punishment and alcohol problems: recidivism among drinking-driving offenders

10
Journal of Criminal Justice 28 (2000) 261–270 0047-2352/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0047-2352(00)00047-7 Punishment and alcohol problems Recidivism among drinking-driving offenders Jiang Yu* New York State Office of Alcoholism and Substance Abuse Services, 1450 Western Avenue, Albany, NY 12203, USA Abstract The degree to which punishment reduces criminal recidivism has been extensively studied, though few efforts have examined the extent to which sanctions are effective for offenders whose criminal behaviors are a result of their alcohol and substance abuse problems. This study was conducted under the hypothesis that the effect of al- cohol problems and the effect of sanctions tend to negate each other. Sanctions reduce the chance of repeat drink- ing-driving offenses, while severe alcohol problems increase such chances. Analysis based on a sample of 521 persons who had prior arrests for drinking-driving offenses indicated that offenders’ alcohol problems are the strongest predictor of future recidivism. When alcohol problems were controlled for, punitive sanctions did not significantly decrease the chance of recidivism. Findings suggest that carefully screening drinking-driving of- fenders’ alcohol-related problems and providing effective treatment to those offenders who are in need of such services, constitute critical strategies to reduce drinking-driving recidivism. Similar strategies should be applied to offenders who committed alcohol/drug-related offenses other than drinking-driving. © 2000 Elsevier Science Ltd. All rights reserved. Introduction According to the deterrence theory, human beings are rational, and their decisions to violate the law are based on their perception of risks for sanctions and on their past experiences of punishment. When pun- ishment is swift, certain, and harsh, the chance for the general public to commit crimes and for criminals to repeat their offenses are low (see review by Liska, 1987; Nagin, 1998). For some crimes, however, cer- tain and harsh punishment does not seem to have the expected impact on recidivism (Paternoster, 1987). For example, even with the dramatic increases in the level of punishment, high recidivism rates are re- ported among drinking-driving (DD) offenders (Si- mon, 1992; Yu & Williford, 1991). The high DD recidivism rate has been a main con- cern of past research. One consensus, however, is that DD offenders, especially repeat DD offenders, tend to display more severe alcohol problems than other criminals. Snowden et al. (1986) reported that convicted DD offenders are likely to use alcohol fre- quently and in large quantities and tend to show signs of alcohol problems. Among known DD offenders, Gruenewald et al. (1990) noted a high likelihood of continued drinking after the consumption of one al- coholic beverage. Controlling high-risk driving be- havior, Yu and Williford (1993) examined the rela- tionship between problem drinking and repeat DD offenses and reported that problem drinking signifi- cantly increases the chances of DD recidivism. In a recent study, Myerholtz and Rosenberg (1997) found consistencies among the results of several alcoholism screening instruments from DD offenders, indicating a potential high prevalence of problem drinking. Another major aspect in DD recidivism research has been the impact of sanctions (see Piquero & Pa- ternoster, 1998). The three main types of sanctions for DD include: fine, confinement, and license with- * Corresponding author. Tel.: 518-485-7524; fax: 518-485-0261. E-mail address: [email protected] (J.Yu).

Upload: jiang-yu

Post on 02-Jul-2016

219 views

Category:

Documents


6 download

TRANSCRIPT

Page 1: Punishment and alcohol problems: Recidivism among drinking-driving offenders

Journal of Criminal Justice 28 (2000) 261–270

0047-2352/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved.

PII: S0047-2352(00)00047-7

Punishment and alcohol problems Recidivism among drinking-driving offenders

Jiang Yu*

New York State Office of Alcoholism and Substance Abuse Services, 1450 Western Avenue,Albany, NY 12203, USA

Abstract

The degree to which punishment reduces criminal recidivism has been extensively studied, though few effortshave examined the extent to which sanctions are effective for offenders whose criminal behaviors are a result oftheir alcohol and substance abuse problems. This study was conducted under the hypothesis that the effect of al-cohol problems and the effect of sanctions tend to negate each other. Sanctions reduce the chance of repeat drink-ing-driving offenses, while severe alcohol problems increase such chances. Analysis based on a sample of 521persons who had prior arrests for drinking-driving offenses indicated that offenders’ alcohol problems are thestrongest predictor of future recidivism. When alcohol problems were controlled for, punitive sanctions did notsignificantly decrease the chance of recidivism. Findings suggest that carefully screening drinking-driving of-fenders’ alcohol-related problems and providing effective treatment to those offenders who are in need of suchservices, constitute critical strategies to reduce drinking-driving recidivism. Similar strategies should be appliedto offenders who committed alcohol/drug-related offenses other than drinking-driving. © 2000 Elsevier Science

Ltd. All rights reserved.

Introduction

According to the deterrence theory, human beingsare rational, and their decisions to violate the law arebased on their perception of risks for sanctions andon their past experiences of punishment. When pun-ishment is swift, certain, and harsh, the chance for thegeneral public to commit crimes and for criminals torepeat their offenses are low (see review by Liska,1987; Nagin, 1998). For some crimes, however, cer-tain and harsh punishment does not seem to have theexpected impact on recidivism (Paternoster, 1987).For example, even with the dramatic increases in thelevel of punishment, high recidivism rates are re-ported among drinking-driving (DD) offenders (Si-mon, 1992; Yu & Williford, 1991).

The high DD recidivism rate has been a main con-cern of past research. One consensus, however, isthat DD offenders, especially repeat DD offenders,tend to display more severe alcohol problems than

other criminals. Snowden et al. (1986) reported thatconvicted DD offenders are likely to use alcohol fre-quently and in large quantities and tend to show signsof alcohol problems. Among known DD offenders,Gruenewald et al. (1990) noted a high likelihood ofcontinued drinking after the consumption of one al-coholic beverage. Controlling high-risk driving be-havior, Yu and Williford (1993) examined the rela-tionship between problem drinking and repeat DDoffenses and reported that problem drinking signifi-cantly increases the chances of DD recidivism. In arecent study, Myerholtz and Rosenberg (1997) foundconsistencies among the results of several alcoholismscreening instruments from DD offenders, indicatinga potential high prevalence of problem drinking.

Another major aspect in DD recidivism researchhas been the impact of sanctions (see Piquero & Pa-ternoster, 1998). The three main types of sanctionsfor DD include: fine, confinement, and license with-

* Corresponding author. Tel.: 518-485-7524; fax: 518-485-0261.

E-mail address

: [email protected] (J.Yu).

Page 2: Punishment and alcohol problems: Recidivism among drinking-driving offenders

262

J. Yu / Journal of Criminal Justice 28 (2000) 261–270

drawal. Some research has noted that longer periodsof license withdrawal tend to have a stronger impacton DD offenders’ recidivistic behavior than shorterperiods of license withdrawal (e.g., less than threemonths) (Homel, 1981; see also: National HighwayTraffic Safety Administration [NHTSA] and Na-tional Institute on Alcohol Abuse and Alcoholism[NIAAA], 1996). Fines appeared to reduce thechance of recidivism in European countries (Homel,1981; Votey & Shapiro, 1985), though relevant stud-ies in the United States—only a few available—gen-erally have indicated a lack of the effect of fines onDD offenders’ recidivism patterns (e.g., Yu, 1994;Yu & Williford, 1995; see National Highway TrafficSafety Administration (NHTSA) and National Insti-tute on Alcohol Abuse and Alcoholism (NIAAA),1996). Little significant evidence has been reportedin regards to the impact of confinement, since only asmall proportion of DD offenders receive jail sen-tences (e.g., National Institute of Justice [NIJ], 1983;Ross, 1992; see also: Nagin, 1998). In addition, Yu(1994) examined the effects of swift license and finesanctions. Results suggest that when license with-drawal is mandatory, swift and severe fine sanctionsmay reduce the chance of DD recidivism. Thereseems a lack of evidence supporting the effectivenessof punishment, though avoidance of punishment tendsto increase the chance of DD recidivism (Piquero &Paternoster, 1998; Yu & Williford, 1995).

One possible explanation for the lack of consis-tency in punishment impact for DD offenders hasbeen that these offenders tend to be heavy alcohol us-ers and demonstrate alcohol problems (Gruenewaldet al., 1990; Mann et al., 1996; Snowden et al., 1986;Yu & Williford, 1992); therefore, it is unlikely forthem to make rational choices not to drink and driveeven after being severely punished. That is, DD sanc-tions and DD offenders’ drinking behavior are twonegating factors for DD recidivism: the former de-creases recidivism, while the latter increases it. Bothfactors are critical in the study of recidivism patternsof DD offenders. Due to many data constraints, re-search on predicting DD recidivism has either fo-cused on offenders’ alcohol problems or their sanc-tions; therefore, little research has been able toexamine the effect of one while partitioning out theeffect of the other. In some attempts to include of-fenders’ problem drinking behavior while estimatingthe impact of punishment on their recidivism pat-terns, researchers employed such surrogate measuresas offenders’ blood alcohol concentration (BAC) lev-els at the arrest. Findings based on such surrogatemeasures do not appear to support the assumptionthat alcohol problems increase the involvement in

DD recidivism (Wieczorek et al., 1992; Yu & Willi-ford, 1995). The inconsistencies may suggest that ar-rests due to BACs, generally only including thoseabove the legal limit, are more of a legal indicatorthat a law has been violated than an indicator of prob-lem drinking. One recent effort by Taxman andPiquero (1998) focused on the relative merits of pun-ishing and rehabilitating DD offenders by incorporat-ing measures for both conditions in the analysis. Thefindings indicated that alcoholism rehabilitative pro-grams tend to reduce DD recidivism and sanctionstend to fail to deter recidivism, though the authorsalso acknowledged that one important factor—offenders’ previous alcohol/drug use—was not in-cluded in the analysis due to data limitations.

This study attempted to take into considerationboth sanctions and drinking behavior while estimat-ing the probability of DD recidivism. The marginaleffects of DD sanctions and offenders’ alcohol-related problems were examined with other factorsheld constant. Two general assumptions were devel-oped for this analysis. One, people fear punishment;therefore, DD sanctions tend to reduce offenders’ in-volvement in future offenses. Two, alcohol problemsare likely to increase the chance of involvement indrinking-driving; therefore, DD offenders who dem-onstrate more severe alcohol problems have a higherchance of recidivism than offenders who have lesssevere alcohol problems. Based on these two as-sumptions, the research hypothesis for this studystated that the effect of alcohol problems and the ef-fect of DD sanctions exert reverse impacts on DD re-cidivism: With sanctions controlled for, offenderswith more severe alcohol problems have a higherchance of recidivism; with problem drinking heldconstant, offenders who receive harsh sanctions havea reduced chance of recidivism.

Methods

Data

Data collected from an adult “high-risk” samplewere employed for this analysis. In order to select alarge enough proportion of subjects who have beeninvolved in DD and DD recidivism, the sample wasstratified into the following subgroups that representdifferent levels of involvement in alcoholism and al-cohol-related behaviors: people in county jails, onprobation, in alcoholism treatment centers, and indrinking-driver programs (DDP). The survey instru-ment was a self-administered questionnaire.

The survey was carried out between August andOctober 1989. Most of the respondents were in cap-

Page 3: Punishment and alcohol problems: Recidivism among drinking-driving offenders

J. Yu / Journal of Criminal Justice 28 (2000) 261–270

263

tive or restrained situations, therefore, various proce-dures had to be followed in order to insure voluntaryparticipation and the anonymity of the participants.In county jails, inmate service workers and creden-tialed alcoholism counselors (CAC) were designatedto administer the survey. The relationship betweenjail inmates and the inmate services and CACs werestrictly confidential and independent of the jail au-thority. The inmates could choose whether or not toparticipate in the survey without feeling coerced bythe jail authority. An inmate participant’s accountwould be credited $10 through the inmate services orthe CACs. In probation departments, personnel whowere not probation officers distributed and collectedthe questionnaire. A probationer received a question-naire to take home after a reporting session, then he/she would return a completed questionnaire, if he/shechose to participate, and be paid a fee of $10 in cashupon attending the next reporting session. CACs inalcoholism treatment centers assisted in administer-ing the survey. Clients were approached at the end ofa counseling session, and a fee of $10 was paid incash to each participant. Staff from New York State Di-vision of Alcoholism and Alcohol Abuse (DAAA) ad-ministered the questionnaire to DDP members. TheDAAA staff contacted DDP participants at the end ofdrinking-driving education classes, distributed the ques-tionnaire, and paid a fee of $10 in cash to each respon-dent. DDP instructors were not involved in the process.

High levels of security and confidentiality wererequired in the survey locations, and it was not possi-ble to acquire assistance and information necessaryfor random sampling designs. The size of the sampleis relatively large, the subgroups are well stratified,and the respondents have diverse backgrounds; there-fore, the analysis based on data from this survey mayprovide information and insights for further researchon alcohol/drug-related problem-prone populations.

A total of 878 respondents participated in the sur-vey. The sample included respondents from facilities/establishments located in the New York State coun-ties of Albany, Schenectady, Rensselaer, Washing-ton, and Suffolk. Of the sample, 26.7 percent wasfrom drinking driver programs, 24 percent fromcounty jails, 22 percent from probation, and 26.5 per-cent from alcohol treatment centers. In this study, thefour subgroups were not separated and were exam-ined as one sample. The group status of a subject atthe time of the survey may not necessarily have beenthe same at another point in time, because those whowere in jail during the survey may have been on pro-bation prior, and in alcoholism treatment afterwards.Analyses on the subgroups will be performed underspecific hypotheses in separate future studies.

For the purpose of this study, a subsample of 521respondents who had at least one DD arrest was se-lected. The data were collected a few years ago, al-though results based on the data should still be appli-cable to the current DD situation, because the lawsagainst DD and DD recidivism in New York Statehave remained unchanged since the late 80s and early90s (Appendix).

Measures

DD Recidivism

The dependent condition for this study was DDrecidivism, measured by two indicators. The first in-dicator was the total number of DD offenses(TOTDD)—respondents were asked to report thenumber of DD arrests they had up to the time of thesurvey. This study only involved those who had beenarrested at least once for DD, so the range of thisvariable was from 1 to 5. The second indicator wascomposed of three variables specifying the second(DD2), third (DD3), and fourth (DD4) offenses. Eachvariable had two values, 0 and 1 (dummy), indicatingwhether or not the respondent was arrested for a par-ticular DD offense. The fifth offense was not in-cluded in this analysis, because only a small numberof the respondents had been arrested for a fifth DDoffense (twenty-four persons). Among the 521 of-fenders, 44.9 percent had a second offense, 24.4 per-cent a third offense, and 10.7 percent a fourth of-fense.

Alcohol problems

Two measures were used for respondents’ alcoholproblems. The short version of the Michigan Alco-holism Screening Test (MAST) was employed to in-dicate problem drinking behavior. The scale wasoriginally composed of 13 of the 25 MAST items

1

;item 7 was omitted because of its lack of reliabilitywith other items (Hedlund & Vieweg, 1984). TheShort MAST Scale was a summated scale of 12 items(

a

5

.88), ranging from 0 to 12.The second indicator of respondents’ alcohol

problems was alcohol consumption. Beer and liquorwere the two most used alcoholic beverages by therespondents. The average number of drinks (onedrink is equivalent to 12 ounces of beer, a shot of li-quor, a glass of wine, or one wine cooler) consumedper drinking occasion by the survey respondents wasas follows: beer, 6.4; liquor, 2.4; wine, .9; wine cool-ers, .3; and other .3. Some studies have noted that thepreference for a particular alcoholic beverage associ-ates with certain behavioral patterns, for example, the“beer subculture” (Berger & Snortum, 1985; 1986;Douglass, 1982). The present analysis focused on

Page 4: Punishment and alcohol problems: Recidivism among drinking-driving offenders

264

J. Yu / Journal of Criminal Justice 28 (2000) 261–270

beer and liquor. The quantity and frequency index ofalcohol consumption (ALCOHOL) was a product ofthe sum of the number of drinks of beer and liquorconsumed on a typical drinking occasion, multipliedby the number of drinking days per month, and thendivided by thirty (total number of days in a month).This index reflected the average daily amount of al-cohol (beer and liquor) a respondent consumed.

Sanctions

The questionnaire asked respondents a series ofquestions regarding each of their DD offenses, in-cluding the dates of the offenses, outcomes of casedispositions, and sanctions. The impact of the sever-ity of sanctions could be estimated at various stagesof an offender’s “DD career.” Three sanction catego-ries were utilized: (1) whether or not offenders re-ceived jail sentences for the first to fourth offenses—JAIL1, JAIL2, JAIL3, and JAIL4—(due to the factthat only a small percentage of DD offenders re-ceived jail sentences, the original continuous mea-sures of number of days respondents were jailed foreach offense were recoded into dummies, where 1

5

having received a jail sentence, and 0

5

otherwise);(2) the number of months of license withdrawal, sus-pension and/or revocation, for the first to fourth of-fenses (LIC1 through LIC4); and (3) the amount ofthe fine for first, second, third, and fourth DD of-fenses (FINE1 through FINE4). The law specifies theamount of fines for DD offenses, so the fine variablestended to cluster around a few values. They were re-coded as: 1

5

$0, 2

5

from $25 to $240, 3

5

$250, 4

5

from $260 to $350, and 5

5

from $360 to $2,500.

Controls

DD recidivism patterns may be affected by condi-tions other than punishment and problem drinking;therefore, two types of controls were included. Yuand Williford (1995) reported that DWI offenderswho are able to delay or avoid convictions tend tohave a higher rate of recidivism. In New York State,driving while intoxicated (DWI) is defined as operat-ing a vehicle with a BAC at or above .10, and drivingwith ability impaired (DWAI) is defined as operatinga vehicle with a BAC between .05 and .09. A convic-tion of DWI is associated with more severe punish-ment than a conviction of DWAI (see Appendix formore information). Conviction type (CNVTPY1through CNVTPY4) was included as a legal control.As a dummy variable, it denoted whether or not therespondents were convicted of DWI (value 1) orDWAI (value 0). [There were four original categoriesfor types of convictions: DWI, DWAI, chargedropped, and other convictions. Very few cases were

in the categories of

charge dropped

and

other convic-tions

(i.e., 49, 11, 6, and 3, respectively, for the first,second, third, and fourth offenses); therefore thesetwo categories were dropped from the analysis.]

Conditional discharge represented the second le-gal control. Also a dummy, this variable indicatedwhether or not the respondents were discharged un-der such conditions as going through DD educationprograms or alcoholism treatment programs. Avail-able for the first to fourth offenses (CD1 throughCD4), this variable separated offenders who were pu-nitively sanctioned from those who might have re-ceived rehabilitative dispositions.

Two nonlegal controls were respondents’ gendersand ages at each DD arrest (ARSTAGE1 throughARSTAGE4). Past research indicated that males andyounger drivers are more likely to be involved in DDand DD recidivism (e.g., Yu et al., 1992). Table 1contains descriptive information about DD offenses,alcohol problems, and sanction variables. To sim-plify the table, sanction variables for the first time of-fense are presented.

Procedure

The analysis was carried out in two procedures.First, the ordinary least-squares (regression) methodwas used to estimate the effects of alcohol conditionsand punishment severity for the first DD offense onthe total number of DD arrests. This analysis wasconducted under the assumption that alcohol prob-lems increase the chance of a prolonged DD career(more DD arrests), and DD sanctions tend to reducethe likelihood of future offenses. Second, the logisticregression was employed to examine the effects ofalcohol problems and sanction severity for a specific

Table 1Descriptive information of the study variables

Variables Mean

SD n

TODD 1.85 1.16 521MAST 5.31 3.80 487Alcohol 3.77 5.39 463Fine1 (dollars) 205 202.85 511Lic1 (months) 5.45 12.92 505Arstage1 (years) 27.94 9.45 495

Dummy variables % yes

n

DD2 44.9 521DD3 24.4 521DD4 10.7 521CD1 48.2 484Cnvtyp1 18.2 470Jail1 12.0 511Gender 86.2 521

Page 5: Punishment and alcohol problems: Recidivism among drinking-driving offenders

J. Yu / Journal of Criminal Justice 28 (2000) 261–270

265

Tab

le 2

Cor

rela

tion

mat

rix

of th

e st

udy

vari

able

s

TO

DD

DD

2D

D3

DD

4M

AST

Alc

ohol

Lic

1Ja

il1Fi

ne1

Cnv

typ1

CD

1A

rsta

ge1

Gen

der

TO

DD

(

n

5

521)

a

DD

2.8

09**

(

n

5

521)

DD

3.8

73**

.620

**(

n

5

521

)D

D4

.781

**.3

84**

.597

**(

n

5

521

)M

AST

.407

**.3

47**

.383

**.2

64**

(

n

5

487

)A

lcoh

ol.2

20**

.149

**.1

83**

.180

**.5

05**

(

n

5

463

)L

ic1

.003

2

.016

.052

2

.042

2

.003

.013

(

n

5

505

)Ja

il1.0

32.0

41.0

64

2

.032

.152

**.0

46.0

49(

n

5

511

)Fi

ne1

2

.049

.011

2

.024

2

.110

*

2

.019

.018

.105

*.1

87**

(

n

5

511

)C

nvty

p1.0

04

2

.003

.020

.006

.143

**.0

81.0

50.1

60**

2

.019

(

n

5

470

)C

D1

2

.134

**

2

.109

*

2

.126

**

2

.073

2

.207

**

2

.095

*.0

27

2

.074

.037

2

.068

(

n

5

484

)A

rsta

ge1

2

.228

**

2

.233

**

2

.188

**

2

.126

**

2

.224

**

2

.190

**

2

.112

*

2

.094

*.0

22

2

.046

.117

*(

n

5

495

)G

ende

r.1

67**

.198

**.1

34**

.094

*.0

49.0

54.0

86.1

08*

.091

*.0

82

2

.061

2

.087

(

n

5

514

)

a

Sam

ple

size

s fo

r co

mpu

tatio

n ar

e pr

esen

ted

in p

aren

thes

es o

n th

e di

agon

al li

ne.

*

p

,

.05;

**

p

,

.01.

Page 6: Punishment and alcohol problems: Recidivism among drinking-driving offenders

266

J. Yu / Journal of Criminal Justice 28 (2000) 261–270

offense on the likelihood of a higher order offense.The logistic regression was performed under the as-sumption that while alcohol problems increase thechance of future DD offenses, harsh sanctions for aspecific DD offense are likely to decrease the chanceof involvement in the next DD offense. Table 2 con-tains zero-order correlations among DD recidivism,sanctions for first offense, and the control variables.

Results

Table 3 shows the results from the regressionanalysis. For the total number of DD offenses,MAST appeared to exert the strongest impact onTOTDD compared to other independent variables. Inaddition, this independent variable alone explained16 percent of the variance in DD recidivism. ALCO-HOL, however, did not have any significant effect onDD recidivism. The beta associated with ALCOHOLwas .027, and almost no variance in the dependentvariable could be explained by ALCOHOL.

None of the three sanction conditions—LIC1,JAIL1, and FINE1—showed an impact on TOTDD.Of the three sanction variables, where FINE1 ac-quired the largest beta (

2

.092), the coefficient wasnot statistically significant—the R

2

coefficient forFINE1 was .003 and hardly contributed any explana-tory power to the equation.

CD1 was statistically significant at the .05 level,though the size of its beta coefficient was relativelysmall (

2

.098) for any substantive significance. Both ofthe nonlegal controls, gender and age at the first DDarrest, produced statistically significant coefficients.The beta coefficient was

2

.146 for ARSTAGE1, indi-cating that the younger the offenders are at the time offirst arrest, the more likely they are to be involved infuture offenses. The coefficient for GENDER (

b

5

.128) suggests that men are more likely than women tobe arrested for multiple DD offenses. Each of the non-

legal controls contributed about 2 percent to the totalexplanatory power of the equation.

Table 4 presents findings from the logistic regres-sion analysis. Among the three logistic equations, re-sults were relatively more consistent for the secondand third offense equations, rather than for the fourth.MAST had a significant impact on the likelihood ofboth second and third offenses. ALCOHOL did notachieve a significant coefficient for either the secondor the third offense equations. None of the sanctionvariables was significant in the second and third of-fense equations. Two controls that produced statisti-cally significant coefficients were gender and age atthe time of arrest. The younger the offenders are atthe previous arrest, the higher their chances for sec-ond and third offenses. Males are more likely than fe-males to be arrested for second and third offenses.

Results from the fourth offense equation were lessconsistent. Of all the independent variables, onlyJAIL and CNVTYP produced statistically significantestimates. Offenders who are not sentenced to jail fortheir third offense are more likely to be involved in afourth DD arrest, and offenders who are not con-victed of DWI for their third offense are more likely(than those who are convicted) to be arrested for afourth offense.

Discussion

This article examines DD offenders’ recidivismpatterns taking into consideration both their sanctionsand their alcohol problems. The specific hypothesis forthis analysis noted that while DD offenders’ alcoholproblems increase their chances of future DD offenses,severe sanctions decrease such chances; therefore,both conditions need to be included in the analysis sothat the effect of one condition can be partitioned out,while the effect of the other condition is estimated.

The hypotheses were partially supported. DD of-fenders’ alcohol problems are by far the best predic-tors of their DD recidivism. DD offenders who reporthigh scores on MAST tend to be involved in moreDD arrests. Not only are alcohol problems effectivepredictors of offenders’ DD careers, they are also thestrongest predictors of specific DD offenses. Bothfirst- and second-time offenders with high MASTscores have a higher likelihood of recidivism. Incon-sistent with the results from the second and third of-fense equations, however, the coefficient for MASTdid not achieve statistical significance for fourth timeoffenses. MAST seems to have lost its predictivepower for fourth time offenses due to the lack of vari-ance. Multiple DD offenders, especially those withthree and four arrests, are likely to demonstrate se-

Table 3Results from the regression analysis (

n

5

415)

Variables

B

b

SE

R

2

change

MAST .149 .376 .022 .163**Alcohol .001 .027 .011 .001Lic1 .00004 .001 .004 .000Jail1

2

.001

2

.003 .166 .000Fine1

2

.076

2

.092 .039 .003Cnvtyp1

2

.226

2

.083 .130 .002CD1

2

.216

2

.098 1.070 .005*Arstage1 2.017 2.146 .006 .020**Gender .421 .128 .156 .017**

* p , .05, **p , .01.

Page 7: Punishment and alcohol problems: Recidivism among drinking-driving offenders

J. Yu / Journal of Criminal Justice 28 (2000) 261–270 267

vere alcohol problems; therefore, the variance inMAST tends to be smaller among third and fourthtime offenders than among first and second time of-fenders. The average MAST scores for higher orderoffenders increase, though the variance decreases.The average MAST scores for the first, second, third,and fourth time offenders were 5.31, 6.79, 7.95, and8.33, respectively; and the variances in the MASTscores were 14.45, 12.81, 10.2, and 10.67, respec-tively. Alcohol consumption has a moderate associa-tion with DD recidivism, though it cannot effectivelypredict DD rearrests when alcohol problems areincluded in the equation. The discrepancies betweenthe results on alcohol problems and alcohol useindicate that offenders with drinking problems, butnot those just using alcohol, have a higher rate of re-cidivism.

Inconsistent with the hypothesis, sanction vari-ables did not significantly reduce DD recidivism,when the effect of problem drinking was controlledfor. Only jail sentences showed a statistically signifi-cant effect for fourth time offenses. Third time of-fenders who received jail sentences were less likelyto be involved in a fourth offense than those who didnot receive jail sentences. As the data showed, 12,22, 41, and 35 percent of the first, second, third, andfourth time offenders, respectively, received jail sen-tences. The impact of jail sentences appears to be ir-relevant to the lower order offenses due to low incar-ceration rates. Multiple offenders, such as third timeoffenders, are more likely to receive jail sentencesthan first and second time offenders, and, therefore,jail sentences produce an effect for fourth time of-fenses but not for lower order offenses.

Of the two legal controls, whether or not the of-fenders are conditionally discharged of a DD offensetended to have a moderately negative relationshipwith DD recidivism. Those offenders who are not

conditionally discharged are more likely than thosewho are to be involved in DD recidivism. The expla-nation for this finding may be that DD offenders whoare not conditionally discharged tend to be more seri-ous offenders (e.g., offenders who show high arrestBACs, have multiple offenses, and cause injuries/fa-talities), and have more severe alcohol problems;therefore, they tend to have a high likelihood of re-cidivism. Offenders’ conviction types display a nega-tive effect for a fourth time offense. This coefficientsuggests that multiple offenders (the third time of-fenders in this particular situation) who are convictedof lesser charges are more likely to display future re-cidivism. This finding is consistent with results fromYu and Williford (1995) in that DD offenders whoare able to delay or avoid convictions are more likelyto be engaged in future recidivism.

A general pattern in this study indicates that maleoffenders and younger offenders have a higher chanceof recidivism than female offenders and older offend-ers, which is consistent with results from past studies.

Conclusion

This study made an attempt to control two negat-ing conditions, sanctions and problem drinking,while estimating the likelihood of DD recidivism.Findings indicate that the chance of DD recidivism isprimarily increased by offenders’ drinking problemsand that sanctions do not seem to effectively decreasethe chance of recidivism of DD offenders who dem-onstrate problem drinking behavior. Effectivelyscreening DD offenders’ drinking problems and pro-viding alcoholism treatment to offenders (who are inneed of such services) are the essential tasks in re-ducing DD recidivism. Another aspect reflected inthis study is that increased certainty of harsh punish-ment may prevent multiple offenders from further of-

Table 4Results from the logistic regression

2d (n 5 352) 3d (n 5 150) 4th (n 5 80)

Variables Lg(odds) SE Lg(odds) SE Lg(odds) SE

MAST .261** .042 .209** .003 .071 .091Alcohol 2.036 .027 2.009 .036 .013 .047Lic 2.008 .008 2.026 .020 .007 .021Jail 2.158 .405 2.075 .482 22.170** .641Fine 2.132 .094 2.051 .132 2.292 .163Cnvtyp 2.495 .320 .356 .406 21.646 .792CD 2.422 .253 2.643 .385 2.625 .646Arstage 2.051** .015 2.046* .022 2.021 .038Gender 1.410** .439 .355 .826 .896 1.085

For each equation, sanction variables for the previous offense are used.* p , .05; ** p , .01.

Page 8: Punishment and alcohol problems: Recidivism among drinking-driving offenders

268 J. Yu / Journal of Criminal Justice 28 (2000) 261–270

fenses. Jail sentences tend to be the least used sanc-tioning method for DD offenders, though theynegatively impact the chance of further DD offenses,when the possibility of a jail sentence is increased formultiple offenders. This analysis focused on DD re-cidivism, but the findings provide insight for other al-cohol/drug-related crimes as well. Punishment alonemay not effectively reduce the chance of recidivismof alcohol/drug-related crimes by offenders whoabuse or are dependent on alcohol and other sub-stances. Criminals who commit alcohol and drug-related offenses should be generally screened forsubstance abuse and dependence problems so thatsanctions and rehabilitative efforts can be effectivelyintegrated into their programs. This will produce thebest results in decreasing their chances of recidivism.

Finally, the author would also like to caution thereader about the limitations of this study. The analysiswas based on self-reported DD offenses, but the lackof validity and reliability of self-reported crime datahas long been noted. Respondents’ may choose not toanswer sensitive questions, and their memory decaymay contribute to measurement errors, creating bias inresults. There is no one remedy for the problems asso-ciated with self-reported crime data. More researchshould be conducted to gather information from vari-ous samples to further examine the relationship amongalcohol problems, sanctions, and DD recidivism. In

this analysis, the MAST scale was used to measuredrinking problems; however, other instruments maygenerate more detailed measures. In this regard, futureresearch may include the Diagnostic and StatisticalManual of Mental Disorders (American PsychiatricAssociation, 1994) to categorize problem drinkingbehavior into alcohol abuse and alcohol depen-dence. Such categorization should further explainthe relationships among sanctions, alcohol prob-lems, and DD recidivism.

Note

1. For a list of the items of the Michigan Alcohol-ism Screening Test (MAST), please see Seizer, 1971.

Acknowledgments

The author thanks Robin W. Shacket for technicalassistance. The data collection was supported in part bythe New York State Governor’s Traffic Safety Commit-tee with funds from the National Highway TrafficSafety Administration. The opinions, findings, and con-clusions expressed in this publication are those of the au-thor and not necessarily those of the National HighwayTraffic Safety Administration, New York State Gover-nor’s Traffic Safety Committee, or New York State Of-fice of Alcoholism and Substance Abuse Services.

Appendix:Penalties for driving while ability impaired and driving while intoxicated (Comparison of the 1978, 1981, 1992, and 1997/1998 Vehicle and Traffic Laws)

Penalty 1978 law 1981 law 1992 law 1997/1998 law

Driving while ability impaired(Section 1192.1)

1st offense Fine $0 min$50 max

$250Mandatory

$250 min$350 max

$300 min$500 max

Jail term (max) 15 days No change No change No changeLicense action 60-day

suspension90-daysuspension

No change No change

2nd offensea Fine $0 min$100 max

$350 min$500 max

No change $500 min$750 max

Jail term (max) 45 days 30 days No change No changeLicense action 120-day

suspension180-daysuspension

6-month minrevocation

No change

3rd offenseb Fine $0 min$250 max

$500 min$1500 max

No change $750 min$1,500 max

Jail term (max) 90 days No change No change 180 daysLicense action 6-month min

revocationNo change No change No change

Driving while intoxicated(Section 1192.2 and 1192.3)1st offense Fine $0 min

$500 max$350 min$500 max

No change $500 min$1,000 max

(Continued)

Page 9: Punishment and alcohol problems: Recidivism among drinking-driving offenders

J. Yu / Journal of Criminal Justice 28 (2000) 261–270 269

References

American Psychiatric Association. (1994). Diagnostic andStatistical Manual of Mental Disorders. 4th ed. Wash-ington, DC: American Psychiatric Association.

Berger, D. E., & Snortum, J. R. (1985). Alcoholic beveragepreferences of drinking-driving violators. J Stud Alcohol46, 232–239.

Berger, D. E., & Snortum, J. R. (1986). A structural model ofdrinking and driving: alcohol consumption, social norms,and moral commitments. Criminology 24, 139–153.

Douglass, R. L. (1982). Youth, alcohol, and traffic accidents.In U.S. Department of Health and Human Services (Ed.)Alcohol and Health Monograph No. 4, Special Popula-tion Issues (DHHS Pub. no. ADM 82-1193, pp. 197–223). Washington, DC: Government Printing Office.

Gruenewald, P. J., Stewart, K., & Klitzner, M. (1990). Alco-hol use and the appearance of alcohol problems amongfirst offender drunk drivers. Br J Addiction 85, 107–117.

Hedlund, J. L., & Vieweg, B. W. (1984). The Michigan Al-coholism Screening Test (MAST): a comprehensive re-view. J Operational Psych 15, 55–65.

Homel, R. (1981). Penalties and the drunk/driver: a study ofone thousand offenders. Aust & NZ J Crim 14, 225–241.

Liska, A. E. (1987). Perspectives on Deviance. EnglewoodCliffs, NJ: Prentice-Hall.

Mann, R. E., Smart, G. R., & Anglin, L. (1996). Alcohol-related measures as factors in traffic fatalities. J Stud Al-cohol 57(6), 646–651.

Myerholtz, L., & Rosenberg, H. (1997). Screening DUI of-fenders for alcohol problem: psychometric assessmentof the Substance Abuse Subtle Screening Inventory.Psych Addict Behav 11(3), 155–165.

Nagin, D. S. (1998). Criminal deterrence research at the out-set of the twenty-first century. Crime & Just 23, 1–42.

National Highway Traffic Safety Administration (NHTSA)and National Institute on Alcohol Abuse and Alcohol-ism (NIAAA). (1996). A Guide to Sentencing DUI Of-

fenders. Rep. no. DOT HS 808 365 (March). Washing-ton, DC: NHTSA.

National Institute of Justice (NIJ). (1983). Impacts of Man-datory Confinement for Drunk Driving on Criminal Jus-tice Operations. Pub. no. 100498. Washington, DC:U.S. Department of Justice.

Paternoster, R. (1987). The deterrence effect of the per-ceived certainty and severity of punishment: a review ofthe evidence and issues. Just Quart 4(2), 173–217.

Piquero, A., & Paternoster, R. (1998). An application ofStafford and Warr’s reconceptualization of deterrence todrinking and driving. J Res Crime & Del 35(1), 3–39.

Ross, H. L. (1992). Are DWI sanctions effective? Alcohol,Drugs & Driving 8(1), 61–69.

Seizer, M. L. (1971). The Michigan Alcoholism ScreeningTest: the quest for a new diagnostic instrument. Am JPsych 127, 1653–1658.

Simon, S. M. (1992). Incapacitation alternatives for repeatDWI offenders. Alcohol, Drugs, & Driving 8, 51–60.

Snowden, L. R., Nelson, L. S., & Campbell, D. (1986). An em-pirical typology of problem drinkers from the MichiganAlcoholism Screening Test. Addictive Behav 11, 37–48.

Taxman, F., & Piquero, A. (1998). On preventing drunkdriving recidivism: an examination of rehabilitation andpunishment approaches. J Crim Just 26(2), 129–143

Votey, H. L., & Shapiro, P. (1985). Cost effectiveness of al-ternative sanctions for control of drunken driving: theSwedish case. In S. Kaye & G. W. Meier (Eds.), Proceed-ings of the Ninth International Conference on Alcohol,Drugs, and Traffic Safety, (San Juan, Puerto Rico: DOTHS-806-8l4, pp. 1449–1466). Washington, DC: NHTSA.

Wieczorek, W. F., Miller, B. A., & Nochajski, T. H. (1992).The limited use of BAL for identifying alcohol problemsamong DWI offenders. J Stud Alcohol 53, 415–419.

Yu, J. (1994). Punishment celerity and severity: testing aspecific deterrence model on drunk driving recidivism. JCrim Just 22(4), 355–366.

Yu, J., & Williford, W. R. (1991). Calculating DWI/

Appendix:(Continued)

Penalty 1978 law 1981 law 1992 law 1997/1998 law

Driving while intoxicatedJail term (max) 1 year No change No change No changeLicense action 6-month min

revocationNo change No change No change

2nd offensec Fine $0 min$5,000 max

$500 min$5,000 max

No change $1,000 min$5,000 max

Jail term (max) 4 years No change No change Varies (accordingto penal law)

License action 6-month minrevocation

No change 1-year minrevocation

No change

a Prior conviction of driving while ability impaired, driving while intoxicated, or driving while ability impaired by drugswithin the past five years.

b Two or more prior convictions of driving while ability impaired, driving while intoxicated, or driving while ability im-paired by drugs within the past ten years.

c Prior conviction of driving while intoxicated, driving while ability impaired by drugs, vehicular assault, or vehicular man-slaughter within the past ten years.

Page 10: Punishment and alcohol problems: Recidivism among drinking-driving offenders

270 J. Yu / Journal of Criminal Justice 28 (2000) 261–270

DWAI recidivism with limited data: using state driverlicense file for drinking-driving research. J Drug Ed21(4), 285–292.

Yu, J., & Williford, W. R. (1992). Perception, style, andconsumption: specifying and testing a structural modelon problem drinking. J Drug Iss 22, 75–90.

Yu, J., & Williford, W. R. (1993). Problem drinking andhigh-risk driving: an analysis of official and self-reported

drinking-driving in New York State. Addiction 88(2),219–228.

Yu, J., & Williford, W. R. (1995). Drunk driving recidi-vism: predicting factors from arrest context and casedisposition. J Stud Alcohol 56(1), 60–66.

Yu, J., Essex, D. T., & Williford, W. R. (1992). DWI/DWAI offenders and recidivism by gender in the 80s: achanging trend? Int J Addictions 27, 635–645.