University of Southern California Dissertations and Theses

Factors In The Conviction Of Law Violators: The Drinking Driver

(USC Thesis Other)

Factors In The Conviction Of Law Violators: The Drinking Driver

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 71-16 ,4-32 PURDY, Ross Lloyd, 1941- FACTORS IN THE CONVICTION OF LAW VIOLATORS: THE DRINKING DRIVER. University of Southern California, Ph.D., 1971 Sociology, criminology University Microfilms, A X E R O X Company , Ann Arbor, Michigan THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED FACTORS IN THE CONVICTION OF LAW VIOLATORS THE DRINKING DRIVER by Ross Lloyd Purdy A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Sociology) January 1971 UNIVERSITY O F SO U T H E R N CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 9 0 0 0 7 This dissertation, written by ............. ROSS LLOYD PURDY_____________ under the direction of hX§ Dissertation Com mittee, and approved by all its members, has been presented to and accepted by The Gradu ate School, in partial fulfillment of require ments of the degree of D O C T O R OF P H I L O S O P H Y a M, O .......................f H z r Date S e P.?e. I P^er. 1970 DISSERTATION COMMITTEE ACKNOWLEDGEMENTS I wish to thank Jack Bishop, Oksana Didenko and the staff of the Drinking Driver and Traffic Safety Study for collecting the data and permitting its use. To Drs. Jon P. Miller and Sanford I. Labovitz I express my thanks for the hours of editing without which this dissertation would be unreadable. Finally, my thanks go to my wife, Joyce, whose typing skills far surpass mine. TABLE OF CONTENTS Page ACKNOWLEDGEMENTS................................... ii LIST OF TABLES..................................... iv Chapter I. DEVIANT BEHAVIOR AND THE LABELLING PERSPECTIVE 1 Discretion in the Legal System......... 4 Factors Related to Arrest and Conviction . . 6 The Crime of Drinking and Driving..... 18 II. RATIONALES AND HYPOTHESES................ 30 III. METHODS AND DATA.......................... 47 The Instrument.......................... 47 Validity of the Data.................... 51 Legal Measurement of Drunkenness ...... 58 Operational Definitions ................... 59 Measures of Association............... 64 IV. FINDINGS................................... 71 Zero-Order Correlations ................... 71 Statistical Control for Actual Illegal Behavior.............................. 79 Discrimination and the Legal System .... 88 V. CONCLUSION................................. 103 Summary................................ 103 Future Research........................ 112 LIST OF TABLES Table 1. Percentage Distributions of the Total DMV Group (1968) and the Population of Los Angeles County (1960) on Age, Race-ethnicity, Family Income, Marital Status, and Education . . . . 2. Occupational Status Cross-Classified by Conviction Status among Admitted Drinking Drivers ..................................... 3. Educational Level Cross-Classified by Conviction Status among Admitted Drinking Drivers ..................................... 4. Total Family Income Cross-Classified by Conviction Status among Admitted Drinking Drivers ..................................... 5. Age Cross-Classified by Conviction Status among Admitted Drinking Drivers ................... 6. Correlations Between Conviction Status and Race, Social Class, Sex and Age Among Admitted Drinking Drivers, Controlling for Frequency of Drinking and Driving, and Frequency of Drunk Driving ................. 7. Correlations Between Police Severity and Race, Social Class, Sex and Age Among Admitted Drinking Drivers, Controlling for Frequency of Drinking and Driving and Frequency of Drunk Driving ............................... 8. Correlations Between Legal Charge and Race, Social Class, Sex and Age Among Convicted Drinking Drivers, Controlling for Blood Alcohol Level and Total Convictions for Drinking and Driving ....................... 9. Correlations Between Fine Charged and Race, Social Class, Sex and Age Among Convicted Drinking Drivers, Controlling for Blood Alcohol Level and Total Convictions for Drinking and Driving ....................... Page 52 73 75 76 78 80 89 97 98 iv Table 10. Correlations Between Number of Days in Jail on a Conviction for Drinking and Driving and Race, Social Class, Sex and Age Among Convicted Drinking Drivers, Controlling for Blood Alcohol Level and Total Convictions for Drinking and Driving ................. Page 100 v CHAPTER I DEVIANT BEHAVIOR AND THE LABELLING PERSPECTIVE Throughout its history the study of deviant behav ior has taken as its basic question: Why do people do things they shouldn't? The history of research in this field has been characterized by a sequence of independent variables, each one being proposed as the cause of deviant activity. In general, the method of study is based on an examination of known deviants, and a comparison of their characteristics with those of the general population. On the basis of such research a sizeable literature has developed describing the wide variety of ways deviants diverge from the non-deviant population (see Cohen, 1966). Characteristics differentiating between these two broad aggregates are offered as the possible causes of deviance, and based on these findings, several theories of deviance have been developed (for example: Merton, 1938; Cohen, 1955; Cloward and Ohlin, 1960; and Miller, 1958). * Most of the descriptive study and theoretical development in this area rests on the questionable assump tion that apprehended deviants either constitute all 1 deviant actors or are a representative sample of them. Hidden deviance studies suggest that this is an erroneous assumption. Such studies are characterized by the use of non-official sources of data; usually they are based upon self-reported deviance. These studies clearly indicate that there are many more deviants than appear in official records. The undetected deviants, furthermore, appear to be much more like the general population than the appre hended offenders. If apprehended deviants are a biased sample of all violators, then existing theories of deviance which are based solely on studies of the apprehended are probably inadequate. The results of these studies also raise a question regarding why this difference between apprehended and non-apprehended offenders occurs. A number of articles representing what is termed the "labelling" perspective-have recently appeared partly as a response to the findings of hidden deviance studies. According to this perspective, the process of becoming seen or labelled as a deviant does not automatically follow the commission of a deviant act. Rather, acquiring the deviant label is somewhat problematic and is viewed as a process of interaction between the actor and some social audience. In explicating the labelling perspective Becker states that: . . . deviance is not a simple quality, present in some kinds of behavior and absent in others. Rather, it is the product of a process which involves 3 responses of other people to that behavior. The same behavior may be an infraction of the rules at one time and not at another; may be an infraction when committed by one person, but not when committed by another; some rules are broken with impunity, others are not. In short, whether a given act is deviant or not depends in part on the nature of the act (that is, whether or not it violates some rule) and in part on what other people do about it. (Becker, 1964: 14) This perspective indicates that the apparent differences between apprehended and non-apprehended rule violators is due to the differentiating variables being used as criteria in assigning the deviant label. That is, certain variables are capable of altering the probability of apprehension and conviction for a deviant act independently of the nature of the act itself. Consistent with the labelling perspective, this study will be concerned with the factors that determine arrest and conviction (i.e., being labelled) for a crime. The remainder of this chapter provides a review of litera ture pertaining to apparent differences between apprehended and non-apprehended law violators. The information reviewed stresses the roles of age, sex, race, social class and actual behavior in affecting the probability of detection and conviction. Following the review, a series of hypoth eses will be offered regarding the role of each variable in determining conviction for a crime. The rule violation to be considered in testing these hypotheses will be the illegal act of drinking and driving. Two samples will be used in the analysis: (1) a sample of the general driving public admitting to having driven after consuming alcohol but never having been convicted for this crime, and (2) a sample of convicted drinking drivers. A comprehensive instrument was administered to each group. On the basis of the data collected, it is possible to relate several variables to the chance of arrest and convic tion holding constant the actual crime itself. It is also possible to examine certain assumptions which characterize studies based on examination of convicted criminals. Discretion in the Legal System Assuming implicitly that there is no discretion (that is, opportunity for exercising individual judgment) in the legal system and that the passage of a subject through the system is Mautomatic," traditional studies of the convicted criminal generally overlook the role of the legal system in determining who is to be convicted. However, it appears that there is a good deal of discretion in the legal structure and that passage through the system is not auto matic. For example, court judges are clearly permitted a range of discretion. The judge not only considers the issue of guilt or innocence but he also decides whether the evidence was gathered legally and the case pursued in an acceptable manner. The discretion exercised by a judge, however, is open and public and thus more likely to be within bounds of legal precedent. Even if discretion is improperly applied it is open for appeal to a higher court. District attorneys also have discretionary power in their relations with alleged legal violators. However, many of their acts are not open to public scrutiny. On the basis of his judgment, the district attorney can decide whether to prosecute an offender, as well as the nature of the charge and the size of bail. Newman (1956) has pointed out that the district attorney has broad discretion in plea bargain ing. Most legal violators do not go into court with a plea of not guilty. Instead, before a court appearance, the alleged offender or his counsel often meet with the district attorney and bargain for a reduced charge or lighter sen tence in exchange for a guilty plea. This procedure is defended by lawyers and judges as necessary to expedite the operation of overcrowded courts. The police officer is the legal agent via vAiich nearly all defendants enter the legal system. Technically, the police have no discretion in arresting a person since almost all states have a full enforcement statute which specifies that police will arrest everyone observed commit ting a crime and that they will report to superiors anyone suspected of violating any law. But this expectation is impossible to meet. Both LeFave (1965) and Goldstein (1969) have noted that police cannot possibly arrest all law violators. Limited personnel and resources, legal limitations on police practices, ambiguous laws, and public pressure all operate to force the development of informal priorities in law enforcement. It is clear, then, that there are several decision points within the legal system where disposition can be based on considerations other than those directly involved with the crime in question. Factors Related to Arrest and Conviction Socioeconomic Status Socioeconomic status is one variable outside of the criminal act which is often considered capable of predicting chance of apprehension and conviction for a crime. For example, Glaser (1960) states that ”. . .it has been well established that official agencies take a more punitive attitude toward misbehavior by low status youths than towards the same behavior in higher status youths” (Glaser, in Ohlin and Cloward, 1960: 12). Research relating class position on an alleged offender to the action taken by the legal system will now be examined. Official data on juvenile delinquency have consist ently shown that lower class youths are approximately five times more likely to commit a delinquent act than are middle and upper class youths (see Gold, 1966). In contrast to this, there are several hidden deviance studies, focusing on the undetected delinquency of juveniles, which indicate that the relation between class and delinquency is less than that shown by official data.1 In one of the first empirically rigorous studies, Short and Nye (1958) administered a questionnaire to two samples of high school youths. The subjects were asked which of a series of delinquent acts they had performed. The same instrument was administered to a group of institutionalized delinquents. Holding constant both sex and age, the investigators found few differences in delinquency between class group (interestingly, the differ ences that were found indicated high rates of delinquency among high status youths as often as high rates of delin quency among those of low status). A number of other investigators using the self- report technique have reported similar findings. For example, Empey and Erickson (1966) found that the number of violations differed little between status groups. However, they found in relating types of crime to status that the most serious violations were more likely to be committed by higher status boys. Vaz (1966), Voss (1966), and Dentler and Monroe (1961), all using self-report instruments, found that all status groups perform many more delinquent acts than are ever apprehended. None of these studies report relationships between class position and delinquency that approach the relationships reported in official information. Other types of studies also suggest that lower status persons are more likely to be convicted for a criminal action. Piliavin and Briar (1964) found in a participant observation study of a police juvenile squad that officers have a great deal of discretion in making arrests. Given this discretion, the criteria for arrest used on the street appear to be such things as demeanor, manners, show of respect, and attire. The youth who is respectful and not dressed like a "hoodlum" is less likely to be arrested and taken in than one who is belligerent, hostile, or disrespectful. These characteristics often reflect status groups. Higher status youths by their appearance and their ability to manipulate authority can therefore escape legal detention while lower status youths are more likely to be arrested. Goldman (1963), in investi gating how youths are selected for court appearances, comes to the same basic conclusion as Piliavin and Briar. He states that "... defiance on the part of a boy will lead to juvenile court faster than anything else" (Goldman, 1963: 106). Winick (1964), studying narcotic addiction among physicians, found that while all 98 doctors in his sample had run afoul of the law, only one had been charged with a crime and only a few had been arrested at all. Winick notes that all were "treated relatively kindly" by the law. Discovery of this violation could have led to serious fines and/or jail sentences, but in no case did this occur. Presumably, if persons of lesser status had been detected committing this offense they would have received much more severe penalties. Winick suggests that physicians were accorded this leniency because of their status and because punishment of another type was felt to be sufficient (for example, suspension of the license to practice and the public embarrassment). Another component of class position is income, which has been shown to be important in accounting for arrest and conviction. Patricia Wald (1967), after review ing relevant research, concludes that "... the poor are arrested more often, convicted more frequently, sentenced more harshly, and rehabilitated less successfully than the rest of society" (Wald, 1967: 58 in Dinitz and Dynes, 1968). Nagel (1966) found considerable differences in the treatment of indigents and non-indigents in the legal system. To illustrate, 34 per cent of the indigents held for felonious assault do not get preliminary hearings while only 21 per cent of the non-indigents failed to get the preliminary hearing. The preliminary hearing prevents undue delay before a trial and, therefore, is an important safeguard in the legal system. Nagel also found that indigents were more likely to be found guilty and less likely to be recommended for probation by a probation officer or granted probation or a suspended sentence by the judge. In fact most research relating class position to chance of apprehension and conviction clearly suggests that lower class persons are at a considerable disadvantage. However, there are exceptions to this conclusion. Recently, Terry (1967) has argued that clear evidence for differential, treatment of classes by the legal system is not extensive. In a detailed study, he correlated class position with the severity of response by each of three legal agencies: the police, the court, and the probation department; he found virtually a zero correlation in every case. The correlation between individual class position and severity of police disposition was -.04 (tau). Correlating class position with the severity of the action taken by the probation department yielded a coefficient of -.02 (tau). The correlation between severity of court action and class was -.09, and this decreased to -.02 vhen the number of previous offenses was held constant. Terry interprets these findings as indi cating that there is very little discrimination on the basis of class in the legal system. His research, however, is vulnerable at the point of entry into the system. He depended on official data, which may serve to filter out the higher class potential entrant. That is, he did not have information on those the police did not pick up. He argues that this is not a problem because most police action is based on citizen complaint. Citizens as well as police, however, may be guilty of discrimination on the basis of the class of the alleged offender. 11 Race Race is frequently mentioned as a variable which is relevant in predicting arrest and conviction for a criminal act. Clinard (1963) states that M. . . it is a generally established fact that Negroes as well as Spanish-speaking peoples, on the whole, are arrested, tried, convicted, and returned to prison more often than others who commit comparable offenses" (Clinard, 1963: 550-551) While there is evidence that police are prejudiced against blacks (see, for example, Skolnick, 1966, and Bayley and Mendelsohn, 1969), the important point in this report is to show that race actually makes a difference in the probability of arrest and conviction apart from its relation to actual criminal behavior. Similar to the find ings based on class position, it appears that race is correlated with official criminality. For example, Gould (1969) found that the correlation (rho) between race and the official delinquency rate is .46, blacks being more likely to be officially delinquent than are whites. In contrast, responses to a self-report questionnaire given to a sample of black and white youths indicated that blacks were only slightly more likely than vdiites to be "actually" delinquent. Therefore, race is a better predictor of official delinquency than of self-reported delinquency. If race affects the probability of performing a crime, and also affects the probability of arrest and conviction, then 12 j the partial correlation between race and official delin quency, holding admitted delinquency constant, should be near zero. However, Gould found a correlation between race and official delinquency of .48 for those with high self- reported delinquency, and .46 for those with low self- reported delinquency. Apparently actual delinquency does not intervene between race and official delinquency. This suggests that race has an independent influence on chance of arrest and conviction apart from its influence on the actual commission of an illegal act. If so, we would expect differences to occur between black and white offi cial criminals. The following studies indicate that this is, in fact, the case. Axelrad (1952), comparing black and white official delinquents, found that black youths were committed younger, for less serious offenses, and with fewer previous court appearances. Wolfgang (1958) studied the outcomes in Philadelphia courts over the interval 1948-1952. He found that of those receiving a court trial, 81 per cent of the blacks were found guilty while only 62 per cent of the whites were found guilty. For each specific offense, blacks received more severe sentences despite the fact that they had been more often provoked by their victims and were less likely to possess a previous police record. After reviewing a substantial amount of literature 13 relating to the blacks' experience in the legal system, Pettigrew (1964) concludes: When compared with white defendants, the accused Negro typically has less access to bail, astute legal counsel, cash fines rather than imprisonment, appeals, and other legal protection. He is more likely to be poorly educated and unaware of his full rights. And in many communities he must face a "white man's court" . . . Such conditions set the stage for impaired justice, even when honest efforts are made by officials to prevent bald racial discrimination. (Pettigrew, 1964: 138) Sex In investigating factors which affect the chance of legal apprehension and conviction the variables class and race have received the major share of attention. There has also been some effort to determine the role of sex in altering the probability of conviction. The general opin ion is that females receive less severe treatment through out the legal process. Reckless, for example, comments that . . . citizens are willing to report the behavior of males much more readily than that of females. The police are supposed to be much more lenient in their arrests of females. Judicial processes in America are supposed to be very much more lenient with women than with men. Consequently female offenders have a much better chance than male offenders of not being reported, of not being arrested, and of dropping out of the judicial process. . . . (Reckless, 1961: 37) In the same vein, Block and Geis (1962) suggest that the very high sex ratio among convicted offenders reflects differential court action more than actual 14 2 differential offense behavior. Official data on crimes indicate, as expected, that men are far more likely to be convicted of a crime than are women. The ratio of males to females is, overall, approxi mately five to one. Of course, certain types of crimes differ greatly from this figure. For example, the crime of prostitution is one for which nearly all convicted offenders are women. Nye and Short (1958), using self-reported data, found that while a higher proportion of boys admit to com mitting nearly all offenses, the "... five to one ratio reported by the Children's Bureau is not found in many cases, suggesting a bias in under-reporting female delin quency on the part of official data" (Nye and Short, 1958: 300). They go on to indicate that for certain crimes girls are more likely than are boys to be brought into court. These crimes are usually sex-related and presumably reflect America's double standard regarding sexual activity. Clark and Haurek (1966), in a more extensive analy sis of differences by sex in juvenile delinquency, also conclude that the actual sex ratio in delinquency is lower than the official ratio (vhich is approximately 5:1). Overall they find a sex ratio of 2.3:1 among the officially non-delinquent youths in their sample. Considering only chronic offenders (those who admit to committing an offense four or more times), the ratio increases to 3.7:1. This is still less than the official ratio. Examining the single offense category of major theft, the sex ratio among admitted offenders is 1.4:1; this only increases to 1.9:1 when chronic offenders alone are considered. Clark ^ d Haurek conclude that official data yield a much higher sex ratio than may actually exist. They suggest that differences by sex in the reporting, enforcement, and court handling of juveniles may account for at least part of these differences. Age In contrast to sex, a research on deviant behavior has often used age as a control variable. Juvenile delin quency is so commonly separated from adult criminality that these two are taught as separate courses in many universi ties. However, very little research has been done using age as an independent variable in predicting the probability of legal conviction. According to Shulman (1961), the general pattern of official criminality by age shows a curve which rises rapidly from childhood through the late teens, then remains at a plateau into the early thirties, finally declining slowly to old age. Clark and Haurek (1966) indicate that there is no clear relationship between age and delinquency. The patterns of age and crime vary greatly by sex and type of 16 offense. They find that offenses associated with the role of child decrease through adolescence. Those offenses that are judged to be less deviant for an adult than for a child increase in frequency throughout adolescence. Further, those offenses that involve aggression and daring increase in frequency to mid-adolescence and then decrease. Apparently there is no published study using data on adults which relates age to the chance of conviction for a crime. Since age is an important social factor, it is reasonable to explore the role of age in legal treatment. Offense Behavior While the variables mentioned previously are to varying degrees predictive of legal conviction, still the actual criminal activity may be of prime importance. That is, even though some demographic characteristic is associ ated with the chance of conviction, it may still be the frequency of commission and the seriousness of the act that are the most important determinants of apprehension and conviction. A number of investigators, after examining other variables, have shown that both frequency and seriousness of violation are predictive of conviction. • * * * ’ Empey and Erickson (1966), in studying actual delinquent behavior, divided their population into four groups using official criteria: non-delinquents, one-time offenders, serious offenders on probation, and incarcerated 171 offenders. They then compared these groups in terms of the actual (admitted) frequency of delinquent acts. They conclude that "... boys whose official position placed them on probation in an institution, reported having been far more delinquent. On many offenses their reported violations exceeded those of non- and one-time offenders by thousands" (Empey and Erickson, 1966: 550). Short and Nye (1958) report in their study of hidden delinquency that for both boys and girls . . . the offenses for which (they) are most often arrested are generally those which they most often admit committing. . . . Also, comparing institu tionalized delinquents with a sample of never- convicted youths, the incarcerated youths admit committing nearly all offenses more frequently. Moreover, the convicted delinquents . . . ranked significantly higher, in terms of seriousness of involvement in delinquent behavior than do high school students. (Short and Nye, 1958: 301) In their pioneering study of undetected delinquency, Murphey et. al. (1946) show that comparing a group of offi cial non-delinquents with a group of official delinquents, the violations of the official delinquents were both more frequent and more serious. The median number of violations (over the period of observation) was 79 for the delinquents and 30 for the non-delinquents. The delinquents also averaged 53 minor violations and six serious infractions of the law while the non-delinquents had averaged only 20 minor violations and no serious offenses (Murphy et al., 1946). 18 Summary of Study Results Recent research has shown that official data regard ing criminal behavior is not representative of all criminal activity. One reason for the discrepancy is suggested by the labelling perspective, namely, that more than the performance of a criminal act is involved in being appre hended and convicted (labelled) for a crime. Such charac teristics as class, race, age, and sex may be relevant. Review of the literature supports the contention that each of these, with the exception of age, has an effect on the probability of conviction independent of any relation to the actual criminal behavior. It is also indicated that both the seriousness and the frequency of criminal behavior influence the chance of conviction. In the following chapter, a set of hypotheses is presented relating each of these variables to conviction for the crime of driving after drinking. The Crime of Drinking and Driving The dependent variable in this study is the crime of driving while under the influence of alcohol. Most research involving this violation has concentrated on the relation between drinking drivers and accidents. Despite the recent attention given to this relation, there is a paucity of simple descriptive information on convicted drunk drivers and even less on the drinking driver who is never 19 I caught. As interest in the drunk driver has increased, there have been a few attempts to describe this type of offender. As the hidden deviance studies suggest, the description of the drinking driver varies greatly by the type of data used. With official data the drinking driver appears to differ from the never-convicted driver by race, class, marital status, and degree of adjustment. On the other hand, the never-convicted drinking driver seems not to differ greatly from the general driving population . Recently Merton Hyman (1968) reported the results of a study of officially convicted drunk drivers in two different police jurisdictions (the midwest and the far west). Hyman concludes that those convicted for drinking and driving are quite different from the general driving public. These differences include both drinking habits and a series of socioeconomic variables. He found that people from poor census tracts with a high proportion of minority group members and a low proportion of home ownership are more likely to be convicted for driving while intoxicated. Coldwell and Grant (1965), in a study of convicted drinking drivers, found that the majority of their subjects were blue collar workers between the ages of 25 and 54. Similarly, Goldberg (1955), in a study of Swedish drinking drivers using all adult male Swedish drivers as a control group, found that convicted drunk drivers were more likely 20 to be in the age group 25-54, to be divorced or married, to be clients of social welfare agencies, to be legally defined alcoholics, and to have been previously arrested for drunkenness. Several studies of drinking drivers involved in accidents have found an over-representation of single men; persons with a background including broken homes and parents who were problem drinkers (Barmack and Payne, 1961); crim inal arrests and known non-support of family (Waller, 1961); and recognizable psychiatric illness exclusive of alcohol ism (Seize, Payne, Gifford, and Kelly, 1963). In sumnary, it appears from the above studies that convicted and accident-involved drinking drivers are of lower status and less stable than the general driving population as measured by several social and demographic characteristics. Studies using self-reports support a quite different conclusion regarding the drinking driver. Harold Mulford (1968) studied the drinking driver using a random sample of all adult drivers in the state of Iowa. These subjects were divided into several categories in terms of the probability that they might appear on the highway after having consumed a measurable amount of alcohol. The distri bution of these categories by demographic characteristics shows a different picture than do studies based on convicted drinking drivers. For all those who are employed, the 21 distribution of the groups by occupational category shows that 52.9 per cent of all non-drinkers are in the lowest status occupational group while only 32.9 per cent are in the highest status occupational group. Conversely, for the highest probability group (those who admit to driving after consuming at least three drinks) only 33 per cent are in the lowest prestige occupational group and 46 per cent are in the highest prestige occupations. Thus the members of the admitted drinking driver group are more likely to hold higher prestige occupations, while abstainers are more likely to be in the least prestigious occupations. The same basic pattern emerged for educational differences. Those persons with more education were more likely to appear on the highway as a drinking driver than those with less education. Finally, there appeared to be no differences among drinking driver categories by marital status. In contrast to Hyman's (1969) findings, these results suggest that the drinking driver is, if anything, of higher status than the average driving citizen. Another study of self-reports by Cosper and Mozersky (1968) generally supports Mulford's findings. They studied the drinking driver using non-random samples of drivers in two cities. In the first city, the percentage of admitted drinking drivers varied from 65 per cent for those who completed college to 15 per cent of those attending only grade school (grades 1-9). The same pattern occurs with 22 regard to occupational prestige; the higher status occupa tions tend to show higher proportions of drinking drivers. The authors note that the bulk of these differences may be due to differences in the population at risk. The lower class, for example, may have fewer drinking drivers because it has fewer people who both drink and drive at all; only those vho drink at all and drive at all are ever potential self-reporters of drinking and driving. If, according to Cosper and Mozersky, the population of each status category is restricted to those who can possibly drive under the influence (the population at risk) the differences found in the total population disappear. This same factor may account for Mulford's finding that drinking drivers tend to be of higher status. Cosper and Mozersky report the same finding for the second city, where there were few differences between the admitted drinking drivers and the general population. The few existing differences suggest that higher status respond ents are more likely to drive after drinking, which is consistent with the above findings. In summary, the crime of driving while under the influence of alcohol appears to be an acceptable criterion variable in testing the role of a number of factors in predicting conviction. This is so because the same differences between official and non-official offenders apparently occurs for drinking drivers as occurs for most other types of crime. FOOTNOTES The early studies which looked at undetected devi ance include Murphey et al. , 1946; Porterfield, 1946; and Wallerstein and Wyle , " ”1957. 2 The sex ratio is the number of males per female. Thus a sex ratio that is greater than 1 indicates more males than females. REFERENCES Axelrad, Sidney. 1952 "Negro and White Male Institutionalized Delin quents." American Journal of Sociology (May): 569-74. Barmack, J. E., and D. E. Payne. 1961 "Injury Producing Private Motor Vehicle Accidents among Airmen." Bulletin of the Highway Research Board 285: 1-22. Bayley, David H., and Harold Mendelsohn. 1969 Minorities and the Police: Confrontation in America. New York: The Free Press. Block, Herbert, and Gilbert Geis. 1962 Man, Crime, and Society. New York: Random House. Bordua, David. 1960 Sociological Theories and Their Implications for Juvenile Delinquency. Washington, D.C.: United States Department of Health, Education and Welfare Children's Bureau. Chevigny, Paul. 1969 Police Power: Police Abuses in New York City. New York: Pantheon Books. Clark, John P., and Edward Haurek. 1966 "Age and Sex Roles of Adolescents and Their Involvement in Misconduct: A Reappraisal." Sociology and Social Research 50 (July): 496-508. Clinard, Marshall, B. 1963 Sociology of Deviant Behavior (rev. ed.). New York: Holt, Rinehart and Winston. Cloward, Richard, and Lloyd E. Ohlin. 1960 Delinquency and Opportunity. Glencoe, Illinois: The Free Press. Cohen, Albert K. 1955 Delinquent Boys: The Culture of the Gang. Glencoe, Illinois: The Free Press. Cohen, Albert K. 1966 Deviance and Control. Engleyrood Cliffs, N.J.: Prentice-Hall, Inc. 26 : Coldwell, B. B., and G. L. Grant. 1963 "The Drinking Driver.” Proceedings of the Third International Conference on Alcohol and Traffic Safety: 54-60. Cosper, Ronald, and Kenneth Mozersky. 1968 "Social Correlates of Drinking and Driving.” Quarterly Journal of Studies on Alcohol. Supple ment No. 4 (May) 58-117. Cressey, Donald R., and David A. Ward. 1969 Delinquency, Crime and Social Process. New York: Harper and Row. Dentler, Robert A., and Lawrence J. Monroe. 1961 "Social Correlates of Early Adolescent Theft." American Sociological Review 26 (October): 733-743. Empey, Lamar, and Maynard Erickson. 1966 "Hidden Deviance and Social Status." Social Forces 44 (June): 546-54. Erickson, Maynard, and Lamar Empey. 1963 "Court Records, Undetected Delinquency and Decision-making." The Journal of Criminal Law, Criminology and Police Science 54 (December): 456-69. Gold, Martin. 1966 "Undetected Delinquent Behavior." Journal of Research in Crime and Delinquency 3 (January): 27-46. Gold, Martin. 1963 "Status Forces in Delinquent Boys." Ann Arbor: University of Michigan, Institute for Social Research. Goldberg, L. 1955 "Drunken Drivers in Sweden." Proceedings of the Second International Conference on Alcohol and Traffic Safety: 112-127. Goldman, Nathan. 1963 The Differential Selection of Juvenile Offenders for Court Appearances, National Council on Crime and Delinquency. 27 Goldstein, Joseph 1969 Police Discretion Not to Invoke the Criminal Process: Low Visibility Decisions in the Adminis tration of Justice." In Cressey and Ward, 1964: 166-84. Gould, LeRoy C. 1969 "Who Defines Delinquency: A Comparison of Self- Reported and Officially Reported Indices of Delinquency for Three Racial Groups." Social Forces 16 (Winter): 325-36. Green, Edward. 1961 Judicial Attitudes in Sentencing. New York: St. Martin's Press. Hyman, Merton M. 1969 "The Social Characteristics of Persons Arrested for Driving While Intoxicated." Quarterly Journal of Studies on Alcohol. Supplement No. 4 (May): 138-177. Johnson, Guy. 1941 "The Negro and Crime." The Annals of the American Academy of Political and Social Science 217: 93- 104. LaFave, Wayne R. 1969 "Noninvocation of the Criminal Law by Police." In Cressey and Ward, 1969: 185-208. Merton, Robert. 1938 "Social Structure and Anomie." American Sociolog ical Review 3 (October): 672-82. Miller, Walter B. 1958 "Lower Class Culture as a Generating Milieu of Gang Delinquency." The Journal of Social Issues 14 (October): 5-19. Morris, Ruth R. 1964 "Female Delinquency and Relational Problems." Social Forces (October): 82-89. Murphy, Fred J., Mary M. Shirley, and Helen L. Witmer. 1946 "The Incidence of Hidden Delinquency." The American Journal of Ortho-psychiatry 16 (October): 686-96. 28 Newman, Donald J. 1956 "Pleading Guilty for Considerations: A Study of Bargain Justice." Journal of Criminal Law, Crime and Police Science 46 (March-April): 780-90. Pettigrew, Thomas F. 1964 A Profile of the Negro American. Princeton, N.J.: D. Van Nostrand. Piliavin, Irving, and Scott Briar. 1964 "Police Encounters with Juveniles." American Journal of Sociology 70 (September): 206-214. Reckless, Walter. 1961 The Crime Problem (third ed.). New York: Appleton-Century-Crofts. Reiss, Albert J., Jr., and Albert L. Rhodes. 1961 "The Distribution of Juvenile Delinquency in the Social Class Structure." American Journal of Sociology 26 (October): 720-32. Shulman, Harry M. 1961 Juvenile Delinquency in American Society. New York: Harper and Bros. Terry, Robert M. 1967 "Discrimination in the Handling of Juvenile Offenders by Social Control Agencies." Journal of Research in Crime and Delinquency 4 (July): 218-30. Vaz, Edmund. 1969 "Juvenile Delinquency in the Middle Class Youth Culture." In Cressey and Ward, 1969. Voss, Harwin L. 1966 "Socio-Economic Status and Reported Delinquent Behavior." Social Problems 13 (Winter): 314-24. Wald, Patricia. 1967 "Poverty and Criminal Justice." The President's Commission on Law-Enforcement and the Administra tion of Justice, Appendix C (in Dinitz and Dynes). Wallerstein, James S., and Clement J. Wyle. 1946 "Our Law-Abiding Law-breakers. Probation 25 (March-April): 107-12. Werthman, Carl, and Irving Piliavin. 1967 "Gang Members and the Police." In David Bordua The Police: Six Sociological Essays. New York John Wiley and Sons: 56-98. Wolfgang, Marvin. 1958 Patterns in Criminal Homocide. Philadelphia: University of Pennsylvania Press. CHAPTER II RATIONALES AND HYPOTHESES According to official statistics, as indicated in Chapter I, crime is among certain groups in the population. Specific aggregates have a rate of criminal convictions that is significantly higher than the average for the population. Blacks, for example, are more likely to be convicted for many types of crime than are Caucasians. Lower class people show a higher official crime rate than those in the upper class. Males and perhaps youths are convicted in higher proportions than females and the aged. The initial concern of this study is whether or not blacks and Chicanos, the poor, males, and youths are overrepre sented in the convicted drinking driving population. The literature reviewed in Chapter I suggested that, in fact, drinking drivers are more likely to be members of one (or more) of these aggregates. If so, the following hypotheses should be supported: 1. Blacks and Chicanos are more likely than Caucasians to be convicted for driving after drinking. 2. Social class is inversely related to the chance of conviction for driving after drinking. 30 3. Males are more likely than females to be convicted for driving after drinking. 4. Age is inversely related to the chance of conviction for driving after drinking. These hypotheses may be supported by two opposing rationales. First, people in these categories may actually be more deviant. They are convicted more often because they engage in drinking and driving more often than others, and they are more inebriated than others when they drive after drinking. The fact that they are overrepresented in official conviction data indicates that the legal system is doing its job by apprehending and convicting the "worst" offenders. The second rationale asserts that discrimination is the basic cause of overrepresentation of certain groups in the convicted category and assumes that conviction for a crime is not related to the actual frequency or severity of the law violation. Instead of law-violating behavior determining the chance of conviction, characteristics such as race, class, sex, or age determine whether or not a law violator will be convicted. Ideally "justice" means that all alleged offenders are given the same rights; conviction is determined only by reference to facts directly related to the alleged crime; and the severity of the penalty is increased with the "severity" of the crime and the pre scribed legal limits. Any measurable departure from these criteria of justice is deemed discrimination (that is, any time unequal treatment is accorded those with equal crimes, discrimination has occurred). Note that "discrimination" as it is used here is only a label to be applied to a particular set of empirical findings. The condition of discrimination may arise from a variety of sources. These will be discussed later in this chapter. In this study, race, social class, sex, and age are examined regarding their relation to conviction. These variables are included because they are significant social categories which differentiate segments of the population which actually behave differently and which are reacted to differentially by others. Regarding differential reac tions , these variables define continua which carry evalua tions of superior and inferior. For example, it has been clearly shown in this country that the rich are evaluated as superior to the poor (see, for example, Tumin, 1967: 26-27). This evaluation, furthermore, is often translated into differential behavior. Most of the research studies cited in Chapter I refer to the idea of preferential treat ment extended to those evaluated as superior. According to the research, race, class, sex, and age are four such criteria for preferential or harsh treatment. There is a third possible explanation for the over- representation of certain aggregates in the convicted category. It is perhaps most reasonable to posit that conviction for drinking and driving is determined by both the nature of the violation and by certain other character istics of the alleged offender. Examination of this general proposition will enable us to weigh the predictive utility of the two rationales outlined above and to deter mine whether social characteristics and behavior combine to produce conviction. The evidence reviewed in Chapter I tends to favor this approach. Certain characteristics of the offender were found to be directly related to convic tion, and the nature of the law-violation was also influ ential in determining conviction. In other words, member ship of an individual in a social category affects the probability of commission of an illegal act, so that some categories have higher rates of actual illegal behavior than others. Once apprehended, these same characteristics which define membership in a given social category (race, class, sex, age) may eventuate in differential treatment by the legal system, so that the proportion convicted, of those apprehended, will also be higher than the proportion convicted (among those apprehended) among members of more favored categories. If race, class, sex, and age have a direct effect on the chance of conviction and on actual illegal behavior, they should be associated with conviction when the effect of actual illegal behavior is statistically "held constant." Assuming this model is correct and the effect of all variables exogenous to the model are randomized, controlling 34 ' for actual illegal behavior should minimize the effect of actual behavior on conviction. The correlation between j individual social characteristics and conviction will remain significantly greater than zero, but lower than the zero-order correlation (within probability limits). This would constitute evidence for the direct effect of member ship in different social categories on conviction for drinking and driving. If there is no direct connection between membership in certain social categories and convic tion (and assuming we have the correct model, the correct level of analysis, and randomized all other factors), then when actual illegal behavior is statistically controlled the correlation between conviction and characteristics such as race, social class, sex, and age should drop. In this case, the only link between membership in these categories and conviction would have been severed. Such evidence would constitute support for the argument that rates of conviction are basically determined by rates of illegal activity. Assuming that race, social class, sex, and age have some direct effect on the probability of conviction inde pendent of actual illegal behavior, hypotheses one through four can be changed to read: 5. Controlling for actual drinking and driving behavior, blacks and Chicanos are more likely than are Caucasians to be convicted for driving after drinking. 35 6. Controlling for actual drinking and driving behavior, social class is inversely related to the chance of conviction for driving after drinking. 7. Controlling for actual drinking and driving behavior, males are more likely than females to be convicted for driving after drinking. 8. Controlling for actual drinking and driving behavior, age is inversely related to the chance of conviction for driving after drinking. These hypotheses suggest only that certain variables alter the chance of conviction independent of actual drink ing and driving behavior. It is often assumed that when discrimination is found it is the result of legal agents acting on their prejudicial attitudes. This assumption may be challenged in several ways; first, the empirical demon stration that police are prejudicial against certain groups does not logically lead one to conclude that these attitudes are acted out to produce discrimination. It remains to be shown that legal officials clearly do act in a way to cause discrimination. Second, even if prejudice in the legal system does produce discrimination, this does not eliminate the possibility of other sources of discrimination. That is, factors other than prejudice could lead to the over conviction of certain groups. Specifically, it is suggested that there are organizational and cultural factors which differentially affect the chance that members of certain categories will be convicted of drinking and driving. 36 Stinchcombe (1963), for example, shows how police practices in arrests and convictions in court are affected by the social location of crime. Specifically, he shows that the proportion of crimes cleared by arrest and convic tion rates vary between those crimes committed in public and those committed in private places. If this aspect of social structure affects the operations of the law, it seems tenable to argue that other features of the social structure may also have an impact on the operation of the legal system. Police density is one such organizational factor which is more closely related to the specific crime of drinking and driving. Most arrests for drinking and driving (in this study about 75 per cent) occur as the result of police observing the offense. Most other arrests occur after an accident. Therefore, the ratio of convic tions per violation should vary directly with the density of police in a given area; those who live in areas with a high police density are more likely to be observed when driving after drinking than those who live in low police density areas. The assignment of police to patrol areas is determined by a formula vdiich includes the number of crimes known to the police in an area during the previous year. Thus areas with high crime rates tend to have a high police density. The distribution of wealth in the population is another organizational factor which may alter the chance of conviction. Attorneys are not equally skilled. The rich are able to afford very able counsel, while the poor settle, at best, for court-appointed legal aid. Not only are the rich able to obtain more skilled legal assistance, they are also able to purchase a larger amount of an attorney's time for the preparation of a legal defense and for the pursuit of an appeal. In drinking and driving cases, legal aid lawyers will rarely appeal a conviction. This increases the overall chance of conviction for the poor. Another factor which may affect the chance of conviction is the response to alcohol. There is some evidence which suggests that there are social and cultural factors which influence how one behaves after drinking (see MacAndrew and Edgerton, 1969). If this evidence is correct, there may be subcultural differences in the response to alcoholic inebriation. In the middle class, as the argument goes, there is an emphasis on control and discipline in social intercourse. The respected middle class person learns and is pressured to behave "properly" even after consuming a good deal of alcohol. This may extend to driv ing behavior where the middle class person can appear to be in control, thus not attracting police attention. Lower class persons and blacks, for example, may not have such prescriptions regarding "proper" behavior. Therefore, these 38 people may be more likely to be expressive and demonstrative when drinking. If this sort of behavior is continued when driving, it can increase the chance of being observed and stopped by police. Factors such as police density, income distribution and response to alcohol potentially produce discrimination, yet they are not directly attributable to biases or preju dices of those in the legal system. In other words, the poverty that prevents the poor from obtaining skilled lawyers is not necessarily the product of official bias. While the source of discrimination may be unimportant to the person suffering such discrimination, it can make real difference in understanding how the legal institution works. A distinction must be made between extra-legal and intra- legal sources of discrimination. This distinction may help clarify some of the contradictory findings regarding discrimination by the legal system. Many who argued that discrimination is produced by the legal system have been measuring the "output" of the legal system. They provide no information about the proportions of various groups entering the legal system. For example, Terry's (1967) findings of little or no discrimination against blacks and the poor is the result of attempting to measure discrimina tion created within the legal structure.^" Yet it is possible for a specific category to experience a higher ratio of convictions per violation than another group in the absence of overt discrimination by legal authorities. While the data at hand do not allow for the direct separation of extra-legal from intra-legal sources of discrimination, it is possible to examine more closely the internal operation of the legal system. It is possible to assess the effect of the social structure indirectly by removing that part of the differen tial conviction rate which can be attributed to the operation of the legal system. To do this it is hypothe sized, as a point of reference, that each legal agent (policeman, district attorney, and judge) is impartial in applying his discretion when making decisions concerning the drinking driver. If no association between social characteristics and legal decisions appears, then we would be justified in concluding that there is no intra-legal source of differential convictions. It would follow that the existing overrepresentation of categories must be the result of extra-legal factors. Conversely, if the hypoth eses are not supported and one or more of the legal agents does show substantial favoritism, this would decrease the relative importance of social structure (extra-legal) factors as variables producing discrimination. While the hypothesis of total impartiality may be unrealistic, it is not unreasonable to suggest that the level of partiality is lower than that necessary to account for all the overrepresentation. This is especially true jregarding the crime of drinking and driving. This law violation is relatively unambiguous. Guilt is determined by a chemical test with a clear boundary below which the i [person is not guilty. Conviction is relatively easy when j (the level of alcohol is high; it is difficult for a defense lawyer to question the veracity of a breathalizer. Because | of this lack of ambiguity, the decision maker is not forced ; ; | to rely on his own judgment as much as with other law violations. i The first contact with the law by a potential ! convicted drinking driver is with the police officer. In terms of the severity of his response, the officer has several options; he can warn the subject and release him; he can give a citation and then release him; or he can arrest the subject and take him into custody. By applying these options differentially to different social categories, police could contribute to the over-conviction of members of these categories. If this does occur, social characteris tics (such as race, class, sex, and age) should be related to the chance of release once stopped by police after drinking. A high correlation may only indicate that certain groups are released less often because they offend more severely. However, the effect of actual drinking and driving behavior can be controlled by statistically holding (it constant. I ; Hypotheses nine through twelve can be derived from ] the assumption that organizational and cultural factors ' make the deviance of some more "observable1 1 from the deviance of others, but that within the pool of equally observable violators police are not biased in picking : offenders for arrest: 9. Controlling for actual drinking and driving behavior, there is no relation between race and the severity of police treatment. 10. Controlling for actual drinking and driving behavior, there is no relation between social class position and the severity of police treatment. 11. Controlling for actual drinking and driving behavior, there is no relation between sex and the severity of police treatment. 12. Controlling for actual drinking and driving behavior, there is no relation between age and the severity of police treatment. After the initial contact with the police, the second important decision point concerns the charge to be brought by the district attorney. The person stopped for drinking and driving can be charged with driving under the influence of alcohol or he may receive the lesser charge of reckless driving, which can be used when the person violated a traffic law and was found to have been drinking, but was not sufficiently inebriated to be legally "drunk" (the legal definition of drunkenness will be discussed in Chapter III). The penalties for reckless driving are substantially less severe than the penalty for a conviction for drunk driving. If the district attorney uses his 42 discretion to favor certain groups, such groups should show a larger percentage receiving the reduced charge than other, less favored, groups. Again, differences in propor tions getting the reduced charge may only be the result of actual differences in the severity of the offense. This factor can be controlled by holding constant the actual degree of intoxication among offenders. These comments lead to four hypotheses relating social characteristics and receipt of a reduced charge: 13. Controlling for the seriousness of the violation, there is no association between race and the legal charge among those convicted for drinking and driving. 14. Controlling for the seriousness of the violation, there is no association between class position and the legal charge among those convicted for drinking and driving. 15. Controlling for the seriousness of the violation, there is no association between sex and the legal charge among those convicted for drinking and driving. 16. Controlling for the seriousness of the violation, there is no association between age and the legal charge among those convicted for drinking and driving. The final decision point for which data are avail able is the sanction levied by the judge. While this decision comes after conviction and thus cannot directly influence who gets convicted, it may nevertheless indicate attitudes or behavior by the judge which operate before, and influence, the conviction decision. Of course, certain !aggregates may receive stiffer fines or longer sentences because their drinking and driving violations are, on the average, more serious. Additionally, the judge may respond to the total number of past convictions for this crime and give more severe treatment to recidivists. However, the : i effect of both past and present involvement in drinking and j driving can be controlled in order to determine if the i judiciary is favoring certain groups. j I As with police and district attorneys, it is assumed that judges give equally severe sanctions to all j j offenders once the number of past offenses and the serious- ; ness of the present offense have been controlled. Though it is anticipated that there will be some differential sentencing by social characteristics, this is assumed not to be as important as the organizational factors discussed earlier. Four hypotheses relate social characteristics to the penalty for a conviction for drinking and driving: 17. Controlling for the past and present involve ment in drinking and driving, the severity of the penalty for a conviction is not related to the race of the convicted person. 18. Controlling for the past and present involve ment in drinking and driving, the severity of the penalty for a conviction is not related to the class position of the convicted person. 19. Controlling for the past and present involve ment in drinking and driving, the severity of the penalty tor a conviction is not related to the sex of the convicted person. 20. Controlling for the past and present involve- j ment in drinking and driving, the severity of the penalty for a conviction is not related to the age of the convicted person. In general, it is expected that certain groups are ; I |convicted more often than is warranted by their illegal j ; l |drinking and driving behavior. Furthermore, it is antici- j I pated that this discrimination is not primarily the result j of the differential application of the discretion given | i legal agents. Rather, it is argued that there are features of the social structure that have as one consequence the alteration of one's vulnerability to the law. The following chapter will explicate the data to be used in testing the hypotheses. FOOTNOTES It should be pointed out that because one is unable to measure discriminatory behavior by legal offi cials does not mean that it doesn't occur. It is possible that many very subtle actions together operate to alter the chance of conviction. Such subtle differences may be impossible to measure without much more complete data on the operation of the legal system. REFERENCES MacAndrew, Craig, and Robert B. Edgerton. 1969 Drunken Comportment: A Social Explanation. Chicago: Aldine Publishing Co. Stinchcombe, Arthur L. 1963 "Institutions of Privacy in the Determination of Police Administrative Practice." American Journal of Sociology 69 (March): 150-60. Tumin, Melvin M. 1967 Social Stratification: The Forms and Functions of Inequality. Englewood Cliffs, N.J.: Prentice- Hall. CHAPTER III METHODS AND DATA This chapter presents the methods to be used to test the hypotheses developed previously. Specifically it will include information on the questionnaire, the samples of convicted and never-convicted drinking drivers, the validity of the resulting information, the measures of the concepts and the statistical techniques used in the analysis. The Instrument The "Drinking Driver and Traffic Safety Project" was originated by the State of California to facilitate "doing something" about drunk drivers.^- After a review of the literature and several pretests a final questionnaire was developed by a group at the University of Southern 2 California. This forced choice instrument, twenty-two pages long, included questions on general level of health, drinking patterns (where, with whom, how much, when began, what, binges, etc.), demographic-social variables, satis faction with life, and attitudes toward the law. 47 The questionnaire was handed out to respondents i during the five-month period from January through May, 1968. i : i The convicted group was composed of drinking drivers j i processed by four courts in Los Angeles County which together handle approximately 60 per cent of the total volume of drinking and driving cases in the county. These ; ! courts included Division 35 and Division 51 of the Los Angeles Municipal Court, the Van Nuys Municipal Court, and | the East Los Angeles Municipal Court. A random sample of the convicted drinking and driving cases passing through these four courts was collected. The research assistants in each court first obtained each day’s arraignment list. Then all those being arraigned on drinking and driving charges were divided into two groups--first offenders and multiple offenders or recidivists, that is, those with one or more drinking and driving convictions. A table of random numbers was used to select one-third of the first offenders and three-fourths of the recidivists. Part of the overall study involved detailed examination of recidi vists; therefore, this group was over-samples to provide a number sufficient for analysis. A total of 863 question naires were completed by the convicted drinking drivers selected in the four courts. To compensate for the over enumeration of multiple offenders each recidivist was q weighted .44. Thus in all computations where any recidi vists appear each has less weight than each first offender. A group of drivers representing the general driving public was selected from the population seeking initial acquisition or renewal of the drivers license at a local Department of Motor Vehicles (DMV) office from January through May, 1968. There are a large number of DMV offices in the county and some offices denied access to their facilities. In order to obtain a sample of this population of drivers, it was decided to draw a sample of the cooper ating DMV offices. This was done by selecting (not at random) five offices in which the surrounding areas Were generally representative of the entire councy as indicated by Census data. The five offices are Van Nuys, Montebello, Torrance, Long Beach, and Compton. The sampling procedure in each DMV office consisted of selecting for interview the first available applicant each morning. As soon as the first interview was completed the next subject chosen was the person just completing the DMV requirements. In this way five or six persons were interviewed each day. The interviewer also selected a person each time one was picked for an interview. This second person was asked to complete the instrument in questionnaire form.^ Due to the relatively small number of female drinking drivers (in official data), a modified quota sampling technique was used. Only males were selected four days a week with both males and females included on the fifth day. The actual number of completed !questionnaires and interviews taken from DMV respondents j was 1,320. To overcome the undercoverage of women, a weight i factor of five was included in all calculations to make the | proportion of women approximately equal to the original proportion in the DMV offices."* Data from public records were also collected on j each individual in both the court and DMV samples. The I records included the driving history contained in the DMV files and the criminal history in the State of California files. Other possible sources were checked during the pilot study but the relative paucity of data precluded their use. These sources included credit systems, alco holic rehabilitation centers, private and public welfare organizations, and the county hospital. The overall sample consists of 2,183 subjects. However, this study is interested in what determines conviction for drinking and driving. The DMV sample, therefore, was reduced to include only those who admit to driving after drinking. Out of the 1,320 DMV respondents, 545 admitted to consuming two or more drinks and then driving, at least once a year. Thus the sample for analy sis consists of 1,408 drinking drivers; 863 convicted offenders, and 545 offenders who have never been convicted. The weight factors applied to number of convictions and sex alters the marginal totals in tables. After the weights have been included the sample size is 1,196. Validity of the Data | The representativeness of the DMV sample (before j removing the non-drinking drivers) is indirectly measured i by comparing this sample with the population of Los Angeles| County (1960). As shown in Table 1, there are several I differences between the two. The DMV sample has both a greater income and a higher level of education than the county population. In Los Angeles County, the percentage of families making less than $3,000 (12.6 per cent) is morej than twice that for families in the DMV group (5.7 per cent). With regard to education, 20.6 per cent of the people in Los Angeles County and only 13.8 per cent of those in the DMV group have completed no more than one to three years of high school. Further, only 9.8 per cent of the people in Los Angeles County have at least a college degree while 19.5 per cent of those in the DMV sample have such a degree. The DMV group has a larger percentage of Caucasians and a smaller percentage of minority group members. In Los Angeles County the population was 7.6 per cent black while the DMV sample is only 4.1 per cent black. Regarding age, the county population showed larger proportions of the total in older age categories than in the DMV sample. There are several factors besides biased sampling that could account for these differences. First, the data | TABLE 1.--Percentage Distributions of the Total DMV Group } (1968) and the Population of Los Angeles County (1960) | on Age, Race-ethnicity, Family Income, Marital Status, and Education* | Variable Los Angeles County DMV Group Total Yearly Family Income (for 1959) (for 1968) None - $2,999 12.6 5.7 $3,000 - $5,000 25.6 9.7 $6,000 - $8,999 30.3 20.0 $15,000 and over 7.8 19.3 No answer 11.4 Race-Ethnic Grouo Caucasian 80.8 84.3 Blacks 7.6 4.1 Chicanos 9.6 7.1 Other 2.0 3.4 No answer 1.1 Age (males only) 15-19 years 9.3 5.7 20-24 years 8.4 14.8 25-29 years 9.6 13.6 30-34 years 10.6 9.8 35-39 years 11.3 9.6 40-44 years 10.3 11.3 45-49 years 9.4 10.5 50-54 years 7.9 8.5 55-59 years 6.8 6.6 60-64 years 5.4 4.1 65-69 years 4.3 2.6 70 years and over 6.7 2.3 JL Los Angeles County data are from the U.S. Bureau of the Census, 1960 Tract Report for Los Angeles County. 53 TABLE 1--Continued Variable Los Angeles County DMV Group Marital Status (males only) Single 23.5 17.1 Married 69.2 65.9 Widowed 3.1 1.4 Divorced 4.2 4.1 No answer 11.4 Education (years completed) (those aged 25 & over) Grades 10-12 20.6 13.8 High School Diploma 28.9 23.2 1-3 Years College 14.4 33.0 College Degree and over 9.8 19.5 No answer ! 54 i |for Los Angeles County were collected by the Census Bureau in 1960 while the data on the DMV sample were gathered in 1968. Some of the differences, especially in income and i jpossibly in education, are probably due to this discrepancy. i Second, the differences in income and education might also be a result of having a smaller percentage of blacks in the j ! DMV group. That is, because blacks are poorer than Caucasians on the average, a smaller proportion of blacks ! may result in a larger proportion of people in the higher jincome and educational categories. A third important ;source of difference between the DMV sample and the popula tion of the county is that everyone in the DMV group either jhas or is applying for a drivers license, while the county i i population includes many people who do not have such a license. Thus, the small percentage of older people in the DMV group may well be due to the small number of people in these age groups having drivers licenses or continuing to I apply for new ones. The fact that everyone in the DMV sample has or is (getting a drivers license may also partially account for |the underrepresentation of blacks and poor people in the IDMV sample. Hyman (1969) ranked census tracts by socio- |economic status in two cities and found that 20-25 per cent imore of the households in the higher status tracts owned (cars than in the lower status tracts. Consequently, the i lower income and education categories, which include a I 55 large proportion of blacks, are less likely to own cars land, hence, perhaps less likely to obtain drivers licenses. | If so, then there should be fewer black, poor, and low | education persons in the DMV group than in the total county j population. The data are consistent with this expectation. I In suranary, the differences between a sample of j drivers and the total population of the county are not I J extensive. The differences that do exist are, at least in i i part, accounted for by the differences in the period of i data collection and by the fact that all those in the |control group are drivers while many in the county popula- Ition are not. i j The validity of the data on both convicted and !never-convicted drivers may be especially questionable j considering the circumstances under \diich they were gathered and the kind of questions being asked. Although it is not possible to measure validity directly, an indirect indicator is the extent to which the data are internally consistent. Several cross-tabulations of the total sample j were made to locate inconsistent responses. First, the admitted frequency of drinking and driving for both DMV and convicted drinking driver samples taken together was cross- j tabulated with the reported frequency of drinking. i |Logically it is impossible for a respondent to drive after Idrinking more frequently than he drinks. Here it was discovered that 76 respondents (or 3.1 per cent of the _____________________________________________________________________ |total) indicated that they drove after drinking more often |than they drank. This may be due in part to the respond ent's inability to recall his own habits accurately especially when response categories are rather close (for |example, "several times a week" vs. "once a week"). If so, | I it is anticipated that most discrepancies would be in I ["error" by no more than one response category. Only 10 respondents (0.4 per cent of the total) were in "error" by | two or more categories. The frequency of drinking and driving was also I |cross-tabulated with the frequency of driving after consum- |ing more than the respondent felt he could hold and still I drive safely. Of the 2,183 respondents, 58 or 2.7 per cent i i gave an illogical answer by stating that they drove while quite drunk more often than they drove after drinking, j Thirty-four of these subjects were in "error" by one !category. The remaining 24 (about 1.1 per cent of the I total) marked a frequency category of drinking and drivings,, that was more than one category less frequent than neces- |sary for a consistent response. | Data were gathered from the general driving popula- :tion at the time they were applying for or renewing a drivers license. Under these circumstances it may be |anticipated that respondents would be unwilling to report their traffic violations accurately. However, over one- third of all the DMV respondents (n=545) admit to driving S 57 I jafter drinking at least once a year, which represents a large portion of the driving public. Most errors are probably an underestimate of drinking and driving. Any i relations that occur consequently are important because 'they exist despite probably underestimation of actual |beuavior. i j In summary, neither the convicted drinking driver !nor the never-convicted drinking driver groups are random samples of their respective populations. However, there is good reason to believe that any departures from randomness I are not sufficient to disqualify the use of these data to |generalize to the population of drivers in Los Angeles. One problem is that a small number of respondents who showed past convictions for drinking and driving were removed from the sample. Thus we are unable to estimate the proportion of the driving public that has been con victed for drinking and driving. The removal of these subjects potentially could bias the sample in the direction of higher status people. However, the number is so small that bias is minimal. Problems of validity in these data jmay not be extensive because cross-tabulations suggest reasonably consistent responses. It seems safe to assume that most error is in the same direction, i.e., toward underestimating violating behavior. j Legal Measurement of Drunkenness j The standard procedure used by legal agencies for [measuring the degree of intoxication is to take a sample of jthe subject's breath (or blood or urine) and submit it to j I test. This test measures the amount of alcohol in the ! blood by volume and is expressed as a percentage of the total blood that is alcohol. A number of states have |established certain blood alcoholv(BA) levels as the bound- !ary between being legally drunk and not being legally drunk. For some states the level is .15 per cent and for others it is .10 per cent (that is, .10 per cent of the total volume of blood is alcohol). In California at the time of this study there was no legal definition of drunkenness in terms of a specific BA level. Discretion was left to individual judges; however, the BA level was determined in all cases where there was a charge for drinking and driving. Anyone with a BA level above .15 per cent was very likely to be | convicted of driving while under the influence of alcohol; for a BA level between .10 per cent and .15 per cent, the i i Iperson may be convicted of drunk driving or of reckless jdriving. Reckless driving is essentially the crime of driv ing after drinking with a BA level lower than that usually thought necessary for a charge of drunk driving. Since jconviction on a reckless charge means that an individual was driving after drinking, those convicted of this charge |are included in the sample of all those convicted for i |drinking and driving. I Operational Definitions | | I Frequency The actual frequency of driving after drinking is measured by the question, "How often do you usually drive I after drinking at least two drinks of hard liquor or three drinks of beer?" Response categories vary from "daily" through "once or twice a year" to "never." All drivers in the DMV sample who admit to driving after drinking at least once or twice a year are included in the rule violating group. Note that not all of these people would necessarily be considered drunk drivers if they were stopped by police. That is, two drinks, for most people, is not sufficient alcohol to produce a BA level much above .05 per cent. It is, however, sufficient alcohol to be charged with reckless driving.^ ! |Seriousness j I The seriousness of drinking and driving is diffi- !cult to measure in instances where the person was not apprehended and convicted. The best measure (and one that would facilitate comparison with convicted drinking i 7 drivers) is the BA level. Obviously, it is impossible to obtain the BA level for each rule violator each time he drove after drinking. Another measure of seriousness is | 60 i therefore required. Mulford (1968) has utilized a quantity-frequency I index to assess the seriousness of driving after drinking. i !There are several problems with this type of index. First, |these indices usually apply to drinking patterns only and |ignore the fact that many times the person will not drive j iafter drinking or will drink less when he knows he is going jto drive. Also the quantity and frequency of drinking may jbe inversely related (at least in some cases). Another possibility is to ask the respondent how i |much he usually drinks when he drives after drinking. It I iis presumed that this indicates what the BA level would be | I if it were measured. This may be a fallacious assumption for two reasons. First, body size greatly affects the BA level with a given amount of alcohol consumed. The larger the size, the greater the volume of blood and hence the greater the dilution of a unit amount of alcohol. Second, the span of time over which the drinking is spread can substantially alter the BA level. The human body burns ialcohol such that if one drinks about two drinks per hour i j over a five to six hour period the BA level will stay quite I low (perhaps never more than .05 per cent). On the other I hand, if ten drinks are consumed in one hour the body !cannot burn very much of the alcohol, resulting in a much I ;higher BA level (in an average-sized man, about .20 per i cent, which is a very high level of intoxication). 61 Another difficulty in defining seriousness is balancing frequency against the degree of intoxication. i Which is more serious, the person who drives when very drunk once a year or the driver who drives in a somewhat less intoxicated condition ten times a year? In order to avoid these difficulties the respondents were asked, "Some people can drink more than others and still drive. How many drinks can you handle and still drive well?" Response categories were provided for each of a variety of alcoholic beverages. The respondents were then asked, "How often have you driven in the past twelve months after having consumed more than that amount?" Based on responses to these questions, seriousness of drinking and driving is measured as the admitted frequency of driv ing after having consumed more than the respondent feels he can hold and still drive well. This is an adequate, though I by no means perfect, indicator of seriousness. It controls indirectly for the differential effects of alcohol by letting the respondent define his capacity to drink and still drive safely. Socioeconomic Status Socioeconomic status is a difficult concept to measure because no one operational indicator is considered adequate. A procedure often used is to combine several indicators such as occupation, education and income into S one index. This may lead to problems in index construction 62 I j which will decrease rather than increase an investigator's j ability to predict. In the present case three separate operational measures of social class are used: occupation (for those working), education, and income. Respondents were asked to indicate the highest level of education achieved in terms of years of formal schooling. Occupation i :is measured by a scale conforming essentially to that used by the United States Bureau of the Census. Finally, the definition of income is total family income as reported by the respondent. This is used in preference to the respond ent 's individual income, because it is felt that the style of life, ability to retain lawyers, and other related factors are more accurately determined by the total family income. i |Race-ethnic Status i | The race-ethnicity of respondents was indicated by |their response to an item listing six categories: Caucasian, Mexican-American, Negro, Oriental, American- Indian, and other. In this analysis the categories of Oriental, American-Indian, and other were recoded into a single category labelled "other." Discretion within the Legal System The measure of police discrimination is the response to the question, "Was there ever a time when a policeman stopped you after you had been drinking but let I 63 you go without arresting you for drunk driving?" Obtaining more adequate information on police operations was not possible. The indicator of discrimination by the district attorney is based on whether the initial charge of drunk idriving was reduced to the lesser charge of reckless driv- ; ing. Technically this should only occur when the BA level of the subject is low (probably less than .10 per cent), jConversely, the reduced charge should not occur in cases of higher BA levels (above .10 per cent). This information i was taken from the official court record. Two measures of the severity of the penalty levied by the judge was used: amount of fine paid, and number of days actually spent in jail. This information was also |gathered from the official court record. j i Drinking and Driving Convictions The total number of both reckless and drunk driving convictions was obtained from the official criminal records of the State of California. This is a conservative esti mate of the actual total, because out-of-state convictions are not included. Measures of Association To test the hypotheses, measures of association are required between pairs of variables. Relating variables would be a relatively simple procedure if all were measured on an interval scale. However, most of the variables are measured ordinally and a few are nominal. Since it has been shown that the more powerful intervally based measures of association can be used on ordinally measured data (Labovitz, 1970), both ordinally-based and intervally-based measures of association are reported. The ordinal measure of association to be used is Somers' d ^ (Somers, 1962). It is an asymmetric measure which can vary from -1.0 to +1.0. This measure has the advantage of an unambiguous operational interpretation as a |proportionate-reduction-in-error (PRE) measure (see Costner, 1965). (Initially Costner argued that Somers' measure did not fit the requirements for the PRE inter pretation but in a later exchange with Somers, Costner agreed that dyX can be interpreted as a PRE measure. See Somers, 1968, and Costner, 1968). The value of the coeffi cient indicates the proportion of the error made in "guessing" the value of the dependent variable that is reduced by knowledge of the independent variable, ignoring pairs tied on the independent variable. Another operational interpretation of dyx is given by Somers, who states that 65 j | "expresses the proportionate excess of concordant pairs j over discordant pairs among pairs not tied on the independ ent variable" (see Somers, 1962; 804). The two forms of the measure are defined by the following equations, where d x indicates that x is the independent variable, and d indicates the reverse. m L and ^ where P = the number of concordant pairs of observations Q = the number of discordant pairs of observations X ■ the number of pairs of observations u not tied on X Y = the number of pairs of observations u not tied on Y Because Somers' measure lacks the sensitivity of ! the Pearson product moment correlation coefficient, r will i be reported along with the d . It has been shown empir- ically that interval level measures of association applied to ordinal data results in small error (provided the ordinal scale is clearly monotonic). Labovitz (1970) assigned 20 different scoring systems to a clearly ordinal scale. Assuming each system to be the "true" score, he correlated it with the remaining 19 and found, in general, Q very high correlations. Labovitz concludes that there are three advantages to using interval statistics on ordinal data: 66 (1) the use of more powerful, sensitive, better developed and interpretable statistics with known sampling error, (2) the retention of more knowl edge about the characteristics of the data, and (3; greater versatility in statistical manipula tion, e.g., partial and multiple correlation and regression, analysis of variance and covariance, and most pictorial presentations. (Labovitz, 1970: 523) Several of the hypotheses require the statistical control of one or more variables. Because there is no adequate partial d , the partial analysis techniques or "modes of elaboration" are used. This involves the phys ical separation of the sample into groups in terms of values on some control variable. The zero-order correla tion is then compared within each sub-group. This procedure has several clear disadvantages: (1) if the i Icontrol variable is divided into more than a few categories, the sample of cases is distributed very thinly, so that leach cross-tabulation has a small N; (2) if the control variable is divided into only a few categories to avoid small sample size much variation is still possible within categories, thus providing only a crude degree of control; (3) finally, it is generally possible to control for the effects of only a few outside variables at a time. If several control variables are used, sub-group sample size is often too small to justify confidence in interpreting the association. Following Labovitz1 (1970) argument, the partial I correlation coefficient (partial r) will be reported along 67 with the partial analysis when using statistical control, and Pearson's r will be reported along with Somers' dyx. In this study tests of significance are not reported. Technically, tests are not applicable when the sample is not a strict random sample of some theoretically important population. In this case neither the DMV group nor the convicted drinking drivers were chosen by purely random methods. Another reason for not reporting tests of significance is the large sample size of this study (N = 1,196 after weighting). In a sample of this size, very small correlations will be statistically significant. Such small correlations, even though significant, may constitute little theoretical support for an hypothesis. For this I study, as a rule of thumb, a correlation of .10 is suffi ciently small to be considered "about zero." A correlation (such as the Pearsonian r) of this size accounts for only i one per cent of the variance in the dependent variable. FOOTNOTES The original State of California great number was STA-1. Later the project was transferred to the Federal Department of Transportation and became grant number FH-11- 7099. The project was carried out by co-directors Jack Bishop (Administrative Director) and William Hill (Research Director). Much of the credit for gathering the data on the DMV sample goes to Mrs. Oksana Didenko. 2 Initially an exploratory study was conducted using an in-depth interview covering a wide variety of areas. It was administered to 54 persons who were renewing or apply ing for a new drivers license. These persons constituted the control group in the exploratory period since none had ever been convicted for driving while under the influence of alcohol. The experimental group in this period was composed of 62 persons convicted for driving while under the influence of alcohol. The results of the interviews were analyzed and a new instrument was constructed which used forced choice response categories. It was adminis tered to another pilot sample of convicted and never- convicted drivers. From this second instrument a final set of questions was prepared. 3 To compute the weight factor it is necessary to find a fraction which when multiplied by 3/4 equals 1/3. Solving for X in the equation, 3/4X = 1/3, yields the fraction .44. i ^Both questionnaire and interview forms of the instrument were used in order to compare them. The litera ture pertaining to data-gathering methodology is not clear regarding which form is more likely to illicit honest responses. It was initially planned that the two forms would be randomly assigned and then any statistically significant differences would indicate which form to use in subsequent applications. However, the assignment was not entirely random. In a number of cases the research assist ant gave the instrument in interview form when the subject lacked a level of literacy sufficient to complete the questionnaire. Analysis of the difference between the two forms indicated few significant differences. This occurred despite the potential for biasing that results from the disproportionate administration of the interview to people with low levels of education. Thus in this study no distinction is made regarding the source of the data. ^The following table indicates the weights applied in each logically possible category of sex and number of convictions: 69 FOOTNOTES (continued) Convictions 0 1 2 3+ Male 1 .44 .44 .44 Female 5 .44 .44 .44 The subroutine which does the weighting also rounds all fractions to whole numbers. This is done only with the total figures and so involves very little error. This is a case weighting, that is, each case is properly weighted even when entered in a table which does not involve either sex or number of convictions. £ Persons with a BA level as low as .05 per cent or lower are, however, much less likely to be convicted even on a reckless driving charge than persons with a higher BA level. ^Even this measure has some drawbacks, however. The effect that alcohol has on people is determined by more than just the concentration in the body. The neophyte drinker may suffer measurably greater loss of judgment and driving capacity with less alcohol than the experienced and knowledgeable drinking driver. Various cultural factors may also affect behavior after drinking. Q Of 190 correlations computed by Labovitz, axi were above .90, and 157 were greater than .97. 70 REFERENCES Costner, Herbert L. 1968 "Reply to Somers." Letter to the Editor, American Sociological Review 33 (April): 292. Costner, Herbert L. 1965 "Criteria for Measures of Association." American Sociological Review 30 (June): 341-53. Labovitz, Sanford. 1968 "The Assignment of Numbers to Rank-Order Cate gories." American Sociological Review 35 (June): 515-24. Somers, Robert. 1968 "On Measurement of Association." Letter to the Editor, American Sociological Review 33 (April): 2ft=V2T~-~ Somers, Robert. 1962 "A New Asymnetric Measure of Association for Ordinal Variables." American Sociological Review 27 (December): 799-811. CHAPTER IV FINDINGS Zero-Order Correlations j This chapter reports the results of testing the hypotheses developed in Chapter II. The first set of hypotheses relates race, social class, sex, and age to conviction for drinking and driving. Race The first hypothesis states that blacks and Chicanos are more likely than Caucasians to be convicted for drinking and driving. The correlations, dyx = -.51 and |r = .47, are moderately high and in the predicted direction. t I According to the correlation coefficient, roughly 25 per cent of the convictions are accounted for by race; the value of Somers' d for this cross-tabulation represents a 26 per cent reduction in error of prediction (using race to predict conviction rates for drinking and driving). These results support the hypothesis.'*' The percentages show marked differences between races; 88.5 per cent of the blacks and only 23.3 per cent of the Caucasians have been convicted for drinking and 71 72 driving in a sample where everyone admits to driving after drinking. The percentage of Chicanos convicted (71.1 per cent), while intermediate between blacks and Caucasians, is much closer to the percentage for blacks. Social Class The second hypothesis posits that social class is negatively correlated with conviction for drinking and driving. Three different indicators of social class are used--occupation, education, and family income. Table 2 presents the cross-tabulation of occupation with conviction. Those with higher status occupations are, as hypothesized, less likely to receive a conviction for drinking and driving than those in less prestigious occupa tions. The dyx measure between occupational level and conviction status is -.21 and r is -.29. Although the |relations are fairly low, each is substantially greater than zero and in the predicted direction; and consequently supports Hypothesis 2. These data indicate a clear difference between white collar and blue collar occupations. The four white collar occupations— professionals, managers, clerks, and salesmen--have about the same percentage convicted (around 25 per cent). The blue collar occupations (all the rest) have a substantially higher proportion of members receiving convictions. Laborers have the highest proportion of TABLE 2.--Occupational Status Cross-Classified by Conviction Status among Admitted Drinking Drivers* Occupation Convicted Never Convicted Professionals, Technicians and Kindred Workers 29.1% (103) 70.9% (252) Managers, Officials, and Proprietors 22.0% (27) 78.0% (97) Clerical Workers 27.3% (24) 72.7% (64) Sales Workers 22.6% (18) 77.4% (63) Craftsmen and Foremen 56.3% (110) 43.7% (85) Operatives 68.5% (96) 31.5% (44) Private Household and Service Workers** 48.0% (49) 52.0% (53) Laborers (Farm and Non-farm) 73.4% (66) 26.6% (24) *The sample N of 1,196 is reduced 21 because of either a response or an uncodable response. **The separate categories of Private Household Workers and Service Workers were combined because there were so few private household workers. members who have been convicted of drinking and driving 2 (73.4 per cent). The cross-tabulation of education with conviction is presented in Table 3. As expected, the percentage of convictions decreases as the level of education increases. This is reflected in the dyx of -.31 and the r of -.41. These measures indicate that education reduces the error in predicting conviction rate by 10 per cent to 15 per cent. This range of error reduction represents a rather small proportion, but it is sufficient to lend support to Hypothesis 2. The "well-educated" show a particularly small proportion in the convicted category. Among those with a college diploma or a professional degree only 10.4 per cent of the respondents (15 of 144) have been convicted 3 iof drinking and driving. | | The relation between the final indicator of class, family income, and conviction is presented in Table 4. The overall trend is for the percentage of convictions to decrease as the level of income increases. The dyx measure is -.24 and r is -.21, which are again fairly low but in the predicted direction.^ In summary, the three indicators of social class point to the same conclusion: although the relations are fairly low, social class is related to conviction for drinking and driving. These results offer moderate support for the second hypothesis. TABLE 3.--Educational Level Cross-Classified by Conviction Status among Admitted Drinking Drivers* Level of Education Completed Convicted Never Convicted Grades 1-6 86.6% (39) 13.4% (6) Grades 7-9 72. 2% (75) 27.8% (29) Grades 10-12 66.7% (148) 33.3% (74) High School Diploma 43.97, (127) 56.1% (162) Some College 25.9% (87) 74.1% (248) A College Diploma 9. 9% (12) 90.1% (106) A Professional Degree 12. 2% (3) 87.7% (23) *The N of 1,139 is 58 short of the total sample size of 1,196. This is caused by uncodable responses. TABLE 4.--Total Family Income Cross-Classified by Conviction Status among Admitted Drinking Drivers* Total Family Income Convicted Never Convicted Under $3,000 60.0% (43) 40.0% (29) $3,000 - $5,999 65.2% (118) 34.8% (63) $6,000 - $8,999 46.5% (143) 53.3% (165) $9,000 - $11,999 44.9% (76) 55.1% (93) $12,000 - $14,999 28.4% (40) 71.6% (102) $15,000 and over 10.9% (23) 89.1% (191) *The N here is 108 less because of subject non- than the total -response. of 1,196 Somers' d = -.24 and r = -.21. 77 i I Sex | The third hypothesis anticipates an association between sex and conviction status, with males overrepre sented in the convicted category. As predicted, the data show that males are more likely to be convicted. Almost 49 per cent of the males and only about 11 per cent of the females have been convicted of drinking and driving. For this cross-tabulation, the d ^ is .38 and r is .28, which support Hypothesis 3. Age A negative association between age and conviction is predicted in Hypothesis 4. Based on the cross-tabulation i in Table 5, d and r both equal .09. These correlations “ JrX — are quite small and not in the predicted direction, that is, the coefficients are positive rather than negative. In i ;Table 5, the percentage of each age group in the convicted category increases as age increases (up to the age of 55 years). The hypothesis specifies the reverse; i.e., a decreasing proportion of each older age category in the convicted group. This departure from the hypothesis perhaps reflects the greater probability of conviction simply as a product of the number of years of driving. In summary, three out of the first four hypotheses are supported. The hypotheses relating race, social class, and sex to conviction for drinking and driving are 78 TABLE 5.--Age Cross-Classified by Conviction Status among Admitted Drinking Drivers* Age Convicted ConvISed Under 26 28.3% (75) 71.7% (190) 26-35 42.7% (145) 57.3% (194) 36-45 49.0% (165) 51.0% (171) 46-55 51.3% (91) 48.7% (86) Over 55 35.6% (29) 64.4% (52) *The N in this table is actually larger than the total sample because the weighting program rounds to an even number of persons. This occasionally produces more people in a cross-tabulation than in the total sample. Somers' d = .09 and r = .09. ¥25 79 supported, while the hypothesis relating age to conviction is not. The finding that blacks and Chicanos, the lower class, and males are overrepresented in the convicted group is consistent with the findings of other studies examining crime in general. S t a t i s t i c a l C o n t r o l f o r A c t u a l I l l e g a l B e h a v i o r The next step is to test a proposed explanation for the overrepresentation of certain categories in the convicted group. One hypothesized explanation asserts that discrimination occurs on the basis of certain social characteristics. In other words, it is expected that rates I of actual violation among certain aggregates will be ! insufficient to account for the greater conviction rates of ■these aggregates. ! i Race In accordance with the above argument, it is antici pated that race will have a direct effect on conviction independent of its influence on actual criminal behavior (although it may influence behavior as well). Specifically, Hypothesis 5 indicates that controlling for actual drinking and driving behavior, blacks and Chicanos are more likely than Caucasians to be convicted for drinking and driving. T he d a t a i n T a b l e 6 l e n d so m e s u p p o r t t o t h e h y p o t h e s i s . T he d y ^ c o e f f i c i e n t s a r e n o t l a r g e b u t a l l a r e i n t h e 80 TABLE 6.--Correlations Between Conviction Status and Race, Social Class, Sex and Age Among Admitted Drinking Drivers, Controlling for Frequency of Drinking and Driving, and Frequency of Drunk Driving Correlation Between Conviction Status and Social Characteristics Variable Controlled Race Occupa tion - Educa tion Income Sex Age Frequency of Drink ing and Driving:* High Medium Low -.35 -.35 -.59 -.12 -.14 -.20 Somers' -.16 -.27 -.29 -.13 -.17 -.23 .31 .31 .30 -.14 -.12 -.08 Frequency of Drunk Driving:** High Low Never -.25 -.52 -.55 -.12 -.20 -.16 -.19 -.32 -.25 -.18 -.29 -.15 -.25 .42 .21 -.10 -.10 -.06 Frequency of Drink- and Driving -.46 -.26 Partial -.37 r -.18 .22 -.10 Frequency of Drunk Driving -.45 -.26 -.36 -.18 .22 -.08 Frequency of Drink ing and Driving and Frequency of Drunk Driving -.45 -.24 -.35 -.17 .19 -.10 *High frequency of drinking and driving is once a day or more frequently. Medium frequency drinking and driving is less than once a day and including "on the average once every two weeks." Low frequency drinking and driving is less than once every two weeks up to and including once or twice a year. **High frequency drunk driving is up to five or six times a year. Low frequency drunk driving is less than 5 or 6 times a year but at least once or twice a year. L N= 1,196____________ ________________ predicted direction under each of three levels of both frequency of drinking and driving and frequency of drunk driving. The first- and second-order partial correlation coefficients also support the hypothesis. The data indicate some statistical interaction between frequency of offense and race in affecting convic tion. The magnitude of d between race and conviction varies among the categories of frequency of drinking and |driving and frequency of drunk driving. The pattern on both control variables is for the coefficient to increase as frequency decreases. This suggests that the degree of i discrimination decreases as frequency and severity of | offense increases. Among the "casual" offenders, those vdio j either rarely drink and drive or never drive when drunk, race accounts for 30-35 per cent of the variance in convic tion. Among those who either drink and drive frequently or are frequent drunk drivers, race accounts for only 6-12 per cent of the variance in conviction. The rationale being tested suggests that both actual offense behavior and social characteristics operate to produce a conviction for drinking and driving. How these two factors operate to determine conviction was not speci fied. These data suggest that the two interact to influence conviction. To illustrate, the overall dyx between race and convictions is .51; thus the smaller coefficients among ;"high" and "medium" frequency offenders indicate that actual behavior does intervene to some degree between race and conviction. In other words, a portion of the zero-order relation between race and conviction seems to be due to the connection between race and conviction through the interven ing variable of offense behavior. In summary, actual behavior seems to intervene to some extent between race and conviction. Race and actual behavior, furthermore, appear to interact to a moderate degree in affecting conviction. The relationship between race and conviction under control conditions of actual violating behavior is, on the average, somewhat lower than the overall correlation but remains well above zero. These results lend partial support to Hypothesis 5. Social Class Hypothesis 6 stipulates that the negative correla tion between social class and conviction status will remain after controlling for the effect of actual offense behavior. The correlations between each of the three indicators of social class and conviction, after appropriate controls, appear in Table 6. The d coefficients are small within categories of frequency and seriousness of drinking and driving but they are consistently in the predicted direction. The partial r1s are also greater than zero and in the predicted direc tion when controlling for either the frequency or the s e r i o u s n e s s o f d r i n k i n g a n d d r i v i n g , o r b o t h . T h e z e r o - o r d e r r b e t w e e n e a c h i n d i c a t o r o f c l a s s a n d c o n v i c t i o n i s o n l y s l i g h t l y g r e a t e r t h a n e i t h e r t h e f i r s t - o r s e c o n d - o r d e r p a r t i a l r . T h i s s u g g e s t s t h a t a c t u a l o f f e n s e b e h a v i o r d o e s n o t c a u s a l l y i n t e r v e n e b e t w e e n s o c i a l c l a s s a n d c o n v i c t i o n f o r d r i n k i n g a n d d r i v i n g . I f o f f e n s e b e h a v i o r w e r e a c a u s a l l i n k b e t w e e n c l a s s a n d c o n v i c t i o n , h o l d i n g i t c o n s t a n t s h o u l d p r o d u c e a p a r t i a l c o r r e l a t i o n n e a r z e r o ; t h i s d i d n o t o c c u r . T h i s f i n d i n g , t h e r e f o r e , t e n d s t o r e f u t e t h e r a t i o n a l e w h ic h i n d i c a t e s t h a t t h e p r im a r y d e t e r m i n a n t o f c o n v i c t i o n i s a c t u a l o f f e n s e b e h a v i o r a n d t h a t s o c i a l c h a r a c t e r i s t i c s a r e u n i m p o r t a n t . T h e c o r r e l a t i o n s i n T a b l e 6 s u g g e s t t h e p r e s e n c e o f i is o m e s t a t i s t i c a l i n t e r a c t i o n b e t w e e n o f f e n s e b e h a v i o r a n d s o c i a l c h a r a c t e r i s t i c s i n a f f e c t i n g c o n v i c t i o n . F o r e a c h I i n d i c a t o r o f s o c i a l c l a s s t h e d „ „ c o e f f i c i e n t s i n c r e a s e a s | j b o t h f r e q u e n c y a n d s e r i o u s n e s s o f d r i n k i n g a n d d r i v i n g d e c r e a s e . T h e e x c e p t i o n i s t h e " n e v e r " c a t e g o r y o f d r u n k d r i v i n g f o r a l l t h r e e i n d i c a t o r s o f s o c i a l c l a s s . I n t e r a c t i o n s u g g e s t s t h a t o f f e n s e b e h a v i o r a n d s o c i a l c l a s s c o m b in e m u l t i p l i c a t i v e l y t o a f f e c t c o n v i c t i o n p r o b a b i l i t y . S t a t e d o t h e r w i s e , t h e v a l u e o f o n e v a r i a b l e a f f e c t s h o w t h e I s e c o n d v a r i a b l e i s r e l a t e d t o c o n v i c t i o n . A p p a r e n t l y t h e d e g r e e o f d i s c r i m i n a t i o n i n c r e a s e s a s t h e f r e q u e n c y a n d s e r i o u s n e s s o f d r i n k i n g a n d d r i v i n g d e c r e a s e s . T h i s i s s i m i l a r t o t h e f i n d i n g o f i n t e r a c t i o n b e t w e e n r a c e a n d 84 i offense behavior in influencing conviction. A rationale posited in Chapter II indicated that both offense behavior and social characteristics operate in I determining conviction for drinking and driving. However, the nature of the connection between these two was not specified. The above results and those for race suggest that the relation between these two causal factors is at least partially multiplicative. That is, the effect of the two variables on the dependent variable is greater than the sum of the independent effects of each variable separately. Sex Hypothesis 7 anticipates an association between sex land conviction after the effect of the frequency and i ;seriousness of drinking and driving has been statistically I controlled. The correlation coefficients in Table 6 gener ally support this hypothesis. The zero-order r between sex and conviction is .28; both first-order partials are .22 and the second-order partial is .19. The dyx measure supports this. The ''zero-order'* d ^ between sex and conviction is .38; as can be seen, it is reduced only slightly under each of three categories of drinking and driving. The offense behavior, therefore, does not appear to intervene directly between sex and conviction. Frequency of drunk driving yields inconsistent results as a control variable. Among high frequency drunk 85 drivers, dyX between sex and conviction is -.25 and among low frequency drunk drivers it is .42. This may well be an artifact of the small number of women who admit to a high frequency of drunken driving. Of the 312 high frequency drunk drivers only 10 are women and all of these have been convicted. Among low frequency drunk drivers (362) there are 61 women and 55 of these have never been convicted. Because of the small number of seriously offending women the negative association should be interpreted cautiously as constituting negative evidence for the hypothesis. There appears to be no interaction between sex and offense behavior in affecting conviction. The coefficients under each category of frequency of drinking and driving are almost exactly the same (.31). This is somewhat puzzling; by showing a relationship to conviction that is i |not diminished by controlling for actual behavior, sex is ishown to be a basis for discrimination just as race and class are. Yet both race and class show interaction with offense behavior in determining conviction while sex does not. In general, these data substantiate the hypothesized relation between sex and conviction controlling for actual offense behavior. Age The eighth hypothesis states that after controlling for illegal drinking and driving behavior there will be a inegative correlation between age and conviction. This_____ 86 hypothesis is complicated by the fact that the "zero-order" hypothesis was not supported. The d and r values between age and conviction are only .09. This quite small but positive association suggests a slight tendency for the proportion convicted to increase as age increases. All the partial r's and the partial-analysis dyx1s, however, are negative in sign as initially predicted. Although the partials are quite small, the consistently negative sign suggests something more than an artifact. It is possible that the positive zero-order relation indicates that older people have had more time to get convicted. Presumably, all other things being equal, the longer one has driven the greater the chance of both drinking and driving, and being apprehended and convicted. When one controls for the number of actual violations, however, the young are slightly more likely than older persons to be convicted. This suggests, as predicted, that a slight amount of discrimina tion against younger drivers occurs. In general, these small but consistently negative correlations provide some support for Hypothesis 3. Hypotheses 5 through 8, in effect, assert that race, social class, sex, and age have a direct effect on convic tion for drinking and driving independent of their effect on actual offense behavior. All four hypotheses receive some support. These results do not support the idea that blacks and Chicanos, the lower class, and males are overrepresented in the convicted category because they offend more fre quently and/or more seriously. Stated otherwise, conviction probability is independently altered by the social charac teristics of the alleged offender. Not only do social categories directly affect the probability of conviction but they also interact with offense behavior in influencing the probability of conviction. Interaction appears to occur only with race and class; sex and age do not appear to interact with offense behavior in affecting conviction probability. These results tend to refute "single factor" explanations of conviction for drinking and driving. The rationale that conviction is determined by actual offense behavior alone is not supported, because social categories are related to conviction after controlling for offense behavior. The second rationale, that conviction is deter mined solely by the social category of the offender, is not supported, because of the presence of statistical interac tion between social categories and drinking and driving behavior in influencing conviction. These findings, there fore, support the rationale that conviction, by both additive and interactive effects, is the result of drinking and driving behavior and the social characteristics of the person. ; 88 Discrimination and the Legal System One possible source of discrimination against racial minorities, the lower class, males and the young lies within the legal system. It was suggested earlier that once discrimination is empirically demonstrated, it is often automatically assumed that legal agents are using their discretion to systematically favor certain aggregates, or, conversely, that they are more harsh with members of other, less "acceptable," groups. Police Discretion The policeman is the first decision maker in the apprehension-conviction sequence who could discriminate. It is anticipated, however, that discrimination stems primarily from socio-structural factors exterior to the | legal system (see Chapter II, pages 36-37). Therefore, Hypotheses 9 through 12 specify that after controlling for actual offense behavior, police do not discriminate on fhe basis of race, class, sex, or age in their treatment of alleged drinking drivers. Stated otherwise, the severity of police response will not vary depending on the race, class, sex, or age of the person stopped. The associations |in Table 7 are generally low, although some are clearly greater than zero. Age and the social class indicators are not related to police severity after the appropriate controls have been 89 TABLE 7.--Correlations Between Police Severity and Race, Social Class, Sex and Age Among Admitted Drinking Drivers Controlling for Frequency of Drinking and Driving and Frequency of Drunk Driving Variable Controlled Correlation Between Police Severity and Social Characteristics_______ Occupa- Educa- Race tion tion Income Sex Age Frequency of Drink ing and Driving:* High .13 Medium .13 Low .20 Frequency of Drunk Driving:** High .05 Low .12 Never .22 Frequency of Drink ing and Driving .16 Somers' .07 .03 .07 -.10 .08 -.10 -.01 -.45 .05 -.08 -.28 .02 -.10 -.15 .03 .00 .11 .08 10 -.01 -.12 -.08 Partial r -.03 -.13 -.05 ,20 ,17 ,14 .07 -.01 -.11 .03 .02 ,02 .00 Frequency of Drunk Driving 11 -.06 -.01 Frequency of Drink ing and Driving and Frequency of Drunk Driving .11 .06 -.01 ,02 -.07 -.02 .02 -.07 -.02 *High frequency drinking and driving is once a day or more frequently. Medium frequency drinking and driving is less than once a day and including "on the average once every two weeks." Low frequency drinking and driving is less than once every two weeks up to and including once or twice a year. **High frequency drunk driving is up to five or six times a year. Low frequency drunk driving is less than 5 or 6 times a year but at least once or twice a year. introduced. This supports Hypotheses 10 and 12; apparently,j j police do not use discretion differentially depending on the age or social class of the person stopped. Race and sex, on the other hand, show small but consistent relations with police severity, both before and after statistical controls have been applied. The "zero-order" dyx and r both equal .19 between police severity and race. This indicates that minority group members are more likely to have been stopped then released after drinking and driving than are Caucasians. On the surface this is a surprising finding. However, this result is ambiguous. It could mean that police discriminate against Caucasians and tend to be more lenient with minority group members. Thus Caucasians are released after being stopped less often because they are arrested a larger proportion of the times they are stopped. Second, these findings may reflect the fact that minority group members drink and drive more frequently than do Caucasians. Among drinking drivers, the more frequently one offends, the greater the overall probability that he will be stopped then released at least once. Thus high frequency offenders should also show the highest release rate. Third, blacks and Chicanos may be stopped by police much more often than are Caucasians, even among those who offend with equal frequency. If, for example, police density is greater in minority group areas, then a black or Chicano offender is more likely than a Caucasian to be stopped. If more offenders are being stopped, then perhaps more are less-serious offenders. It is the less-serious offender who is most likely to be released. Thus minority group offenders are released more often because police stop non-serious, minority group offenders in larger proportions than Caucasians. These possibilities can be examined with the data at hand. The results from testing Hypotheses 1 and 5 make it seem unlikely that Caucasians experience discrimination. It was shown that, controlling for actual offense behavior, blacks and Chicanos are much more likely than Caucasians to be convicted for drinking and driving. If minority group members are more likely to be convicted, it does not seem consistent that they are less likely to be arrested. Furthermore, if Caucasians are being discriminated against, then among those never stopped and then released after drinking, a larger proportion of Caucasians than minority group members should be convicted for drinking and driving. That is, if Caucasians are receiving more severe treatment they are less likely to be stopped and then released (after drinking), because every time they are stopped they are arrested (i.e., not released). The data indicate just the opposite. Among those never stopped and then released after drinking, 14 per cent of the Caucasians, 68 per cent of the Chicanos, and 81 per cent of the blacks have been convicted of drinking and driving. 92 I ; The second possible interpretation of the correla tion between race and police severity— that Chicanos and ‘blacks are released more often than Caucasians because they offend more often--is also unsupported by the data. Controlling for the frequency of offense should have reduced the correlations between race and police severity to near zero. This did not occur. The zero-order r between race and police severity is .19. It is reduced only to .16 when the partial r is computed, controlling for actual frequency of offense. Thus differential rates of offense do not account for the greater release rate of minority group members. The third and last interpretation for the greater release rate for minority groups is that these people, ibeing more vulnerable to the law, are more often stopped j for less serious offenses. It can be reasonably expected i that those who are the least serious offenders are most likely to be released. Thus, a larger proportion of all minority group members have been released after being stopped for drinking and driving because they are more likely to be stopped for a minor infraction of the law and it is this group that is most likely to be released. The [data are consistent with this interpretation; among non- serious convicted offenders (BA level of .09 or less) only 112 per cent of the Caucasians, but 35 per cent of the minority group members, have been stopped and then released after drinking. Thus, among non-serious offenders, for example, blacks are stopped and then released more often because they are stopped more often. If blacks and Chicanos are released more often only because they are more often stopped on non-serious viola tions , then they should not show a higher release rate j among serious offenders. The data support this by showing that, for example, among serious drinking drivers (BA level = .20+), Caucasians are more likely to be released (58 per cent) than blacks (45 per cent). There is apparently some discrimination by police in releasing Caucasian offenders more often than non-Caucasian offenders. This is masked by the fact that a high proportion of minorities are stopped for non-serious violations, and it is the people in this ;category who are most likely to be released, i The results tend to refute the hypothesis of no i |relation between police severity and race and instead support the assumption that some discrimination is created within the legal structure. It is not possible to specify the degree of discrimination, because of the lack of adequate data bearing on the BA level of the offender each time he was stopped and then released. However, by infer ring past BA levels from the BA level at the time of arrest among convicted offenders, it appears that the degree of ‘ discrimination is small and much less than that necessary to account for the markedly higher conviction rate of blacks and Chicanos. A similar difficulty in interpretation occurs in examining the coefficients between sex and police severity. Both the d^ (-.25) and the zero-order r (-.17) indicate that men are more likely to be released after being stopped when drinking than are women. The situation is somewhat different than for race because the partial r drops to near zero (-.07) when controlling simultaneously for the BA level and frequency and seriousness of offense. The dyx measures, on the other hand, do not drop appreciably within categories of frequency of drinking and driving. The interpretations analogous to those suggested for race apply here. The data tend to refute the notion that women experience discrimina tion. The data for Hypothesis 7 indicate that after con trolling for the effects of actual offense behavior, men are much more likely to be convicted than women rather than the reverse. However, differentials in the frequency and seriousness of drinking and driving apparently account for part of the greater release rate for men. Controlling for frequency or especially seriousness of offense, the corre lation between sex and police severity drops (but not to zero). Perhaps men are released more often because they are stopped more often. This cannot be demonstrated so clearly for sex as for race because of the small number of women in the convicted category. At any rate these results tend to refute the hypothesis of no relation between police severity and sex. To summarize, examination of the data concerning the relationships between police severity and social characteristics indicates that some race and sex discrim ination occurs, but there is no discrimination based on class or age. Apparent racial discrimination by police involves stopping minority group members more often than Caucasians. Rather than assuming that police officers intend to stop minorities more often, this may merely reflect a factor like differential police density. The greater number of police in ghetto areas could result in closer surveillance and the apprehension of more of the less serious offenders. If so, discrimination is, in part, structurally induced, as well as resulting from prejudicial i beliefs. Discretion by the District Attorney The district attorney determines the charge to be brought against a subject arrested for drinking and driving. There are two crimes for which alleged drinking drivers are prosecuted; drunk driving and the lesser charge of reckless drinking. The legal charge, presumably, is determined by the seriousness of the drinking and driving violation and perhaps by the number of past convictions for this offense. Hypotheses 13 through 16 posit that receiving the reduced charge (reckless driving) is not related to race, class, sex, or age after controlling for seriousness and for the 96 past record of drinking and driving. The data in Table 8 support these hypotheses. The partial r's and the partial-analysis dyx*s are near zero, as are the zero-order correlations. In short, virtually no discrimination on the basis of race, class, sex, or age is apparent in the assignment of the legal charge for which the alleged offender is to be prosecuted. This finding supports the rationale which specifies that discrimination is not produced by the legal system (see Chapter II, pages 36-37). Discretion by the Judge Decisions made by judges may also be a source of discrimination in the legal system. If, in fact, socio- structural factors are more salient in determining discrim ination, then judges should show relatively little discrim ination against drinking drivers. Following this line of thought, Hypotheses 17, 18, 19 and 20 indicate that the severity of the penalty after a conviction for drinking and driving is not related to race, class, sex, or age. While these decisions are made after the conviction decision, they 'do indicate apparent judicial discrimination which may operate in other decisions as well. One indicator of the possible partiality of judicial decisions is the amount of the fine levied. The correlation coefficients in Table 9 indicate that judges do not discrim inate by race, class, sex, or age in levying fines against__ 97 TABLE 8.--Correlations Between Legal Charge* and Race, Social Class, Sex and Age Among Convicted Drinking Drivers, Controlling for Blood Alcohol Level and Total Convictions for Drinking and Driving Correlation Between Legal Charge and Social Characteristics Variable Controlled Race Occupa tion ■ Educa tion Income Sex Age Somers1 % None .00 .01 -.04 Pearson - .06 's r .01 .05 None .04 -.02 -.06 Somers1 -.03 .02 .06 Blood Alcohol Leve1:** High Medium Low .02 -.04 .05 -.02 -.01 -.01 -.01 -.01 -.03 -.01 -.01 -.06 .04 .13 .15 .01 .03 .13 Total Convictions for Drinking and Driving: 1 2 3 or more - .02 .04 -.03 .00 -.03 .03 -.02 .05 .03 Partial -.03 .00 .01 r .09 *>Wr .05 .01 .03 Blood Alcohol Level .04 -.01 - .05 -.02 .03 .04 Total Convictions for Drinking and Driving .03 -.02 -.05 -.03 .03 .05 Blood Alcohol Level and Total Convic tions for Drinking and Driving .03 .00 -.05 -.02 .03 .04 *Legal charge is a dichotomous variable. The alterna tives are drunk driving and reckless driving. **A high blood alcohol level is .20 and above. A medium blood alcohol level is .10 to .19. A low blood alcohol level is .09 or less. ***There were too few women in these categories to compute the correlation. ... ’ ...... 98 TABLE 9.--Correlations Between Fine Charged and Race, j Social Class, Sex and Age Among Convicted Drinking Drivers, Controlling for Blood Alcohol Level and Total Convictions for Drinking and Driving Correlation Between Fine Charged and Social Characteristics Variable Controlled Race Occupa tion - Educa tion Income Sex Age Somers1 V None .28 -.11 -.13 Pearson -.11 's r -.21 .05 None .03 .02 .05 Somers1 .01 .01 .05 Blood Alcohol Leve1:* High Medium Low .07 .00 .09 -.02 -.06 -.11 .06 .08 .08 .10 .06 -.01 .26 .04 -.26 .01 -.04 .01 Total Convictions for Drinking and Driving: 1 2 3 or more .01 .13 .17 -.06 -.05 .09 .05 .07 -.03 Partial .02 .07 .02 r -.03 -.24 ** .04 .01 .02 Blood Alcohol Level .03 .04 .06 .03 .01 .03 Total Convictions for Drinking and Driving .04 -.02 .04 .00 .00 .06 Blood Alcohol Level and Total Convic tions for Drinking and Driving .04 .04 -.05 .01 .00 .04 *A high blood alcohol level is .20 and above. A medium blood alcohol level is .10 to .19. A low blood alcohol level is .09 or less. **There were too few women in this category to compute the correlation. 99 convicted offenders. In all cases the r1s and partial r's are virtually zero. The "zero-order" dyx's are somewhat larger, suggesting some relation betx<reen the race, class, and sex of the convicted offender and the size of the fine. However, when the partial analysis is examined, controlling for either BA level or number of past convictions, these relations drop to near zero. Not only are the coefficients quite small but the signs are inconsistent suggesting random fluctuations around zero. A second indicator of potential judicial discrimina tion is the number of days actually spent in jail by convicted offenders. The hypotheses suggest that after control is made for the seriousness of the offense and past convictions, there is no relation between race, class, sex, and age and the number of days spent in jail. The partial correlations and the partial-analysis dyx coefficients in Table 10 indicate that class, sex, and age are not related to the number of days spent in jail. Race, however, is slightly related to days-in-jail after controlling for the BA level or the number of past convictions, or both. The zero-order r between race and days-in-jail is .12. This coefficient is not diminished at all by any of the controls. The partial-analysis d coefficients are somewhat smaller J o b than the "zero-order" dyx of .19 but still are greater than zero. These results suggest that judges show a small 100 TABLE 10.--Correlations Between Number of Days in Jail on a Conviction for Drinking and Driving and Race, Social Class, Sex and Age Among Convicted Drinking Drivers, Controlling for Blood Alcohol Level and Total Convictions for Drinking and Driving Correlation Between Number of Days in Jail on a Conviction for Drinking and Driving and Social Characteristics Variable Controlled Race Occupa' tion - Educa tion Income Sex Age Somers' d -yx None .19 -.06 -.08 Pearson -.06 ' s r -.06 .02 None .12 .02 -.04 Somers' -.06 -.05 .07 Blood Alcohol Level:* High Medium Low .11 .03 .10 -.01 -.07 -.11 -.04 .01 - .01 -.05 .01 -.03 .16 .07 -.49 -.04 .02 -.01 Total Convictions for Drinking and Driving: 1 2 3 or more .03 .17 .16 - .08 .01 .10 -.01 -.02 -.03 Partial -.03 .01 -.10 r -.02 VoV ** -.05 .02 .03 Blood Alcohol Level .12 .04 -.03 -.04 -.05 .04 Total Convictions for Drinking and Driving .12 .02 - .04 -.06 -.05 .07 Blood Alcohol Level and Total Convic tions for Drinking and Driving .12 .04 -.03 -.04 -.05 .04 *A high blood alcohol level is! .20 and above. A medium blood alcohol level is .10 to .19. I A low blood alcohol level is .09 or less. **There were too few women in these categories to compute the correlation. 101 degree of discrimination against minority group members in assigning the number of days to be spent in jail but not in determining the fine paid for conviction for drinking and driving. Such a finding does not account for the overrepre sentation of blacks and Chicanos in the convicted category but it may measure the judicial behavior in the conviction decision. In any case, the low level of these associations does not indicate sufficient discriminatory behavior to account for the marked overrepresentation which was shown in the data reported earlier (see pages 79-82 in this chapter). FOOTNOTES 102 ^"This moderately high association between race and conviction is only slightly reduced when controlled for the effect of class. That is, a traditional problem is to separate the effects of class from the effects of race. In this case, using the three indicators of social class as controls the correlations between race and conviction are as follows: Low Medium High Education -.32 -.44 -.33 Occupation -.48 -.36 -.38 Income -.41 -.40 -.42 2 It may be that this correlation is, in part, spuri ous. That is, because racial minorities tend to hold low status occupations, the correlation between occupation and convictions is "really" due to the overrepresentation of blacks and Chicanos in these positions. This can be exam ined by holding race constant in correlating occupation and conviction. Thus, among Caucasians the correlation between occupation and conviction is -.10 (dyx with conviction dependent); it is -.15 among Chicanos and -.08 among blacks. All three of these are lower than the uncontrolled correla tion, suggesting that race did affect the relation. Note especially that among blacks occupation bears very little relation to conviction. Either nearly all blacks are in the same low status occupations, thus truncating the distribu tion or when a man is black it is such an important charac teristic that other factors such as occupation are over looked . 3 As with occupation, the association between educa tion and conviction may be spurious, both being influenced by race. Controlling for race the correlations between education and conviction drop. Among Caucasians the associ ation (dyx with conviction dependent) is -.13; it is 0.16 for Chicanos and among blacks the relation is only -.08. As before, blacks show the smallest relation between an indicator of class and conviction. ^As indicated above, this association between income and conviction may be due to the mutual influence of race. This can be examined by controlling for the effect of race. Among Caucasians the association (dyx with conviction dependent) between income and conviction is -.10; it is -.15 among Chicanos and this association is only -.07 among blacks. Thus, when the association between each of the indicators of class and conviction is controlled for race, ithe correlations drop. Also in every case blacks show the smallest association. CHAPTER V CONCLUSION - Summary This study demonstrates that minority group members, the lower class, males, and youths are more likely to be convicted of drinking and driving than are Caucasians, the upper class, females and the aged. These results are due to both discriminatory practices and actual offense behav ior. Furthermore, since there is greater discrimination in conviction than can be accounted for by the discriminatory behavior of police, district attorneys, and judges, it is concluded that the social structure external to the legal system functions to create discrimination in the arrest- conviction process. Beginning with the notion that criminal behavior and criminal conviction are basically different phenomena, which cannot be explained by the same conditions, this study has explored three possible explanations for criminal conviction. In the first explanation, conviction is seen to be the result of illegal behavior--the greater the seriousness and/or the frequency of violation the greater the conviction rate. Different rates of conviction between I 103 104 groups are, therefore, the reflection of different rates and degrees of involvement in illegal behavior. The second explanation asserts that conviction is determined by the social characteristics of alleged offenders rather than by law-violating behavior. Consequently, the greater convic tion rates of certain aggregates are the result of discrim ination, not the frequency or seriousness of law-violating behavior. The third explanation is a combination of the first two wherein both social characteristics and actual behavior influence the chance of conviction for drinking and driving. The first four hypotheses tested suggest only that, for whatever reason, ethnic minorities, the lower class, males, and the young are more likely to be convicted for drinking and driving than those not in these categories. The data support the first three hypotheses but the fourth hypothesis is not confirmed; age is not related to the chance of conviction. The question of which of the three possible explanations account for the greater conviction rates of these aggregates is next explored. The literature has tended to support the third explanation by showing that both the social characteristics of the alleged offender and his actual behavior appear to have a direct influence on conviction. Hypotheses 5 through 8 thus posit that the correlations between each social characteristic and conviction for drinking and driving will remain greater than 105 jzero after controlling for actual violating behavior. As predicted, both the partial correlations and the partial- analysis Somers' d remain greater than zero. This find- ing indicates that there is discrimination against certain aggregates in conviction for drinking and driving. The data also show the presence of some unanticipated statis tical interaction between social category and offense behavior in influencing conviction. This suggests that the third explanation for conviction is most appropriate. That is, offense behavior and social category combine to deter mine conviction. What this means is that for minority group members or the lower class to have the same proba bility of conviction as Caucasians or the upper class, they must offend less often or less seriously than members of either of the latter categories. Conversely, Caucasians i jand the upper class can offend more frequently or more seriously and still have the same chance of conviction as do racial minorities or members of the lower class. Several other studies have found that some social categories suffer conviction for illegal behavior beyond what their behavior would warrant. However, few have attempted to locate the source of the discrimination. The usual assumption is that the discrimination is a function of the operation of the legal system. In contrast to this assumption, several studies have found little or no discrimination being exercised by legal officials. 106 A p o s s i b l e a l t e r n a t i v e s o u r c e o f d i s c r i m i n a t i o n i s t h e s t r u c t u r e o f t h e s o c i e t y a t l a r g e . A n um ber o f f a c t o r s w e r e d i s c u s s e d i n c l u d i n g s u c h f a c t o r s a s p o l i c e d e n s i t y , c u l t u r a l r e s p o n s e t o a l c o h o l , a n d in c o m e d i s t r i b u t i o n , a l l o f w h ic h p o t e n t i a l l y a l t e r t h e c h a n c e o f c o n v i c t i o n . I f s u c h f a c t o r s do h a v e s u c h a n e f f e c t , t h e n i t w o u ld b e l e x p e c t e d t h a t t h e a c t i o n s o f l e g a l o f f i c i a l s w o u ld n o t a c c o u n t f o r a l l o f t h e d i s c r i m i n a t i o n i n c o n v i c t i o n . A s a p o i n t o f r e f e r e n c e i t i s h y p o t h e s i z e d t h a t o f f i c i a l s i n t h e l e g a l s y s t e m w i l l n o t d i s c r i m i n a t e i n t e r m s o f r a c e , s o c i a l c l a s s , s e x , o r a g e . T he e x a m i n a t i o n o f p o l i c e b e h a v i o r r e v e a l e d t h a t p o l i c e a p p a r e n t l y do n o t d i s c r i m i n a t e o n t h e b a s i s o f a g e o r s o c i a l c l a s s . T h e r e w a s , h o w e v e r , som e i n d i c a t i o n o f d i s c r i m i n a t i o n a g a i n s t C a u c a s i a n s a n d f e m a l e s . T h i s i s i n c o n s i s t e n t w i t h t h e d a t a w h ic h s h o w a l o w e r I p r o b a b i l i t y o f c o n v i c t i o n f o r t h e s e tw o c a t e g o r i e s . F u r t h e r a n a l y s i s i n d i c a t e d t h a t t h i s r e s u l t o c c u r s b e c a u s e p o l i c e s t o p m i n o r i t y g r o u p m em b ers a n d ( t o a l e s s e r e x t e n t ) f e m a l e s m o r e o f t e n f o r l e s s s e r i o u s o f f e n s e s , and i t i s t h e l e s s s e r i o u s o f f e n d e r s w ho a r e m o s t l i k e l y t o b e r e l e a s e d . I f o n l y s e r i o u s o f f e n d e r s a r e c o n s i d e r e d , i t i s f o u n d t h a t C a u c a s i a n s a r e m ore o f t e n r e l e a s e d b y p o l i c e t h a n a r e jmembers o f r a c i a l o r e t h n i c m i n o r i t i e s . T h i s l a t t e r f i n d i n g d o e s n o t h o l d f o r f e m a l e s . I t i s c o n c l u d e d t h a t p o l i c e do sh o w a s m a l l d e g r e e o f d i s c r i m i n a t i o n a g a i n s t m i n o r i t i e s and f e m a l e s ; t h i s r e f u t e s t h e h y p o t h e s i s o f n o d i s c r i m i n a t i o n 107 by police. These data indicate further that district attorneys do not discriminate on the basis of race, class, sex, or age in determining the legal charge to be brought against alleged drinking drivers. In no case was the correlation between a social characteristic and \diether the charge was reduced to reckless driving sufficiently greater than zero to refute the hypothesis of no discrimination. The data also indicate that judges are unbiased in terms of social class, sex, and age in determining the penalty for a conviction. None of these three variables was related either to the amount of the fine paid or to the number of days spent in jail by convicted drinking drivers. There was, however, an association between judicial severity and race, even after controlling for the seriousness of the I offense and the number of past convictions. Judges appar ently discriminate against blacks and Chicanos in assigning the sentence for a conviction. Twelve separate hypotheses have predicted no rela tion between a social characteristic and legal severity; nine of these are supported by the data. The three hypoth eses which are not supported show correlations in a direction that help account for the overrepresentation of males and especially minority group members in the convicted category. Overall, the relations between legal severity and race are not large enough to account for the correlation loaT1 !(d = .51) between race and conviction after controlling for actual offense behavior. The over-conviction of the lower class clearly is not accounted for by the behavior of legal officials. i The argument that the social structure contributes to discrimination in conviction is strengthened by this ! indication that the behavior of legal officials is not sufficient to account for all of the discrimination in !conviction. The possibility remains that all discrimination is generated within the legal structure, but the measures used here are not sensitive enough to test this assertion. Still, the possibility that factors outside the legal system influence the chance of conviction is reasonable and worthy of investigation. It is beyond the scope of this study to i examine each element of the social structure as a possible factor in discrimination, but it is within our range to speculate on the possible links between the social structure and conviction for drinking and driving. One general way that the social structure can affect the probability of conviction is by altering the probability of initial apprehension by police. The probability of apprehension may be influenced by a number of factors which affect the observability of behavior. People who are committing the same offense can differ in chance of appre hension because of the degree to which they are observable to police differs. Residence, for example, can affect the observability of drinking and driving behavior. At one extreme are people who live in very isolated areas where there are few if any police, spread over great distances. I Such people are not very observable and as a consequence they will have a very small probability of conviction for each drinking and driving offense. On the other extreme iare people who live in ghettos; they are more observable and have a greater chance of apprehension because there are more police who watch closely for infractions of the law. Residence itself is affected by socio-structural factors such as income, race, and occupation. It was suggested in Chapter II that one's response to alcohol may also alter one's observability. The more adequately a person can control his car when he is intoxi cated, the less observable he is. This may serve to explain I part of the discrimination experienced by younger people; they are less experienced drinkers and cannot "hold their liquor" as well as older, more experienced drinkers. Yet another factor that may affect observability is the time of day that driving after drinking occurs. Late at night, after the bars close, police may be more likely to observe a drinking driver because there are fewer people on the road and because the officer is more attuned to the possibility of a drinking driver. This study does not provide information on people who were arrested for drinking and driving but vdio were 110 I never convicted. Discrimination may occur when socio- structural factors influence the chance of dismissal of a case after apprehension has occurred. Once apprehended the alleged offender's ability to affect his own treatment differs depending upon several factors; the most obvious factor is the amount of money available for bail and attorney's fees. A related factor is the limited number of court personnel and the very crowded court calendars. In some cases the charge is dismissed completely because an attorney for the defendant indicates that he will demand a jury trial and tie up as much court time as possible if the charge is not dismissed. Of course, the people most likely to make use of such leverage are those who can afford skilled attorneys. Clearly, future research related to the issue of conviction and the legal process should include ! extra-legal factors in conceptualizing the problem and ] should develop procedures to test hypotheses derived from such conceptualizations. It was found in this study that the bases of discrimination interact statistically with the frequency and seriousness of drinking and driving behavior in deter mining conviction for drinking and driving. This indicates that the degree of discrimination against members of a particular group is not constant, but rather varies depend ing upon the frequency and/or seriousness of each person's offense. One way to explain why this occurs is to observe Ill that each person has a certain probability of apprehension and conviction each time he (or she) drives after drinking. This probability differs between individuals because they occupy different positions in the social structure (e.g., the offenses of some are more observable than the offenses of others. If, for example, the probability that a poor black will be convicted of drinking and driving is .1 and for a rich Caucasian the probability is .01, then in a given amount of time the black is ten times more likely to be convicted if each offends only once. However, in the same amount of time if each offends 50 times the chance that the black will be convicted is less than 10 times greater than that for the Caucasian. In general, as the number of offenses per unit of time increases, the probability of conviction asymptotically approaches one regardless of the conviction probability for a single offense. Stated other wise, regardless of the advantage that any individual has over any other in terms of the chance of conviction, that advantage will decrease as the number of offenses committed by each person in a unit amount of time increases. Future Research Future study in this area should explore the rela tive importance of actual violating behavior versus social categories in influencing conviction. There are a number of ways in which subsequent research could improve upon the procedures used in this study. The major problems encoun tered here can be divided into two categories; the first concerns the accuracy of the data and the second concerns the availability of measures. The problem of accuracy may be inherent where data gathering involves obtaining answers from respondents about apparently unacceptable behavior or behavior that may involve legal sanctions. This difficulty is exacerbated by the increasing attention being paid by citizens and politicians to the "invasion of privacy" issue. Interestingly, there were only two or three refusals among all DMV respondents, despite careful assurance by the interviewer that the subject could freely refuse to cooper ate. The setting adjacent to the official process of acquiring a drivers license may still have intimidated people but such intimidation does not guarantee truthful responses. Accuracy may be increased by separating data collection from the official location. This probably entails greater costs. Other possibilities for data gather ing include random checks of drivers on the road to obtain an estimate of the actual proportion of drinking drivers and their characteristics. Any time one asks a respondent to recall past behavior there is a chance of selective or faulty memory. Perhaps accuracy could be increased by choosing a panel of drivers to be followed over time. They could be asked to recount behavior that occurred recently and may come to have greater trust in the researcher. While this may introduce the biasing caused by the "Hawthorne effect," the importance of getting accurate measures of human behavior (especially that which is "deviant") should encourage researchers to experiment with a variety of techniques. The second major problem involves the lack of infor mation about much of what occurs within the legal system. We are forced to look at the end product of the legal process and then to infer what went on within the legal system. There is no direct information on who police stop and then release (or why they do so), nor is there any ! available data on those arrested but dismissed without being convicted. The most important advantage of this study-is that there are data on the actual offending population prior to the selective process of the legal system. In future research, effort should be made to obtain more detailed and complete data on the way each segment of the legal system functions, including the criteria for decisions, who goes where, who is dismissed, and so forth. Future research should also include several variables which were not employed in this study. The most obvious of these is the home address of the offender and the place of arrest. This j I information could then be plotted by census tract and combined with other tract measures. For example, with such data it would be possible to correlate arrest rates by tract with such factors as per cent non-white, class level, or police density. Another variable which should be included is the respondent's estimate of the number of miles he drives per year (or month). High mileage drivers may be more experienced and better able to drive when inebriated (thus decreasing observability). Alternatively, perhaps those vrtio drive only short distances after drinking are less apt to be detected than those who travel long distances under the influence of alcohol.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A Sociological Approach To The Etiology Of Female Homosexuality And The Lesbian Social Scene

PDF

The Deterrent Effect Of Criminal Justice Agencies On Felony Offenses

PDF

S.T.P.: A Simulation Of Treatment Processes

PDF

Structural Factors Affecting Fertility In Large United States Cities

PDF

Normative values of selected law enforcement officers and adult male offenders

PDF

Social-Psychological Factors In Organizational Attachment

PDF

Bam: An Innovative Change Model--Barriers Encountered In The Implementation Of A Classical Research Design To Modify The Behavior And Attitude Of Staff And Inmates In A Correctional Institution

PDF

Social Components Of Housing Cost In The Western Metropolis

PDF

Referential Dissociation And Response To Stress

PDF

The Veritism Of Hamlin Garland

PDF

An Analysis Of Factors In The Family'S Withdrawal Of A Patient From A Hospital For The Mentally Retarded

PDF

A Typological Study Of Juvenile Correctional Organizations

PDF

Rigidity Factors And Value Choices

PDF

Educational Expectations And Problems As Perceived By Headstart Parents And Teachers

PDF

The Philosophies Of Ralph Tyler Flewelling And Edgar Sheffield Brightman: A Comparison And A Critique

PDF

Social Factors Related To Dentistry As A Career

PDF

An Analysis Of The Positive And Negative Functions In The Contemporary Usage Of An Ancient Rite Of Passage

PDF

Status consistency and its effects on marital adjustment and stability

PDF

The Parameters Of Singleness: An Inquiry Into Some Factors Influencing The Choice Of Singleness Over Marriage As A Way Of Life

PDF

A Comparison Of Perceptions Reported By Nondelinquents And Delinquents Regarding Their Identification With Selected Socialization Agents And Normative Prescriptions

Asset Metadata

Creator Purdy, Ross Lloyd (author)

Core Title Factors In The Conviction Of Law Violators: The Drinking Driver

Degree Doctor of Philosophy

Degree Program Sociology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,sociology, criminology and penology

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Labovitz, Sanford I. (committee chair), Miller, Howard S. (committee member), Sklar, Bernard (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-468559

Unique identifier UC11362262

Identifier 7116432.pdf (filename),usctheses-c18-468559 (legacy record id)

Legacy Identifier 7116432.pdf

Dmrecord 468559

Document Type Dissertation

Rights Purdy, Ross Lloyd

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA