling.Studies that focus solely on the experiences ofAfrican American motorists, such as Harris(1999a), further support this idea. However,other racial and ethnic minorities are alsolikely to be targeted. As Mann (1993) noted,Americans stereotype the race or ethnicity ofcriminals as African American, Hispanic, orNative American, depending on the type ofcrime and area of the country.

The extent to which various minority groupsare targeted in racial profiling likely depends onthe racial and ethnic composition of the popula-tion. Particular minority groups may be atgreater risk in areas where their numbers are rel-atively large. For example, African Americansmay be at greater risk than others in the Southbecause they comprise a relatively large propor-tion of the minority population. According to theU.S. Census Bureau (2000, pp. 25-27), in 2000,90% to 95% of the minority population wascomprised of African Americans in states likeLouisiana, Mississippi, and Alabama. The com-position of the minority population differs instates such as Texas and California, whereAfrican Americans comprised 39.6% and16.5% of the 2000 minority population, respec-tively. In both of these states, approximately onethird of the population in 2000 was comprised ofHispanics, and in California about 10% of thetotal population and 27% of the minority popu-lation was comprised of Asians.

The race/ethnicity of known drug couriersand drug-related offenders should also beconsidered in conceptualizing the at-risk pop-ulation. Officers are likely to rely on their

own experiences as well as the collectiveexperiences of the agency, which they areexposed to through radio communications,supervisor reports and messages, and infor-mal exchanges of information. Thus, if themajority of drug trafficking arrestees in anarea are young, Asian males, they may be atgreater risk for racial profiling than AfricanAmericans or Hispanics, even if the lattercomprise a relatively large proportion of theminority population in that jurisdiction.

Characteristics of the Incident

Anecdotal evidence suggests that thenature and quality of the encounter is a crucialaspect of racial profiling (Fridell et al., 2001).Although discussions frequently focus on theuse of pretext stops, what happens during thestop is as important as the “reason” for it.Several factors are identified below as poten-tial defining characteristics of racial profilingincidents.

Length of Delay

Anecdotal evidence suggests that minori-ties are more likely to be detained by lawenforcement officers for traffic violations andfor longer periods of time (Fridell et al, 2001;Harris, 1999a; Ramirez et al., 2000). Althougha White driver may be detained relativelybriefly for a traffic violation for the amount oftime it takes to issue a warning or a citation, ithas been argued that minorities are likely to bedetained longer as more extensive questioning,background checks, and/or searches are con-ducted. Investigating the extent to whichminorities are subjected to longer delays andthe reason for them is an important aspect ofracial profiling research.

Searches

Anecdotal evidence suggests that searchesare an important aspect of racial profiling.Several facets of searches need to be moreclosely examined, including racial/ethnic dif-ferences in the likelihood of being searched,


the rationales underlying searches, and natureof the search. Clearly, the most basic questionis whether minority motorists and theirvehicles are more likely than Whites to besearched, presumably because of a suspicionthat drugs will be found. In a study of police-public contact in 1999, Langan, Greenfeld,Smith, Durose, and Levin (2001, p. 2) foundthat Blacks and Hispanics were twice as likelyas Whites to be searched during a traffic stop.They also found that younger drivers weresearched more often.

The circumstances surrounding the searchare also important-specifically, the basis forinitiating the search (i.e., visible contraband,odor, canine alert, inventory search prior toimpoundment, consent search) (Fridell et al.,2001; Ramirez et al., 2000). If an individualrefused consent to a search, was a canine unitsummoned (Cole, 1999)? If so, how was thedog used (i.e., interior vehicle search, exteriorvehicle search)? Were the search results posi-tive, and if so, what was recovered (e.g., drugs,paraphernalia, weapons) (Fridell et al., 2001)?Another important aspect is whether themotorist felt pressured to consent to a searchout of fear of receiving a citation or ignoranceof the right to refuse consent (Cole, 1999;Harris, 1997; Meeks, 2000). However, it maynot be possible to accurately measure thisfacet of searches.

Property Damaged

As alluded to above, allegations of propertydamage have frequently been made by minor-ity motorists claiming that either their vehicleor personal belongings were damaged in thecourse of officers or canine units conducting asearch of their vehicle (Harris, 1999a).Personal property might be damaged if belong-ings are carelessly rifled through, thrown onthe ground, or exposed to inclement conditions(e.g., rain, snow, mud). Property may also bedamaged as a result of police dogs scratchingthe exterior of the vehicle or entering the vehi-cle and damaging the upholstery. Damagedproperty may be indicative of being treated

with a lack of respect and therefore may tapinto the qualitative nature of the encounter.

Information and Demeanor

The extent to which the officer provides orrequests information may also be important inconceptualizing racial profiling. For example,there is evidence that officers frequently donot give a reason for the stop, even whenasked directly by the motorist. This may indi-cate a lack of professionalism, but Fridell et al.(2000, p. 14) noted that minorities tend toattribute officer rudeness, discourtesy, andunwillingness to provide the reason for thestop to racial bias. It should be noted, how-ever, that Langan et al. (2001, p. 15) found fewracial differences (i.e., 1.5% of Whites, 2% ofBlacks, and 2.3% of Hispanics) in driversreporting that no reason was given for the stop.Researchers should also document whetherofficers ask inappropriate or irrelevant ques-tions about where the individuals have been orwhere they are going. In cases involvingminorities driving more expensive cars, peoplereport being asked who the car belongs to, asif the car did not belong to them. These ques-tions may be particularly revealing when theyhave little to do with the underlying traffic vio-lation (Harris, 1999a).

The demeanor of the officer is an importantaspect of the interaction. This refers to the man-ner in which the officer treats and interacts withthe individual; is the person treated with cour-tesy and respect or is the officer rude and suspi-cious? Information on officer demeanor wouldlikely provide important insights into the issueof racial profiling, but this facet of the interac-tion is virtually impossible to measure withoutdirect observation, which may itself have animpact on officer behavior. Furthermore, offi-cer demeanor is likely to be partially a functionof motorists’ demeanor. As Kennedy (1999, p.7) noted, a “downward spiral” may occur if themotorist expresses resentment at having beenstopped because of racial/ethnic heritage andthe officer responds to the perceived hostility inkind. On a related note, Engel et al. (2002)


reported that non-White suspects may be moredisrespectful toward police, which could belinked with higher arrest rates for minorities,especially young Black males. AlthoughLundman (1996) found that citizen demeanoraffects the likelihood of arrest, other factorsmust also be considered, such as severity of theoffense (Klinger, 1996; Worden, Shepard, &Mastrofski, 1996).

Restraint and Use of Force

Another aspect of the incident to consider isthe use of restraints on the driver and/or pas-sengers. Restraints might include handcuffs,leg shackles, or being placed in the patrol car.The use of restraints typically suggests that theofficer had reason to believe the person wasdangerous or posed a threat to his or hersafety. The more frequent use of restraintswith minority motorists may be an indicationof racial bias in that the decision to use themmay stem from preconceptions on the part ofthe officer about racial differences in criminal-ity, in their study of police-public contact.Langan et al. (2001) found that Blacks (6.4%)and Hispanics (5%) were more likely to behandcuffed during a stop than Whites (2.5%).Blacks (2%) and Hispanics (2%) were alsomore likely than Whites (<1%) to have forceor the threat of force used against them.

Disposition of Stop

Racial profiling may be linked withracial/ethnic differences in the disposition oroutcome of a traffic stop (e.g., winning, cita-tion) (Fridell et al., 2001). Differences in inci-dent dispositions may reflect the extent towhich the stop was based on an actual trafficviolation as opposed to a traffic violationbeing the pretext for the stop. Stops initiatedon the basis of a minor infraction that do notresult in a citation, but do result in a search orattempted search, are consistent with DEAtraining on the use of pretext stops. Ramirezet al. (2000, p. 50) suggested the following dis-position codes: oral warning, written warning,

arrest made, arrest by warrant, criminal cita-tion, traffic citation-hazardous, traffic citation-nonhazardous, courtesy service/citizen arrest,and no action taken.

METHODOLOGICAL IMPLICATIONS OF

CONCEPTUAL DEFINITION

The conceptualization of a problem has impli-cations for conducting research on that phe-nomenon. In the following, a variety ofmethodological issues that follow from theconceptual framework on racial profilingestablished above are discussed.

Operationalization

In establishing the conceptual framework,four factors were discussed as central toracial profiling research: realm of activity,level of aggregation, population of interest,and characteristics of the incident. In the fol-lowing, the operationalization of these con-cepts is discussed as well as the implicationsfor data collection.

Realm of Activity

The conceptualization of racial profilinghas implications both for its operationalizationas a variable and for data collection. Withrespect to the realm of activity, we argue thatthe data collected should pertain to interac-tions between law enforcement officers andmotorists that occur during officer-initiatedvehicle stops. Ramirez et al. (2000) defined astop “as any time an officer initiates contactwith a vehicle resulting in the detention of anindividual and/or vehicle” (p. 43). Althoughthe majority of stops will likely involve trafficlaw violations, other highly discretionary vehi-cle stops (e.g., investigative stops) should alsobe included (Fridell et al, 2001). For example,in their investigation of Operation Valkyrie,the ACLU found that Illinois state troopersassigned to drug interdiction units stopped


Hispanic motorists 2 to 3 times more frequentlythan troopers in other units (Harris, 1999a).However, as Fridell et al. (2001, p. 123)noted, collecting data on vehicle stops asopposed to traffic stops compounds the prob-lem of identifying an appropriate baselinefor comparison.

Although data on the outcome of the stopshould be recorded, it should not serve as thebasis for sample selection. Instead, data shouldbe recorded on all vehicle stops to avoid biasand accurately assess the nature and extent ofracial profiling (Cleary, 2000; Fridell et al.,2001). Data should be collected on the date,time, and location of the stop to fully under-stand the context in which the stop occurred(Cordner, Williams, & Zuniga, 2001; Ramirezet al., 2000).

Level of Aggregation

Researchers should be cognizant of thelevel of aggregation that characterizes theirdata. Most racial profiling studies have exam-ined department-level data (Cleary, 2000),with the exception of legal research, such asLaraberth’s study of the Maryland State Police(ACLU, 1996). Agency-level data are likely touncover important information and may benecessary to protect the anonymity and confi-dentiality of the officers being studied, espe-cially in relatively small departments. AsCleary (2000, p. 28) noted, officers often per-ceive the collection of identification informa-tion as a serious threat, which can affect bothbehavior and morale.

Although these are legitimate concerns,identifying the parameters of the issue isimportant and agency-level data may obscuredifferences among officers and districts. Tothe extent that research findings serve as thebasis for policy and program reforms, it isimportant to know, for example, whetherracial profiling involves a large group of offi-cers, all officers working in a particular districtor shift, or only a few officers. Previousresearch indicates that the percentage of trafficstops involving minorities varies by officer

(ACLU, 1996) and district (Cordner et al.,2001). Therefore, whenever possible, datashould be collected in a way that allows foraggregation at the officer, district, and agencylevel. At the same time, researchers shouldtake precautions to protect the confidentialityand anonymity of participants and be cog-nizant of how the data collection processmight impact the behavior of officers(Ramirez et al., 2000). Researchers shouldalso control for factors such as beat assign-ments, especially with data aggregated at theofficer level, that may be related to the likeli-hood of encountering minority motorists.

Population of Interest

Identifying the population of interest andestablishing a baseline measure of this group isperhaps the most difficult aspect of operational-izing racial profiling (Ramirez et al., 2000). AsCleary (2000) noted, “there is typically no per-fect baseline measure against which to gaugepolice stop practices” (p. 24). One commonapproach is to use the racial/ethnic compositionof the community as a baseline against which tocompare the racial distribution of traffic stops(Cleary, 2000; Engel et al., 2002). See, forexample Cox, Pease, Miller, and Tyson (2001);Nixon (2001); Texas Department of PublicSafety (2000), and Verniero and Zoubek (1999).However, this approach is problematic because itassumes that all members of racial/ethnic sub-groups drive and thus are at risk of being stopped(Walker, 2001). Also implicit is the assumptionof vehicle ownership and a valid driver’s license,both of which vary by race/ethnicity (Doyle &Taylor, 2000; Polzin, Chu, & Rey, 2000).

Given these issues, some researchers haveused the racial/ethnic composition of the dri-ving population as a baseline measure (Harris,1999a). For example, a study of NorthCarolina State Highway Patrol data used dis-trict residents holding a valid driver’s licenseas a baseline for comparisons (Zingraff et al.,2000). Although an improvement, this mea-sure is also problematic in that the racial/ethnic composition of the driving population


within a particular jurisdiction varies acrossdifferent stretches of roadway, times of the day,and days of the week (Cleary, 2000; Guiliano,2000; Valenzuela, 2000). Furthermore, itassumes no racial/ethnic differences in trafficviolations and equipment failure rates (Zingraffet al., 2000). As Engel et al. (2002) noted, it ispossible that certain subgroups of the popula-tion (e.g., young Black males) drive moreaggressively and commit more moving viola-tions. Finally, both measures fail to account fornonresidents traveling in the area under study,which itself impacts baseline measures (Coxet al., 2001). Walker (2001) argued that suchproblems may well be insurmountable and sug-gested an “early warning system” approach asan alternative to attempts to establish a baseline.

Another challenge is the determination ofrace/ethnicity by law enforcement officers(Cleary, 2000). The racial/ethnic categoriesused in police record keeping vary acrossagencies (Cleary, 2000; Ramirez et al., 2000),which is logical given regional variations inthe composition of the population. Researchon racial profiling should employ racial/ethniccategories that are socially meaningful withinthe jurisdiction of the agency. However, tofacilitate comparisons with previous research,the categories should be able to be collapsed toreflect those used in the UCRs. Also, thenumber of categories should be relativelysmall to enhance the reliability and validity ofthe measures and to minimize error in the datacollection process. Although they may notmatch motorists’ self-assessments, officerdeterminations of race/ethnicity are generallybest because their perceptions (rather thanactual race/ethnicity) are what is central to theissue of racial profiling (Cleary, 2000).However, as Walker (2001, p. 74) noted, con-textual factors impede the ability of officers todetermine race/ethnicity, including tinted carwindows and limited vision at night.

In addition to recording the motorists’race/ethnicity, data on age and sex should alsobe collected to determine whether disparitiesexist along these dimensions as well (Fridellet al., 2001). For example, the Washington

State Patrol and Criminal Justice TrainingCommission (2001) found that drivers youngerthan 30 were more likely to be cited, regardlessof race. In their North Carolina study, Zingraffet al. (2000) found that age played a role inracial differences in that Black males aged 23to 49 (23%) and 49 and older (70%) were morelikely to receive a citation than their Whitecounterparts. Finally, Cordner et al. (2001)found a greater proportion of San Diego trafficstops involved males rather than females.

Characteristics of the Incident

Research on racial profiling must determinethe extent to which racial disparities exist inthe characteristics of stops. In looking at thecharacteristics of the incident, the focus is onwhat happens from the time the stop is initi-ated to when it is completed. To the extent thatthe data are available in agency records andactivity logs, collecting and coding data onseveral factors should be relatively straightfor-ward. For example, Zingraff et al. (2000) mea-sured length of delay in minutes and thencomputed mean delay times by race. It may bepossible to collect data on length of delay ifthe times that the stop was initiated and thatthe officer returned to service were recordedon activity logs or could be determined byreviewing in-car camera tapes or communica-tions with dispatch, both of which may betime-date stamped.

Data should be collected on whether a vehi-cle search was conducted. With Missouri data,Nixon (2001) calculated search rates (i.e.,number of searches/number of stops) by race.Similarly, Verniero and Zoubek (1999) andthe Texas Department of Public Safety (2000)looked at search percentages by race. Datashould also be collected on the basis or reasonfor the search and whether it involved a canineunit. For example, Cordner et al. (2001) dis-tinguished between nine types of searches inSan Diego (i.e., inventory, incident to arrest,consent, Fourth Amendment waiver, otherbasis, odor of contraband, contraband visible,evidence of criminal activity, and canine


alert). Finally, information on the search out-come is important. In their North Carolinastudy, Zingraff et al. (2000) looked at the per-centage of searches resulting in citations andwarnings by race.

Researchers should attempt to determinewhether restraints were used and, if so, whattype of restraint (handcuffs, placed in policecar). Data on this facet of the incident shouldbe available in police reports. Researchersshould also attempt to determine whetherforce (e.g., held at gunpoint) was used by theofficer.

Data on the disposition of the stop shouldbe available in activity logs and/or warningand citation records. Information should becollected on whether the motorist was releasedwithout a ticket, given a verbal warning,issued a warning or citation for a traffic viola-tion, or arrested as a result of a vehicle search.Although several researchers have looked atwhether or not citations or warnings wereissued, with the goal of discerning racial dis-parities (Texas Department of Public Safety,2000), information on the nature of the warn-ing, citation, or arrest is also desirable. See, forexample, Cox et al. (2001, p. 4) and Ramirezet al. (2000, p. 19). It is also important to dis-tinguish between warnings and citationsissued in the context of traffic accidents, as didZingraff et al. (2000) in their study of NorthCarolina Highway Patrol data.

Although data collection on some incidentcharacteristics may be relatively straightfor-ward, gathering information on other charac-teristics is likely to prove more challenging, ifnot impossible. For example, it would be diffi-cult to identify a valid measure of propertydamage given that both motorists and officersmay falsify information to further their owninterests. Furthermore, in cases in which prop-erty damage is disputed, it would be virtuallyimpossible to determine the truth of the situa-tion. Although it may be possible to rely on in-car video cameras, if available, to some extent,the angle or quality of video footage is ofteninsufficient to allow for a clear determination.When possible, data should be collected on

whether damage occurred to either the vehicleor its contents during the course of a search.

It is also likely to be very difficult, if notimpossible, to collect data on the nature of theofficer-citizen interaction to get at requests forinformation and demeanor. Demeanor ishighly subjective. Although it may be possiblein some cases to use videos from in-car cam-eras to see basic aspects of officer-motoristinteraction during stops, the quality of thefootage is often rather poor, as noted above,and typically does not include sound. Withbetter quality videos that include sound, it maybe possible to see facial expressions and bodylanguage and to hear the nature of theexchange of information as well as the tone.Although it may be possible for a separateobserver to record information about thenature of the interaction, his or her presence islikely to have its own independent effects onthe outcome of the interaction.

SUMMARY AND CONCLUSIONS

In recent years, racial profiling has generated asignificant amount of controversy and socialconcern in American society. Although storiesabout racial profiling abound, very little isknown about the etiology of this phenomenon.In this article, we attempt to summarize andorganize the existing literature on racial profil-ing and establish a strong conceptual foundationfor future racial profiling research. We arguethat the failure to adequately conceptualizeracial profiling is one of the primary problemswith current research. In an effort to fill this gap,we identify and describe four issues that must beexplicitly addressed in the conceptualizationprocess of research on racial profiling: realm ofactivity, level of aggregation, population ofinterest, and characteristics of the incident.These factors are central to the concept of racialprofiling and have implications for the futuredirection of racial profiling research.

One of the primary tasks for future researchis to accurately document the nature andextent of racial profiling in American society,


which is no small task. Although it may be dif-ficult (if not impossible) to accurately assessall of the factors that impact racial profiling,there is still a significant amount of work thatcan be done to further knowledge on the issueand enhance our understanding of the factorsthat influence it. Of particular importance isthe development of multivariate models thatattempt to control for the effects of a variety offactors with the goal of constructing morecomplete and accurate causal models. It isimportant to move beyond simple compar-isons across racial/ethnic categories of thelikelihood of being pulled over or of beingissued a citation. We have identified severalfactors that should be controlled for in racialprofiling research, but note that this list is notexhaustive. Instead, as more is learned aboutvehicle stops and the factors that affect policedecision making within that context, addi-tional potential control variables are likely tobe identified. As models continue to berefined, a better sense of the relative impact ofthe factors discussed will emerge.

APPENDIX

The table below categorizes the different waysthat the term racial profiling is used in the liter-ature. In addition to variation between sourcesin terms of whether an explicit definition ofracial profiling is provided, there are subtle andnot-so-subtle variations in the implied andexplicitly stated meanings of the term.

Definitions of Racial Profiling

Cleary (2000, pp. 5-6)—Conceptual

“Under the narrow definition, racial profilingoccurs when a police officer stops, questions,arrests, and/or searches someone solely on thebasis of the person’s race or ethnicity. . . . Underthe broader definition, racial profiling occurswhen a law enforcement officer uses race or eth-nicity as one of several factors in deciding tostop, question, arrest, and/or search someone.”

Fridell et al. (2001, p. 5)—Conceptual

“Racially biased policing occurs when lawenforcement inappropriately considers race orethnicity in deciding how to intervene in a lawenforcement capacity.”

Harris (1999a)—Conceptual and Operational

“African-Americans call it ‘driving whileBlack’-police officers stopping, questioning,and even searching Black drivers who havecommitted no crime, based on the excuse of atraffic offense.” (p. 265)

“The task was the same as Lamberth’s . . . usestatistics to test whether Blacks in Ohio werebeing stopped in numbers disproportionate totheir presence in the driving population.” (p. 281)

Kennedy (1999, p. 11)—Conceptual

“Properly understood, then, racial profilingoccurs whenever police routinely use race as anegative signal that, along with an accumulationof other signals, causes an officer to react withsuspicion.”

Langan, Greenfeld, Smith, Durose, and Levin(2001, p. 20)—Operational

“One definition of racial profiling is ‘using raceas a key factor in deciding whether to make atraffic stop.’ Another definition is ‘using race asa key factor in deciding whether, during a trafficstop, to search the vehicle or the driver.’”

Meeks (2000, pp. 4-5)—Conceptual

Racial profiling is “the tactic of stopping some-one only because of the color of his or her skinand a fleeting suspicion that the person is engag-ing in criminal behavior.”

Minnesota Department of Public Safety(2000, p. 2)—Conceptual

“To help determine whether motor vehicle dri-vers in Minnesota are being stopped by lawenforcement officers because of their race.”


Muharrar (1998, p. 1)—Conceptual

“Racial profiling—the discriminatory practiceby police of treating Blackness (or brownness)as an indication of possible criminality.”

Newport (1999) for Gallup Organization—Operational

“It has been reported that some police officersstop motorists of certain racial or ethnic groupsbecause the officers believe that these groupsare more likely than others to commit certaintypes of crimes. Do you believe that this prac-tice, known as ‘racial profiling,’ is widespreador not?”

Nixon (2001, p. 1)—Operational

“Racial profiling . . . defined as the inappropri-ate use of race when making a decision to stop,search, cite, or arrest a person.”

Ramirez, McDevitt, and Farrell (2000, p. 3)—Conceptual

“Racial profiling is defined as any police-initiated action that relies on the race, ethnicity,or national origin rather than the behavior ofan individual or information that leads thepolice to a particular individual who has beenidentified as being, or having been, engaged incriminal activity.”

Taylor and Whitney (1999, p. 507)—Conceptual

“‘Racial profiling’ by the police, that is, thepractice of questioning Blacks in suspicious cir-cumstances in disproportionate numbers in theexpectation that they are more likely than peopleof other races to be criminals.”

United States General Accounting Office(2000, p. 1)—Conceptual

“Racial profiling of motorists by law enforce-ment-that is, using race as a key factor in decid-ing whether to make a traffic stop.”

Verniero and Zoubek (1999, p. 5)—Conceptual

“We choose to define ‘racial profiling’ broadlyenough to encompass any action taken by a soletrooper during a traffic stop that is based uponracial or ethnic stereotypes and that has theeffect of treating minority motorists differentlythan nonminority motorists.”

Walker (2001, p. 64)—Conceptual

“Racial profiling refers to allegations that policeofficers stop African American drivers foralleged traffic violations on the basis of race andnot because of legitimate suspicion of any lawviolation.”

Washington State Patrol (2001, p. 1)—Conceptual and Operational

“‘Racial profiling,’ or the targeting of certainracial groups during the course of conductingtraffic stops”

“Officer-initiated contacts of violators . . . [shouldbe] . . . racially proportionate to two standards:driving-age populations and collisions.”

Meaning of Racial Profiling Assumed orInferred Through Examples

Cohen, Lennon, and Wasserman (2000, p. 2)—No Definition

“The well-founded belief that authorities useracial profiles to justify more intensive observa-tion and questioning of people of color has fedescalating tensions between police and minoritycommunities.”

Cordner, Williams, and Zuniga (2001)—NoDefinition

Used the term racial profiling but did not defineit. Compared the percentage of minorities andWhites in the population to the percentage ofminority and White drivers stopped.


Harris (1999b)—No Definition

Used the term racial profiling but did notdefine it. Instead provided examples, talkedabout “driving while Black” and the use of racein pretext stops.

Leitzel (2001)—No Definition

Used term racial profiling but did not define it.Used term race-based policing more often,inferring that it is the practice of using racealone or in concert with other factors as a reasonto stop or detain a person.

Texas Department of Public Safety (2000)—No Definition

Used term racial profiling but did not define it.Referred to instances of inappropriate use ofrace in decisions regarding motorists.

REFERENCES

American Civil Liberties Union. (1996). Report ofJohn Lamberth, Ph.D. (In the Courts).Retrieved February 12, 2002, from www.aclu.org/court/lamberth.html

Anderson, E. (1990). Streetwise: Race, class, andchange in an urban community. Chicago:University of Chicago Press.

Anderson, E. (1999). The code of the street:Decency, violence, and the moral life of theinner city. New York: Norton.

Bursik, R. J., & Webb, J. (1982). Communitychange and patterns of delinquency. AmericanJournal of Sociology, 88, 24-42.

Chambliss, W. (1994). Policing the ghetto under-class: The politics of law and law enforcement.Social Problems, 41, 177-194.

Cleary, J. (2000). Racial profiling studies in lawenforcement; Issues and methodology. St. Paul;Minnesota House of Representatives, ResearchDepartment.

Cohen, J. D., Lennon, J. J., & Wasserman. R.(2000). End racial profiling. Blueprint,volume 8. Retrieved July 13,2002 from http://www.ndol.org/blueprint/fall2000/cohen-lennon-wasserman.html

Cole, D. (1999). No equal justice; Race and classin the American criminal justice system.New York: New Press.

Commission on Accreditation of Law EnforcementAgencies. (2002, March 29). Racial profilingupdate—2000 state actions. Retrieved April13, 2002, from www.calea.org/newweb/news/2000Stateactions.htm

Cordner, G., Williams, B., & Zuniga, M. (2001).Vehicle stops for the year 2000: Executivesummary. San Diego, CA: San Diego PoliceDepartment.

Council on Crime and Justice. (2001). MinneapolisPolice traffic stops and driver’s race; Analysisand recommendations. Minneapolis, MN:Author.

Cox, S., Pease, S., Miller, D., & Tyson, C. B.(2001). Interim report of traffic stops statisticsfor the state of Connecticut. Rocky Hill, CT:Office of the Chief State’s Attorney, State ofConnecticut Retrieved June 24, 2002 fromhttp://web.wtnh.com/Report_Narrative.pdf

Decker, S. (1981). Citizen attitudes toward thepolice: A review of past findings and sugges-tions for future policy. Journal of PoliceScience and Administration. 9, 80-87.

Donziger, S. R. (1996). The real war on crime: Thereport of the National Criminal JusticeCommission. New York; HarperCollins.

Doyle, D. G., & Taylor, B. D. (2000). Variation inmetropolitan travel behavior by sex and eth-nicity. In Battelle (Ed.), Travel patterns ofpeople of color (pp. 181-244). Washington,DC: Federal Highway Administration.

Engel, R. S., Calnon, J., & Bernard, T. J. (2002).Theory and racial profiling: Shortcomings andfuture directions in research. Justice Quarterly,19, 249-273.

Fridell, L., Lunney, R., Diamond, D., Kubu, B., withScott, M., & Laing, C. (2001). Racially biasedpolicing: A principled response. Washington,DC: Police Executive Research Forum.

Free, M. D. (1996). African Americans and thecriminal justice system. New York: Garland.

Fyfe, J. J. (1982). Blind justice; Police shootings inMemphis. The Journal of Criminal Law andCriminology, 73,707-722.

Gaynes, E. (1993). The urban criminal justice system:Where young + Black + male = probable cause.Fordham Urban Law Journal, 20, 621.

Guiliano, G. (2000). Residential location dif-ferences in people of color. In Battelle


(Ed.), Travel patterns of people of color (pp.91-136). Washington, DC; Federal HighwayAdministration.

Harris, D. A. (1997). “Driving while Black” and allother traffic offenses: The Supreme Court andpretextual traffic stops. The Journal ofCriminal Law and Criminology, 87, 544-582.

Harris, D. A. (1999a). The stories, the statistics, andthe law; Why ‘driving while Black’ matters.Minnesota Law Review, 84, 265-325.

Harris, D. A. (1999b). Driving while Black: Racialprofiling on our nation’s highways. RetrievedMay 23, 2001, from the American CivilLiberties Union Web site: www.aclu.org/profiling/report/index.html

Harris, D. A. (2002). Profiles in injustice: Whyracial profiling cannot work. New York: NewPress.

Hemmens, C., & Levin, D. (2000). Resistance isfutile; The right to resist unlawful arrest in anera of aggressive policing. Crime andDelinquency, 46, 472-496.

James, J. (2000). The dysfunctional and the disap-pearing: Democracy, race and imprisonment.Social Identities, 6, 483-493.

Jensen, G. (2000). Prohibition, alcohol, and mur-der; Untangling countervailing mechanisms.Homicide Studies, 4, 18-36.

Johnson, S.L. (1983). Race and the decision todetain a suspect. Yale Law Journal, 93, 214-258.Kennedy, R. (1997). Race, crime and the law.New York; Pantheon.

Kennedy, R. (1999, September 13). Suspect policy.The New Republic, 13, 30. Retrieved March29, 2002, from http://www.tnr.com/archive/0999/09l399/coverstoryO91399.html

Klinger, D, A. (1996). Bringing crime back in:Toward a better understanding of police arrestdecisions. Journal of Research in Crime andDelinquency, 33, 333-336.

Klinger, D. A. (1997). Negotiating order in patrolwork: An ecological theory of police responseto deviance. Criminology, 35,277-306.

Kreft, I., & de Leeuw, J. (1998). Introducing multi-level modeling. Thousand Oaks, CA: Sage.Langan, P. A., Greenfeld, L. A., Smith, S. K.,Durose, M. R., & Levin, D. J. (2001). Contactsbetween the police and the public: Findingsfrom the 1999 National Survey (No. NCJ184957). Washington, DC: Bureau of JusticeStatistics, U.S. Department of Justice.

Lawson, B. (1992). The underclass question.Philadelphia: Temple University Press.

Legal Information Institute. (2002). Civil rights; Anoverview. Ithaca, NY: Cornell Law School.Retrieved March 2, 2002 from www.law.cornell.edu/topics.civilrights.html

Leitzel, J. (2001). Race and policing. Society, 38,38-43.

Lundman, R. J. (1996). Demeanor and arrest:Additional evidence from previously unpub-lished data. Journal of Research in Crime andDelinquency, 33, 306-323.

Maclin, T. (1998). Terry v. Ohio’s FourthAmendment legacy: Black men and police dis-cretion, St. John’s Law Review, 72, 1271-1321.

Mann, C. R. (1993). Unequal justice: A question ofcolor. Bloomington: Indiana University Press.

Mastrofski, S. D., Parks, R, B., Reiss, A. J.,Warden, R. E., DeJong, C., Snipes, J. B., et al.(1998). Systematic observation of public pol-icy: Applying field research methods to policyissues (NIJ No. NCJ 172859). Washington,DC: Government Printing Office.

Mastrofski, S. D., Ritti, R. R., & Hoffmaster, D.(1987). Organizational determinants of policediscretion: The case of drinking-driving.Journal of Criminal Justice, 15, 387-402.

Mauer, M. (1994). Americans behind bars; Theinternational use of incarceration, 1992-93,Washington, DC: The Sentencing Project.

McNamara, J. H. (1967). Uncertainties in policework; The relevance of police, recruits’ back-grounds and training. In D. J. Bordua (Ed.),The police: Six sociological essays (pp. 163-252).New York: John Wiley.

Meeks, K. (2000). Driving while Black: Highways,shopping malls, taxicabs, sidewalks; how tofight back if you are a victim of racial profil-ing. New York; Broadway Books.

Milazzo, C., & Hansen, R. (1999, October 30-November 3). Race relations in police opera-tions: A legal and ethical perspective. Paperpresented at the 106th Annual Conference ofthe International Association of Chiefs ofPolice, Charlotte, NC.

Mills, C. W. (1961). The sociological imagination.New York: Grove.

Minnesota Department of Public Safety. (2000,November), Racial profiling report. RetrievedSeptember 13, 2001, from mnclu.org/nr_racial_profiling.html


Muharrar, M. (1998, September/October). Mediablackface; “Racialprofiling” in news report-ing. Retrieved May 23, 2001, from www.fair.org/extra/9809/media-blackface.html

Newport, F. (1999). Racial profiling is seen aswidespread, particularly among young Blackmen. Princeton, NJ: The Gallup Organization.Retrieved May 23, 2001 from www.gallup.com/poll/releases/pr991209.asp

Niederhoffer, A. (1967). Behind the shield: Thepolice in urban society. Garden City, NY:Doubleday.

Nixon, J. (2001). 2000 annual report of Missouritraffic stops. Jefferson City: Missouri AttorneyGeneral’s Office.

Polzin, S. R, Chu, X., & Rey, J. R, (2000).Demographics of people of color. In Battelle(Ed.), Travel patterns of people of color (pp.27-44). Washington, DC: Federal HighwayAdministration.

Ramirez, D., McDevitt, J., & Farrell, A. (2000). Aresource guide on racial profiling data collec-tion systems; Promising practices and lessonslearned (USDOJ No. NCJ 184768).Washington, DC: Government Printing Office.

Reichel, P. L. (1988). Southern slave patrols as atransitional police type. American Journal ofPolice, 7, 51-77.

Reiss, A. J., Jr. (1986). Why are communities impor-tant in understanding crime? In A. J. J. Reiss &M. Tonry (Eds.), Communities and crime(pp. 1-34). Chicago: University of Chicago Press.

Robinson, M. (2000). The construction and rein-forcement of myths of race and crime. Journalof Contemporary Criminal Justice, 16, 133-156.

Schuerman, L., & Kobrin, S. (1986), Communitycareers in crime. In A. J. J. Reiss & M. Tonry(Eds.), Communities and crime (pp. 67-100).Chicago: University of Chicago Press.

Shaw, C, & McKay, H. (1942). Juvenile delin-quency and urban areas. Chicago: Universityof Chicago Press.

Sherman, L., Shaw, J. W., & Rogan, D.P. (1995).The Kansas City gun experiment (NIJ No. NCJ150855). Washington, DC: National Instituteof Justice.

Skolnick, J. (1994), Justice without trial (3rd ed.).New York: Macmillan.

Smith, D. A., & Visher, C. A. (1981). Street-leveljustice: Situational determinants of police arrestdecisions. Social Problems, 29(2), 167-177.

Strom, K., Brien, P., & Smith, S. (2001). Trafficstop data collection policies for state police,2001 (BJS No. NCJ 191158;. Washington,DC: U.S. Department of Justice, Bureau ofJustice Statistics.

Taylor, J., & Whitney, G. (1999). Crime and racialprofiling by U.S. police: Is there an empiricalbasis? The Journal of Social, Political, andEconomic Studies, 24, 485-509.

Texas Department of Public Safety. (2000, October).Traffic stop data report. Retrieved June 20,2002, from www.txdps.state.tx.us/director_staff/public_information/ trafrep2q00.pdf

Toby, J. (2000). Are police the enemy? Society, 37,38-43.

Tonry, M. (1995). Malign neglect: Race, crime,and punishment in America. New York:Oxford University Press.

Torres, D. A. (1987). Handbook of state police,highway patrols, and investigative agencies.New York: Greenwood.

U.S. Census Bureau. (2000). Statistical abstract ofthe United States 2000. Washington, DC:Government Printing Office.

U.S. General Accounting Office. (2000). Racialprofiling limited data available on motoriststops (Tech. Rep. No. GAO/GGD-00-41).Washington. DC: Author.

Valenzuela, A. (2000). Race, inequality, and travelpatterns among people of color. In Battelle(Ed.), Travel patterns of people of color(pp. 1-26). Washington, DC: FederalHighway Administration.

Verniero, P., & Zoubek, P. H. (1999). Interim reportof the state police review team regarding alle-gations of racial profiling. Office of theAttorney General, New Jersey Department ofLaw and Public Safety. Retrieved June 19,2002 from www.state.nj.us/lps/intm_419.pdf

Walker, S. (1999). The police in America: An intro-duction (3rd ed.). Boston; McGraw-Hill College.

Walker, S. (2001). Searching for the denominator:Problems with police traffic stop data and anearly warning system solution. JusticeResearch and Policy, 3, 63-95.

Walker, S., Spohn, C., & DeLone, M. (2000). Thecolor of justice: Race, ethnicity, and crime inAmerica. Belmont, CA: Wadsworth/ThomsonLearning.

Washington State Patrol & Criminal JusticeTraining Commission. (2001, January). Report


to the legislature on routine traffic stop data.Retrieved June 20, 2002, from WashingtonState Patrol Web site: www.wa.gov/wsp/reports/demogra2.doc

Wilbanks, W. (1987). The Myth of a racist criminaljustice system. Monterey, CA: Brooks/Cole.

Williams, H., &, Murphy. P. (1990). The evolv-ing strategy of police: A minority view (N1JNo. 13). Washington, DC: GovernmentPrinting Office.

Wilson, W. J (1992). The ghetto underclass.Newbury Park, CA: Sage.

Worden, R. E. (1989). Situational and attitudinalexplanations of police behavior: A theoretical

reappraisal and empirical reassessment. Lawand Society Review, 23, 668-711.

Worden, R. E., Shepard, R. L., & Mastrofski, S. D.(1996). On themeaning andmeasurement of sus-pects’ demeanor toward the police: A commenton “demeanor and arrest.” Journal of Researchin Crime and Delinquency, 33, 324-332.

Zingraff, M. T., Mason, H. M., Smith, W. R.,Tomuskovic-Devey, D., Warren, P.,McMurray, H. L., et al. (2000, November 1).Evaluating North Carolina State HighwayPatrol data; Citations, warnings, and searchesin 1998. Retrieved June 20, 2002, fromwww.nccrimecontrol.org/shp/neshpreport.htm



1. According to these authors, is racial profiling a relatively new phenomenon?

2. Batton and Kadleck outline four factors they believe are central to conceptualizingracial profiling. They point out that in broader contexts, race and sex influence socialinteractions. Why do these authors argue that police interactions are different?

3. What is problematic about using the incident (individual traffic stop) as the unit ofanalysis in racial profiling research?

4. What is it about the incidents (the traffic stops) themselves that is important to analyze?

5. Why might racial profiling be greater in some areas of the country than others?

6. What, according to these authors, is the most difficult component of measuring racialprofiling?

7. What are some of the ethical issues that might arise in research on racial profiling?

8. Could the persons who are stopped by law enforcement also have a role in the out-come of that stop?

Levitt Sample.qxd - SAGE Publications Ltd

Documents

Transcript of Levitt Sample.qxd - SAGE Publications Ltd