Policy Analysis

While the present political environment has grown less supportive of economic regulation or poli- cies that distribute economic wealth, policies

regulating individual behavior still abound. Whether to pro- mote public health or social order, public policies designed to discourage undesired individual behavior frequently frame such behavior as “sinful.” Research on morality poli- cies and the politics that surround them has focused on such issues as the death penalty (Mooney and Lee 1999), alcohol and illegal drug use (Meier 1994), and cigarettes (Meier and Licari 1997a).

Two questions are at the core of research on morality policy. First, what types of policies are likely to be adopted to discourage sinful behavior? Second, are these policies effective at modifying undesired behavior? The work of Schneider and Ingram (1990, 1993), as well as Meier (1999), provides a conceptual foundation to guide research addressing these questions. For instance, Schneider and Ingram (1993) argue that socially constructing those engaged in undesired behavior as “deviants” is likely to result in the adoption of punitive tools that employ sanc- tions. Similarly, Meier (1999) contends that morality politics likely lead to coercive tools that impose increased costs on sin through law enforcement. Furthermore, he concludes that these anti-sin policies are likely to fail because the behavioral assumptions they are built on are inappropriate for the primary target of these policies—the frequent sinner.

We consider policies aimed at reducing the incidence of drinking-and-driving as another example of morality policy. First, we consider the social construction of the drink-driver t a rget group and how it is an example of morality politics. Second, we examine the assumptions of deterrence theory that underlie the punitive tools frequently adopted to discourage drink-driving and suggest an alternative set of behavioral

assumptions to guide policy. Third, using data from a national s u rvey of 4,008 individuals that focused on attitudes and behaviors related to drinking-and-driving we examine the assumptions of deterrent policies as they relate to non-drink- drivers, occasional drink-drivers, and frequent drink-drivers. This individual-level analysis is useful for understanding the e ffectiveness of punitive tools because these policies are designed to modify behavior by increasing individual perc e p- tions of costs associated with drinking-and-driving.

EFFORTS TO DETER THE DRINKING DRIVER

It is estimated that the involvement of alcohol accounted for more than 300,000 injuries and nearly 16,000 fatalities (or 38 percent of all fatalities) in 1996 alone (Royal 2000) and that alcohol is the single largest contributing factor in fatal crashes (Evans 1991). Over the last few decades, states have increased legislative efforts to address the problem of drinking-and-driving and its tragic consequences, and they have employed numerous policy tools to reduce this risky behavior. These tools have included education campaigns, designated driver programs, and increased rehabilitation efforts, but most of the attention has been on the increased use of sanctions, such as license revocation, mandatory jail time, and mandatory minimum fines. The politics sur- rounding drinking-and-driving explain the increased use of punitive policies instead of preventive measures.

In policy debates, drinking drivers typically are por- trayed as social deviants. The focus is on the individual not the societal conditions that may contribute to the behavior. It is a fight against “drunk driving,” not “drinking-and-driv- ing.” Drinking drivers are portrayed as the “killer-drunk” who is morally flawed and has committed more than an ordinary traffic offense (Gusfield 1981; Homel 1988). They are repeat-offenders who engage in sinful behavior that poses a threat to us all, not upstanding individuals who happen to make a bad decision. These individuals are

53

Drinking-and-Driving in America: A Test of Behavioral Assumptions Underlying Public Policy

DAVID J. HOUSTON, UNIVERSITY OF TENNESSEE, KNOXVILLE

LILLIARD E. RICHARDSON, JR., UNIVERSITY OF MISSOURI, COLUMBIA

Since the 1980s, states have increasingly used sanctions to deter people from drinking-and-driving, but the effectiveness of these policies is questionable. The use of sanctions as policy tools rests on deterrence theory, but little is known about the appropriateness of its behavioral assumptions for the group targeted by policy— the drinking-driver. Employing a national survey of 4,008 respondents, we use logistic regression analysis to examine perceptions of punishment costs, the importance of individual versus societal costs, and policy pref- erences related to drink-driving. It was found that the perceptions of the punishment costs of drinking-and- driving are not consistent with basic hypotheses of deterrence theory. The results suggest that policies based on deterrence theory are likely to be least effective for the main target of these policies (frequent drink-driv- ers) and are likely to be unnecessary for non-drink-drivers. An alternative set of behavioral assumptions is pro- vided that more closely fit the results obtained.

Political Research Quarterly, Vol. 57, No. 1 (March 2004): pp. 53-64

responsible for their actions, and should be punished. After all, the only thing such an individual will respond to is strict punishment. Thus, the immorality of drinking-driving is the frame accepted by many in the policy debate (Haight 1985).

Public policy addressing drinking-driving fre q u e n t l y involves the use of sanctions that increase the certainty (e.g., illegal per se laws, sobriety checkpoints, enforcement crack- downs), severity (e.g., mandatory jail time, mandatory min- imum fines, vehicle confiscation), or celerity (e.g., adminis- trative license revocation) of punishment. Deviants are punished with sanctions and do not receive policy benefits (Schneider and Ingram 1993), so programs designed to help drinking drivers will receive much less attention and fewer resources than punitive policy tools. Consistent with the implications of a deviant social construction, a survey found that MADD chapter presidents believe that strong legal penalties and strict enforcement are the most effective coun- termeasures while prevention or treatment-oriented options are the least effective (Ungerleider and Bloch 1988).

This negative social construct largely shapes policy because activist groups wielding a deviant construction of drinking-driving are quite influential in policy debate (Homel 1988; Evans 1991; Meier 1994). As Meier (1999) hypothesizes about the adoption of morality policies, drink- ing-and-driving laws typically are adopted with little discus- sion of their merits (Ross 1985b; Gordis 1991; Meier 1994).

Despite the pervasiveness of the deviant social construc- tion in public policy debates, research findings on the impact of punitive drinking-and-driving laws are inconclu- sive. Some research has reported increases in traffic safety associated with policies that increase the certainty, severity, and celerity of punishment (Kenkel 1993; Neustrom and Norton 1993; Sloan and Githens 1994; Voas, Tippetts, and Fell 2000). In a meta-analysis of 125 academic and govern- ment studies, Wagenaar et al. (1995: 307) conclude: “All of the DWI control efforts were associated with reductions in drink-driving and traffic crashes.”1

Other analyses indicate that coercive tools have not reduced alcohol-impaired driving (Evans, Neville, and Graham 1991; Meier 1994; Ross and Klette 1995; Ruhm 1996). Whetten-Goldstein et al. (2000: 732) represent this collection of evaluation studies when they write: “On bal- ance, we are left with a pessimistic view about the influence of criminal laws in this context.”

Still other studies find that eff o rts to increase the deterre n t e ffect of policies produce initial reductions in drunken driv- ing, but the effects are temporary or short - t e rm in duration ( Voas and Hause 1987; Moskowitz 1989; Ross 1992; Chaloupka, Saff e r, and Grossman 1993). It is likely that the ability to increase perceptions of the probability of detection (i.e., punishment certainty) through accompanying publicity, not the severity of the punishment, creates this short - t e rm

e ffect. Ross (1985a, 1992) contends that perceptions of the likelihood of apprehension for drunken driving diminish for two reasons. First, individual experience indicates that initial p e rceptions are exaggerated. Second, eff o rts to increase cer- tainty of punishment are not maintained.

In sum, the results of research examining the efficacy of criminal laws designed to deter drink-driving have been mixed. Further, no sanction has been found to have a uni- form significant impact on alcohol-related traffic fatalities across studies (Whetten-Goldstein et al. 2000). It appears that empirical findings related to the impact of these policies are highly dependent on model specification (Ruhm 1996). To understand these mixed findings it is necessary to move from the aggregate-level to the individual-level to examine the appropriateness of deterrent-based policies for combat- ing drinking-and-driving.

BEHAVIORAL ASSUMPTIONS AND ANTI-SIN POLICIES

Schneider and Ingram (1993) contend that incentive policy tools employing sanctions are likely to be adopted when target groups are portrayed with a negative social con- s t ruction such as “deviants.” But are punitive policies likely to be successful in discouraging sin? In part i c u l a r, are fre q u e n t sinners likely to abstain from sin in the face of these sanc- tions? We contend that deterrent-based anti-sin policies are likely to have limited success in achieving this goal. In this section we delve into the assumptions of deterrence theory to explain why and develop an alternative set of behavioral assumptions that more accurately describes sinful behavior.

Behavioral Assumptions of Deterrence-Based Policy

Based on deterrence theory, policies that employ sanc- tions are characterized by the certainty, severity, and celerity of punishment (Schneider and Ingram 1990). Increasing the p robability of detection, the harshness of the penalty imposed, and the swiftness with which punishment is meted out increases the costs for the undesired behavior. The behavioral assumptions that underlie deterrence theory portray individuals as utility maximizers, who perform a cost assessment related to the sinful behavior. Once costs exceed benefits (or pleasure) individuals will choose to no longer engage in the act. Even though individuals vary in the level of costs that are required to discourage certain behaviors, all engage in such a cost calculus.

The heart of this process is the conscious perception of the punishment costs imposed by government sanctions (Gibbs 1975; Homel 1988). For deterrent policies to work, individuals must be aware of the legal punishment for engag- ing in the undesired behavior (Vingilis 1990) and perc e i v e the cost of this punishment to be significant. It is only when the costs are perceived to exceed the utility gained from sin that individuals will abstain. This further assumes that indi- viduals are capable of controlling their behavior and of making choices among behavioral alternatives based on a cost assessment (Vingilis 1990). Thus, the challenge posed

54 POLITICAL RESEARCH QUARTERLY

1 They temper their conclusion by noting that the studies examined were of uniform poor quality such that a “. . . small proportion of the total lit- erature on DWI control is of significant use to scientists, professionals, and public policy decision makers” (313).

for crafting policy is to estimate the utility that sinners derive f rom sin and to set punishment costs at a higher level.

Two key behavioral hypotheses are derived from this understanding of the specific effect of deterrence. First, the more aware one is of government sanctions on sin, the higher one perceives the punishment costs of sin to be. Second, the higher one perceives the punishment costs of sin to be, the less likely one is to sin.

Alternative Behavioral Assumptions of Sin

While deterrence theory portrays individuals as utility maximizers, sinful behavior is more complex than this approach suggests. For instance, Meier (1999: 687) argues that a deterrent-based anti-sin policy “will fail because the policy does not consider the implications of the heterogene- ity of demand for sin” that exists among individuals. To illustrate his point, Meier presents three archetypes that are characterized by different slopes in their demand curves for sin: non-sinners, occasional sinners, and frequent sinners.2

Due to different demand elasticities, each group will have a very different response to the same price shift instituted by a new policy.

Non-sinners derive little or no utility from the sinful behavior and thereby either have no demand for sin or have highly elastic demand curves. In the latter case, these non- sinners will be highly responsive to the price shift from gov- ernment sanctions that can reduce the quantity demanded to zero. Occasional sinners gain some utility from the sinful activity and possess moderate demand elasticity. These indi- viduals will be moderately responsive to government sanc- tions and will sin less as punishment costs increase. Essen- tially, occasional sinners act in a manner consistent with deterrence theory assumptions. They will engage in the undesired behavior when it is easy to do so with few conse- quences (Meier 1999).

In contrast, frequent sinners derive a higher level of util- ity from the sinful behavior than do others. These individu- als also possess inelastic demand curves that make them almost impervious to the price changes caused by policy tools (Meier and Licari 1997b). Hence, frequent sinners are likely to be the least responsive group to increases in the cost of sin. Coupled with a higher utility for sin, the inelas- tic nature of their demand curve means that large increases in costs in the form of government sanctions will achieve only a small change in behavior. This suggests the impracti- cality of using government sanctions alone to modify the behavior of this group.

As additional deterrent-based policies are adopted to increase the costs of a particular sinful behavior, more of the non-sinners and occasional sinners reduce demand for sin. Each new law focuses attention on a smaller portion of the population still engaged in the behavior. Frequent sinners

will become an increasingly larger portion of the remaining target group (Meier 1999). Therefore, government will have to expend more resources for each marginal unit of sin reduction to try to change the behavior of individuals who are less and less likely to respond to punitive policy tools. For this reason, Meier (1999) argues that anti-sin policies are likely to fail and be very expensive policy failures.

In addition to the elasticity of the demand for sin, we extend Meier’s treatment by considering two other character- istics that distinguish these three archetypes of sinners. We contend that each archetype considers diff e rent types of costs in deciding whether to sin or not and that each possesses dif- f e rent levels of knowledge about relevant public policies. D e t e rrence theory focuses on the punishment costs of sin imposed by government (e.g., loss of freedom, fines and legal expenses), but sinful behavior also incurs personal and soci- etal costs. Personal costs are those borne by the individual that are the direct result of sin (e.g., personal financial loss, guilt, health risk to the individual, and loss of social status). Societal costs are borne by others or the community (e.g., financial losses to others not engaged in the sinful behavior, health risks to innocents, and violation of social norms).

Coupled with little or no demand for the sinful behavior, non-sinners are very sensitive to the individual and societal costs associated with sin. For non-sinners, government sanctions are irrelevant as their perceptions of the personal and societal costs associated with sin are sufficient justifica- tion for avoiding sin. While occasional sinners also recog- nize both the personal and societal costs associated with sin, the sum of these two costs is often perceived to be less than the utility derived from sin. For these individuals, punish- ment costs are a key factor in the decision of whether to sin or not, and likely make the difference in their decision.

Drawing on social control theory, it is the interactive eff e c t between personal/societal costs and the threat of punishment costs that deter the occasional sinner (Sherman 1984). The recognition and consideration of personal and societal costs indicates a connection to mainstream society. These individ- uals possess a stake in social conform i t y. However, it is not the government sanction alone that gets the attention of this individual, rather it is resulting personal and social costs that one bears for having been punished (e.g., loss of status, loss of employment). For this reason “more socially bonded people are more deterrable” (Sherman 1984: 78) for “[i]t is shame which doubles most punishments” (credited to Durkheim in Lukes and Scull 1983: 62).

In contrast, frequent sinners are less likely to possess the same connection with mainstream society. Fre q u e n t l y engaging in behavior labeled “immoral” indicates that fre- quent sinners possess less of a stake in social conformity and less acceptance of the values embraced by mainstream society. For this reason they assign little weight to societal costs and are unlikely to believe the message that sin poses significant personal risk. Because of an inelastic demand curve and a high utility for sin, it is also unlikely that pun- ishment costs can effectively modify their behavior. Fre- quent sinners could be aware of government policies,

DRINKING-AND-DRIVING IN AMERICA 55

2 Meier (1999) labels the three archetypes: nerds, vicarious sinners, and perverts. Our labels are broader in application but reflect similar charac- teristics as Meier’s archetypes.

believe the punishment costs to be high, and yet still sin. While frequent sinners may be aware of existing sanctions, they perceive that anti-sin policies unjustly target them and unfairly label their lifestyle as “criminal.” This is one reason why frequent sinners possess an inelastic demand curve.

A subgroup of frequent sinners may not perceive or weigh the costs of sin in a manner consistent with deter- rence theory as sinning may be pathological. Whether due to an emotional or psychological disorder, they may be inca- pable of rationally calculating the costs and benefits of sin or be unable to control their behavior. Punishment costs play no role in the decision to sin. Such pathological behav- ior may be especially prominent when the undesired behav- ior involves addictive substances.

The distinction between occasional and frequent sinners is supported by research on the deterrent effect of manda- tory arrest policies for domestic violence. These policies have been found to deter subsequent episodes of violence for those tied to conventional society (e.g., through employ- ment). For individuals lacking these ties, mandatory arrest policies actually were associated with increased levels of vio- lence following the initial arrest (Berk et al. 1992; Pate and Hamilton 1992; Sherman et al. 1992).

Therefore, anti-sin policies based on deterrence theory do not fit the behavior of all segments of the population. Coercive tools are unnecessary for the non-sinner, who is responsive to the authoritative or educational component of policy that teaches “the right thing to do.” Conversely, increasing the costs of sin through government-imposed sanctions is likely to have little influence on the frequent sinner, the typical target of policy. It is only the occasional sinner whose behavior is consistent with the assumptions of deterrence-based policy tools.

Finally, knowledge of anti-sin policies is likely to vary across the three archetypes. Research generally has found citizens to be uninformed about public policy and politics (Sniderman 1993; Delli Carpini and Keeter 1996). Explana- tion of decisionmaking by an uninformed public rests on the theory of the individual as a “cognitive miser” who relies on information shortcuts (Sniderman, Brody, and Tetlock 1991; Lupia and McCubbins 1998). The opinions of family, friends, coworkers, and others with whom one comes in contact regularly provide these information shortcuts.

Social networks are the linkages that develop between individuals that provide lasting patterns of relationships and serve as channels for the flow of information, attitudes, and other resources (Wasserman and Galaskiewicz 1994). Polit- ical scientists have documented the importance of social networks for the acquisition of information about politics (Huckfeldt and Sprague 1987; Huckfeldt et al. 1995), public goods (Schneider, Teske, and Marschall 2000), and public policy (Roch, Scholz, and McGraw 2000).

Because they belong to different social networks, mem- bers of the three archetypes likely differ in the amount of knowledge possessed about public policy directed at unde- sired behavior. Non-sinners likely belong to social networks that have little tolerance for sin. Members of these networks

are not likely to know individuals who have personally experienced the imposition of sanctions for sinning. Also, public policies designed to deter sin are not relevant for members of this archetype. Therefore, information about anti-sin policy is unlikely to be transmitted through the social networks to which non-sinners are connected.

In contrast, occasional-sinners are connected to social networks that tolerate some level of sin. These individuals a re more attuned to the punishment costs imposed by gov- e rnment and are more likely to at least know of someone who has had a personal encounter with law enforc e m e n t over sinful behavior. For this reason, occasional-sinners will possess more awareness of anti-sin policies than non-sinners.

The archetype likely to possess the greatest knowledge of anti-sin policies is the frequent-sinner. Due to their propen- sity to sin, these individuals have social networks where sin- ning is accepted and even valued. Thus, frequent-sinners are more likely than others to have first-hand experience with law enforcement regarding sin and to know others with similar experiences. The result is a social network that is relatively rich in information about anti-sin policies.

Based on the above portrayal of non-sinners, occasional sinners, and frequent sinners, several behavioral hypotheses a re off e red. First, frequent sinners are likely to be more aware of anti-sin policies than either non-sinners or occasional sin- ners. Second, frequent sinners are more likely to sin in the p resence of high punishment costs than occasional sinners and non-sinners. Third, frequent sinners are likely to be less responsive to government imposed punishment costs than either non-sinners or occasional sinners. Fourth, fre q u e n t sinners are less likely to recognize the personal and societal costs of sin than are non-sinners or occasional sinners. F i n a l l y, we expect that occasional sinners will always fall between these other two groups in their attitudes.

Implications for Policy Preferences

After a target group has been successfully constructed as “deviants” and their behavior as “sinful,” few publicly stand up and support members of this group. Because no amount of sin can be tolerated, advocates of anti-sin policy will push for more stringent laws to protect cherished values. The lack of public opposition to more stringent policy means that anti-sin laws are adopted with little debate. “Politicians, there f o re, always per- ceive there is greater support for being tougher on sin than t h e re really is” (Meier 1999, 683). The result is that politicians seeking the median stated pre f e rence for sin erroneously seek m o re extreme policies than the public truly wants.

Variations in the perceptions of problem severity and policy preferences are likely to exist across the three arche- types. Non-sinners are the least tolerant of sin and likely view even small amounts of sin as troubling. Members of this archetype are likely to perceive the severity of the p ro blem posed by sin to be greater than others, and will be more supportive of harsher sanctions.

In contrast, frequent sinners are less troubled by the presence of sin and will be less likely to perceive any related

56 POLITICAL RESEARCH QUARTERLY

problems as severe. As a result, frequent sinners will be less supportive of public policies that increase the punishment costs for engaging in sinful behavior. Somewhere between the non-sinner and the frequent sinner are the perceptions and policy preferences of occasional sinners. It is for these reasons that policies sought by advocates and politicians are likely to be more extreme than public preferences dictate.

METHODOLOGICAL ISSUES

The data for this project come from the National Survey of Drinking and Driving Attitudes and Behavior conducted by the Gallup Organization from September 25 to November 27, 1995 (NHTSA 1996). Telephone interviews averaging 20 minutes in length were administered to a national sample of 4,008 individuals using a stratified random digit dialing sample design. Young adults (16 to 20 years of age) w e re oversampled and comprise 23.6 percent of the unweighted sample.3 The survey was sponsored by the National Highway Traffic Safety Administration (NHTSA).

The analysis presented below focuses on self-reported drinking-and-driving behavior. An individual who reported to have driven within two hours of drinking any alcohol in the past year is considered a drink-driver. It is important to focus attention on all drink-driving behavior, not just drunken driving, for three reasons. First, the aim of public policy typically is to discourage the consumption of any alcohol before operating a motor vehicle, not merely exces- sive consumption.

Second, the consumption of any alcohol increases the risk of a traffic crash. For instance, Zador (1991) estimates the risk of a fatal accident for drivers with BAC levels of 0.05 to 0.09 to be 9 times greater than for drivers with no alco- hol in their blood, 50 times greater for those with BACs between 0.10 and 0.15, and 300 to 600 times greater for those with BACs in excess of 0.15. While providing lower estimates of risk, Levitt and Porter (2001) nonetheless indi- cate that even moderate alcohol consumption poses safety consequences. They conservatively estimate that the risk of causing a fatal crash is over 7 times higher for all drinking drivers than non-drinking drivers. This risk rises to being 13 to 14 times greater for those generally considered legally drunk (BAC > 0.10).

T h i rd, difficulties measuring behavior are lessened by focusing on self-re p o rted drinking-and-driving rather than d runken-driving. It is far more difficult for one to assess whether they “drove while impaired” or “while drunk” than it is to recall that they consumed alcohol before driving a motor vehicle. Additionally, social desirability bias is much less of a c o n c e rn when re f e rring to “driving after drinking” than it is for “drunk driving.” To the extent that such a bias influences s e l f - re p o rts of drink-driving, it should not undermine the identification of the archetypes. Because they may ignore this social desirability bias, it is even more apparent that those

falling into the extreme category of the frequent drink-driver re p resent a group that differs from the rest of the population.

To classify individuals according to our typology, it was necessary to operationalize an individual’s preference for drinking-and-driving. This was done by using self-reports of the frequency of driving within two hours of drinking any alcohol in the past year.4 Non-drink-drivers are individuals reporting that in the previous year they never drove within two hours after drinking. The other two categories were dis- tinguished using two assumptions: occasional drink-drivers outnumber frequent drink-drivers in the population and frequent drink-drivers are extreme in their propensity to drink-and-drive. The frequency distribution for this vari- able indicated a clustering of individuals who drove after drinking one to six times during the previous year. We labeled this group as occasional drink-drivers.5 The remain- ing respondents who did so on more than six occasions are classified as frequent drink-drivers.

Three sets of dichotomous dependent variables are used in the estimated models. Because of heavily skewed response distributions, multiple response categories were collapsed to create these dichotomous variables. One set measures individual perceptions about the costs imposed by public policy for driving after drinking. One variable in this category addresses the respondent’s knowledge of state drinking-and-driving laws. Knowledge of punishment is measured by whether a respondent correctly identified the state’s legal blood alcohol concentration (BAC) limit for evi- dence of guilt for driving under the influence of alcohol.

Punitive policy tools increase the costs of sinful behavior by increasing the certainty, severity, and celerity of punish- ment. Certainty pertains to how likely drink-driving behav- ior is to be detected by police and, if detected, how likely it is to receive punishment. Thus, two variables are created to measure perceptions of the certainty of punishment using the following question responses: it is almost certain or very likely that you will be stopped by police for driving after drinking too much, and punishment is almost certain if charged with breaking the law. Perceptions of the severity of punishment are measured by the following question response: punishment for drinking-and-driving is most likely to be very severe.6

A second set of dependent variables addresses the re l a t i v e i m p o rtance of policy-imposed punishment costs versus indi- vidual and societal costs of sin. For this assessment, thre e variables were used with the following responses: driving after drinking is more likely to result in being stopped by police than having an accident, drinking-and-driving is a

DRINKING-AND-DRIVING IN AMERICA 57

3 The unweighted sample is used in this study.

4 NHTSA (1996) defines a drinking-driver as one who has driven within two hours of consuming alcohol.

5 We tested the effect of various cutoffs for categorizing individuals as occa- sional drink-drivers. Classifying individuals who had re p o rted driving within two hours of drinking in the previous year 1-9 times (rather than 1-6 times) generated results similar to those re p o rted in this analysis.

6 The survey does not contain questions that pertain to perceptions of the celerity of punishment.

major threat to personal safety, and drinking-and-driving by n o n - p roblem drinkers is a serious highway safety issue.7

A third set of dependent variables addresses attitudes t o w a rds policy tools for dealing with drink-driving. The four variables in this category are: current laws are very eff e c t i v e , it is very important that something be done to reduce drink- ing-and-driving, sobriety checkpoints should be used more f re q u e n t l y, and penalties should be much more severe .

Due to the dichotomous nature of the dependent vari- ables, we used logistic regression analysis. We included an age variable and several dichotomous variables as controls in the models below: male, white, single, education (“some college or more”), and children (“a child 20 years or younger in the house”). Finally, a dichotomous independent variable is included in the models to control for whether an individual drinks alcohol regularly. Individuals were classi- fied as whether or not they are regular drinkers (“drink alco- hol at least 2-3 days per month”).8 It is important to ensure that the classification of individuals according to the utility they receive from drinking-and-driving is not merely an artifact of the frequency they consume alcohol. Thus, the inclusion of this variable provides a stiffer test of the behav- ioral assumptions described above.

RESULTS

After eliminating missing data, the final sample consists of 3,973 individuals. The distribution of individuals by target group is: non-drink-drivers, 3249; occasional drink- drivers, 523; and frequent drink-drivers, 201. On average, occasional drink-drivers reported driving within two hours of drinking 2.7 times in the previous year, compared to an average of 26.4 times for frequent drink-drivers. It is clear that frequent drink-drivers differ significantly from occa- sional drink-drivers in their propensity to drink-and-drive. The discussion of the logistic regression results is organized according to the three categories of dependent variables: perceptions of punishment costs, perceptions of individual and societal costs, and policy preferences.

Perceptions of Punishment Costs

Four logistic regression models were estimated to exam- ine individual perceptions of punishment costs imposed by drinking-and-driving laws and are presented in Table 1. These individual perceptions are the key to understanding how the threat of sanctions is assumed to influence behav- ior under deterrence theory. Only when individuals perceive the punishment costs for banned behavior to be high rela- tive to the pleasure it affords them will they not sin. This

presumes that individuals are aware of the legal sanctions for being caught engaging in a banned activity. If one is to be deterred, then one needs at least some minimal informa- tion about the laws that impose legal sanctions.

In the case of drinking-and-driving, one of the most basic legal concepts for a driver to be aware of is the BAC level that is evidence of guilt for DUI. One might hypothesize that individuals drink-and-drive because they are unaware of the laws restricting such activity, but the results in the first column of Table 1 show that both groups of drink-drivers are significantly more likely to know the legal BAC level of their state than non-drink-drivers. Both groups are more than one and a half times more likely to correctly identify the state BAC level. Incidentally, only 14 percent of non-drink-drivers c o rrectly identified the legal BAC limit for their state, com- p a red to 27 percent and 31 percent of occasional drink-driv- ers and frequent drink-drivers, re s p e c t i v e l y.

P e rhaps individuals who drink-and-drive, even though m o re aware of the legal BAC of their state, do not pro p e r l y p e rceive the nature of the punishment they will receive if caught driving after drinking too much. Altern a t i v e l y, it is possible that non-drink-drivers rely on environmental cues to judge the punishment costs of drink-driving. In either case, deterrence theory assumes that an increase in the per- ception of the certainty and severity of punishment will deter undesirable behavior. Here perceptions of the certainty of punishment are measured in two ways. The second model in Table 1 examines perceptions of how likely drinking-and- driving behavior will be detected by law enforcement. This model indicates that occasional drink-drivers and fre q u e n t drink-drivers are significantly diff e rent from non-drink-driv- ers in their assessments of the likelihood of being stopped by the police after drinking too much. Frequent drink-drivers and occasional drink-drivers are about half as likely as non- drink-drivers to think it is “almost certain” or “very likely” they would be stopped.9 This finding is consistent with d e t e rrence theory because a lower subjective probability of being caught reduces the cost of engaging in the behavior so individuals will drink-and-drive more often.

Although there is a statistically significant difference in the perception of certainty of detection among the groups, only 29.5 percent of non-drink drivers think it is “almost certain” or “very likely” to be stopped. This small percent suggests that perceptions of certainty do not fully explain differences in the behavior of the three groups.

F u rt h e rm o re, individuals who frequently drink-and- drive are expected to have lower subjective probabilities of being stopped by police than those who only occasionally drink-and-drive. The similarity of the coefficients and odds ratios for the occasional drink-driver and frequent drink- driver variables is not consistent with this hypothesis of deterrence theory. In this case, deterrence theory cannot explain the different levels of drinking-and-driving between the two groups with similar assessments.

58 POLITICAL RESEARCH QUARTERLY

7 For each question a value of 1 was assigned for the response identified in the text and a 0 for all other responses.

8 This concept was also operationalized using dummy variables for “fre- quent drinker” (at least one or two days a week) and “heavy drinker” (at least three or four days a week) but these variables did not produce results significantly different from those reported. 9 Overall, 27 percent of the total sample offered one of these responses.

Certainty of punishment can also be represented by per- ceptions of the likelihood of being punished if arrested. If one believes that punishment is likely, then one should be less likely to drink-and-drive. The results in the third column of Table 1 conflict sharply with this assumption. Both occasional drink-drivers and frequent drink-drivers are significantly more likely than non-drink-drivers to believe that punishment is “almost certain” if charged with breaking the law.10

If these drinking drivers know punishment is likely, then perhaps the issue is the severity of the punishment they will receive. Coercive tools are based on the assumption that as one increases the severity of punishment the deterrent effect of policy increases. Clearly, the results in column 4 of Table 1 contradict this hypothesis. Occasional drink-drivers are likely to believe punishment will be “very severe” but are

not significantly different from non-drink-drivers in this assessment (i.e., an insignificant coefficient for occasional drink-drivers).11 The coefficient for frequent drink-drivers, however, varies considerably from the expectations of deter- rence theory. Frequent drink-drivers are significantly more likely to believe that punishment will be very severe. They know it will be costly, and yet they still drink-and-drive. The behavioral assumptions of coercive tools are not con- sistent with the idea that the individuals who most readily identify the severity of punishment will also be the ones who engage most frequently in the banned activity.

While generally contradicting deterrence theory, these results are more consistent with the alternative theory of sinful behavior developed above. This alternative view p redicts that frequent drink-drivers will be more knowl- edgeable of DUI policies than non-drink-drivers. Addi- t i o n a l l y, frequent drink-drivers are expected to have higher

DRINKING-AND-DRIVING IN AMERICA 59

; TABLE 1 LOGISTIC REGRESSION MODELS OF THE PERCEPTIONS OF PUNISHMENT COSTS

Almost certain Knows the Likely to be to receive Punishment legal BAC stopped punishment very severe

Constant –0.732** –1.849*** 0.238 –1.533*** — — — —

Male 0.726*** –0.108 0.134* 0.123* (2.066) (0.898) (1.143) (1.131)

Single –0.184 0.390*** 0.061 0.162* (0.832) (1.477) (1.063) (1.176)

White –0.786*** 0.493*** –0.034 0.517*** (0.456) (1.636) (0.966) (1.676)

Age –0.017*** 0.009*** –0.012*** 0.002 (0.983) (1.009) (0.988) (1.002)

Education 0.498*** –0.635*** –0.233*** –0.497*** (1.645) (0.530) (0.792) (0.608)

Has child under age 20 –0.135 0.210** 0.134 0.023 (0.873) (1.233) (1.144) (1.024)

Regular drinker 0.144 0.029 0.298*** 0.378*** (1.155) (1.030) (1.347) (1.459)

Occasional drink-driver 0.434*** –0.531*** 0.247** 0.093 (1.543) (0.588) (1.280) (1.098)

Frequent drink-driver 0.476** –0.518** 0.278* 0.284* (1.608) (0.596) (1.320) (1.329)

Sample size 3885 3768 3751 3751 Model Chi Square 230.9 190.0 126.3 120.2 Probability 0.000 0.000 0.000 0.000 Correctly predicted 83.5% 73.0% 57.4% 68.0%

Coefficients are marked as follows: *< 0.10, **< 0.05, ***< 0.01 probability. Figures in parentheses are odds ratios.

10 Forty-seven percent of non-drink-drivers offered this response com- pared to 56 percent of occasional drink-drivers and 57 percent of fre- quent drink-drivers.

11 Thirty-one percent of the total sample believes that punishment would be very severe.

p e rceptions of punishment costs than non-drink-drivers. H o w e v e r, it cannot account for the findings related to cer- tainty of detection nor the fact that frequent- and occa- sional-sinners are equally likely to be aware of a state’s BAC l a w. The following section further examines these altern a- tive behavioral assumptions in re g a rds to perceptions of personal and societal costs among the three arc h e t y p e s .

Perceptions of Individual and Societal Costs

Beyond punishment costs, perceptions of the individual and societal costs of drink-driving also are involved in an individual’s decision to not drive after consuming alcohol. The three archetypes of sinners suggest an important dis- tinction between non-drink-drivers and others in how they weigh individual and societal costs in the decision to drink- and-drive. We expect that non-drink-drivers will have little or no utility for the activity even before policy tools are put in place, so their motivation to abstain from this behavior is not a function of policy-imposed sanctions. Instead, their action is based on their perceptions of the personal and societal costs of drink-driving. On the other hand, occa- sional or frequent drink-drivers are expected to place less emphasis on the personal and societal costs. Therefore, we expect that non-drink-drivers will perceive the threat to personal and public safety to be higher than others.

Two models examine the association of personal costs with drinking-and-driving among the three archetypes. First, in assessing the more likely impact of driving after drinking, does an individual focus more on the possibility of being stopped by the policy or the likelihood of an acci- dent? We expect that non-drink-drivers will focus more on the personal and societal costs incurred by an accident rather than the punishment costs imposed by government. Conversely, drink-drivers are more likely to recognize the threat of sanctions rather than personal costs. Therefore, when asked to judge whether being “stopped by police” or having “an accident” is more likely to occur when driving after drinking, drink-drivers are hypothesized to judge a police stop as the more likely event. As the results in the first column of Table 2 show, compared to non-drink-driv- ers both occasional and frequent drink-drivers are signifi- cantly more likely to expect the police stop as a result of drinking-and-driving than an accident. The odds ratios indicate that frequent drink-drivers are one and a half times more likely to offer this response than non-drink-drivers.

Perceptions of personal costs imposed by drinking-and- driving are also measured by how serious a threat drinking- and-driving poses to one’s personal safety and that of his/her family. The second model in Table 2 confirms the idea that occasional drink-drivers and frequent drink-drivers are less likely to see drinking-and-driving as a significant threat to personal safety when compared to non-drink-drivers. Both groups of drink-drivers are about half as likely as non- drink-drivers to see this behavior as a “major” threat to per- sonal safety. This difference in attitudes suggests that the personal costs of drinking-and-driving are less of a factor in

the decision to engage in this behavior for drink-drivers than it is for non-drink-drivers.

The third model in Table 2 addresses perceptions of soci- etal costs. Consistent with what we hypothesized, fre q u e n t drink-drivers are much less likely than others to “stro n g l y a g ree” that drinking-and-driving by people who are not p roblem drinkers is a serious highway safety problem. Mem- bers of this group are less than half as likely as non-drink- drivers to offer this opinion, while occasional drink-drivers a re a third less likely than non-drink-drivers to do so.

In sum, those who drive after drinking often do not see their actions as dangerous or wrong. This result is consistent with studies that find DWI offenders more likely to believe that some people drive better after drinking (Macdonald and Dooley 1993), that many believe they were able to drive safely after their drinking episode (Roebuck and Murty 1996; Wiliszowski et al. 1996), that the risks of drinking- driving are overrated, and that it is okay to drink-and-drive as long as one is not caught (Baum 1999). Furthermore, the above results are consistent with conclusions drawn in other research that frequent drink-drivers are largely indifferent to social norms and cannot easily be deterred through legal sanctions (Laurell 1991; Ross and Klette 1995).

Policy Preferences

Differences in the preference for sinful behavior also translate into differences in policy preferences among the three archetypes. Non-sinners are more likely to perceive drinking-and-driving as sinful behavior and to see those who engage in it through a deviant social construction. They are likely to be less tolerant of sinful behavior than others. For this reason, we expect non-drink-drivers to per- ceive the presence of any drinking-and-driving as a sign of ineffective laws and to believe that harsher coercive tools should be employed to stop the sinful behavior.

As hypothesized, the results of the first model in Table 3 indicate that drink-drivers indeed are significantly more likely than non-drink-drivers to think that current laws are “very effective.” Among the three groups frequent drink- drivers are the most likely to hold this opinion.12 In the second model a similar pattern is found when individuals are asked how important it is that something be done to reduce drinking-and-driving. Frequent drink-drivers are over two-thirds less likely than non-drink-drivers to respond that it is “very important” to do something about it, while occasional drink-drivers are a third less likely than non-drink-drivers to offer this opinion.

Consistent with attitudes about the effectiveness of curre n t laws, frequent drink-drivers are the least likely to think more stringent policy is needed while non-drink-drivers are most likely to support such policies. A frequent drink-driver is about a third less likely than a non-drink-driver to think that

60 POLITICAL RESEARCH QUARTERLY

12 Of non-drink-drivers, 11 percent believe laws to be very effective com- pared to 15 percent of occasional drink-drivers and 17 percent of fre- quent drink-drivers.

sobriety checkpoints should be used “more frequently” or that punishment for violating drinking-and-driving laws should be “much more severe” (see models 3 and 4 in Ta b l e 3 ) .1 3 This finding is consistent with previous comparisons of attitudes held by convicted DUI offenders and the entire com- munity (Baum 1999). Once again occasional drink-drivers fall between the other two groups in their stated pre f e re n c e s .

D i ff e rences between the archetypes are more than just behavioral in nature; there also are significant diff e rences in p e rceptions of problem severity and policy pre f e rences. There is greater diversity in attitudes about the nature of drinking- and-driving as a public problem and the need for more strin- gent policy than may be re p resented in policy debate.

Although we have not discussed the control variables in the above models, it is important to point out that the occa- sional drink-drivers and frequent drink-drivers are less sup- portive of more coercive policies even holding constant the

effect of drinking behavior. Throughout these models, regu- lar drinkers have very different opinions than others (infre- quent drinkers and nondrinkers) on punishment costs, individual and societal costs, and policy preferences. For instance, regular drinkers are much less likely to support the increased use of sobriety checkpoints or more severe punishment. This finding might be a representation of the silent majority on morality politics that Meier (1999) posits. Regular drinkers and not just the problem drinkers appear to see the potential risk to their own liberties from more checkpoints and more severe penalties.14 Thus, over time policy becomes more extreme in an effort to eliminate any visible sign of sinful behavior, yet policy becomes less reflec- tive of actual societal preferences and poses increasing threats to the civil liberties of citizens with even moderate levels of utility for the sinful behavior.

DRINKING-AND-DRIVING IN AMERICA 61

; TABLE 2 LOGISTIC REGRESSION MODELS OF PERCEPTIONS OF INDIVIDUAL AND SOCIETAL COSTS

Police stop more Major threat to Serious safety likely than accident personal safety problem

Constant –1.533*** 1.721*** 1.695*** — — —

Male 0.550*** –0.481*** –0.145** (1.733) (0.618) (0.865)

Single 0.278*** –0.219* –0.289*** (1.320) (0.803) (0.749)

White –0.218** 0.496*** –0.292*** (0.804) (1.641) (0.746)

Age 0.002 –0.001 –0.005 (1.002) (0.999) (0.995)

Education 0.068 –0.241*** 0.184** (1.070) (0.786) (1.202)

Has child under age 20 0.167* –0.007 –0.039 (1.181) (0.993) (0.962)

Regular drinker 0.390*** –0.534*** –0.342*** (1.477) (0.587) (0.710)

Occasional drink-driver 0.306*** –0.585*** –0.412*** (1.358) (0.557) (0.663)

Frequent drink-driver 0.403** –0.772*** –0.828*** (1.497) (0.462) (0.437)

Sample size 3590 3839 3841 Model Chi Square 150.2 244.3 117.2 Probability 0.000 0.000 0.000 Correctly predicted 72.7% 80.9% 69.3%

Coefficients are marked as follows: * < 0.10, ** < 0.05, *** < 0.01 probability.

13 Of non-drink-drivers, 53 percent think penalties should be much more severe compared to only 26 percent of occasional drink-drivers and 19 percent of frequent drink-drivers.

14 Because of their concerns over judicial equity and the constitutionality of mandatory jail terms for impaired drivers, Norway and Sweden abol- ished such laws (Ross and Klette 1995).

DISCUSSION

What do these findings mean for the utility of deterre n c e t h e o ry for combating drinking-and-driving? It appears that its assumptions do not characterize the behavior of significant p o rtions of the population. Deterrence theory is based on simplistic assumptions that do not apply to all populations nor to all behaviors (Ross 1985a). Public policy based prima- rily on the use of sanctions thus will have limited success.

For instance, those who do not drink-and-drive are not avoiding this sin because of the fear of govern m e n t - i m p o s e d sanctions. Conversely, government sanctions are unlikely to deter those who frequently drink-and-drive. These individu- als are the most likely to be aware of state law and to perc e i v e punishment costs to be high. In this way frequent drink- drivers are not the utility-maximizers that deterrence theory assumes. Additionally, diff e rences in perceptions of punish- ment costs do not account for the diff e rent levels of behavior o b s e rved between occasional and frequent drink-drivers.

The results are more consistent with an alternative set of behavioral assumptions built on Meier’s (1999) notion that

the population of sinners is characterized by three arche- types, each possessing a different elasticity of demand for sinful behavior. For those who have little or no utility for drinking-and-driving, deterrent-based policy tools are unnecessary. Education campaigns warning about the dan- gers of drinking-and-driving (hortatory and capacity policy tools) or governmental statements about desired behavior (authority tools) likely are enough to influence those in this group who might have considered driving after drinking.

C o n v e r s e l y, an increase in the certainty or severity of pun- ishment is unlikely to deter the frequent drink-driver. For instance, in a study of DWI offenders in the south, Roebuck and Murty (1996) conclude that the recidivist is impulsive and possesses low self-control. For the most extreme DWI subtype drinking-and-driving is part of a larger pattern of criminal and anti-social behavior. Yet, it is this individual who poses the greatest danger on the road by engaging in “the most dangerous and reckless type of drunk driving” (Jacobs 1989: 126). Because the habitual drunk driver is undeterre d by the risk of loss of life to themselves and others, Jacobs (1989) concludes that it is unlikely government sanctions will

62 POLITICAL RESEARCH QUARTERLY

; TABLE 3 LOGISTIC REGRESSION MODELS OF POLICY PREFERENCES

Current laws are Something should Use checkpoints Penalties much very effective be done about it more often more severe

Constant 0.115 2.548 1.071*** -0.041 — — — —

Male 0.317*** -0.800*** -0.654*** -0.373*** (1.373) (0.449) (0.520) (0.689)

Single 0.168* -0.213 -0.384*** -0.286*** (1.183) (0.809) (0.681) (0.751)

White 0.073 0.447*** 0.238** 0.256*** (1.076) (1.564) (1.268) (1.292)

Age 0.001 -0.001 0.002 0.006** (1.001) (0.999) (1.002) (1.006)

Education -0.281*** -0.214** 0.190** -0.039 (0.755) (0.807) (1.209) (0.962)

Has child under age 20 0.011 0.160 0.104 0.059 (1.011) (1.174) (1.110) (1.060)

Regular drinker 0.401*** -0.849*** -0.783*** -0.680*** (1.494) (0.428) (0.457) (0.506)

Occasional drink-driver 0.247** -0.585*** -0.285*** -0.750*** (1.280) (0.557) (0.752) (0.473)

Frequent drink-driver 0.367** -1.114*** -0.572*** -0.901*** (1.443) (0.328) (0.564) (0.406)

Sample size 3821 3865 3740 3771 Model Chi Square 104.1 362.7 342.1 352.1 Probability 0.000 0.000 0.000 0.000 Correctly predicted 62.6% 86.7% 70.7% 62.1%

Coefficients are marked as follows: * < 0.10, **< 0.05, ***< 0.01 probability. Figures in parentheses are odds ratios.

be effective deterrents. To the extent that this behavior is symptomatic of other individual or social problems, deterre n t tools are likely to be even more ineffective. In these cases sub- stance abuse treatment, counseling or ignition interlock devices (capacity tools) may be more appropriate. Meier’s (1994, 1999) theory of morality politics, however, suggests that these capacity tools are not likely to be adopted or used regularly because they appear to confer benefits on sinners.

The implication of individual-level re s e a rch is that deter- rent policies may influence the occasional drink-driver but not the habitual one. As Levitt and Porter (2001: 1231, 1234) argue: “These estimates [of the danger drinking driv- ers pose] are consistent with a scenario in which such poli- cies have the greatest impact on the least dangero u s / m o s t m a rginal drinking drivers . . . [and provide] some circ u m- stantial evidence in favor of the hypothesis that such laws do not keep hard - c o re drunk drivers off the roads.” Jacobs (1989) surmises that the most that can be hoped for is a mar- ginal reduction in drunk driving due to deterrent policies.

Because we are not able to map out the demand curves of the respondents in the survey used here, nor are we able to d e t e rmine the amount of benefit an individual gains fro m drinking-and-driving, an alternative interpretation may merit consideration. Finding that frequent drink-drivers are the most likely to perceive punishment costs to be very high could mean that policies have not pushed sanctions high enough to discourage this group. Thus, these individuals may be acting in a manner consistent with deterrence theory but re q u i re significantly higher costs to be deterred from drink- driving. This interpretation assumes that perceptions of pun- ishment costs are unrelated to respondents’ perceptions of the utility derived from drink-driving. Another key for deterre n t anti-sin policies would be to increase the certainty of detec- tion, the one assessment of punishment costs that drink-driv- ers perceive to be lower than non-drink-drivers.

Three caveats should be mentioned regarding the latter interpretation of the findings. First, the severity of punish- ment often imposed by state law may already be at levels comparable to the utility frequent drink-drivers derive from this behavior. Based on an assessment of the risk and exter- nalities of drinking-and-driving, Levitt and Porter (2001: 1229) suggest that “existing punishments are not radically out of line with what might be optimal.” If Levitt and Porter are correct, clearly frequent drink-drivers are not behaving in accordance with deterrence theory.

Second, the importance of increasing the certainty of detection for deterring drink-driving has been questioned. For instance, Williams and Lund (1984) found the likelihood of arrest to be unrelated to drinking-and-driving episodes. Also, Wiliszowski et al. (1996) state that prior to their first incident drunk driving offenders often do not think about the possibility of detection and arrest by the police.

T h i rd, attempting to deter drinking-and-driving by incre a s- ing the certainty of detection is an expensive pro p o s i t i o n . Levitt and Porter (2001: 1229) estimate there to be “ro u g h l y one arrest for every 27,000 miles driven by drunk drivers.” Evans (1991) places the probability of arrest on an impaire d

driving trip at one in one thousand. A doubling of govern m e n t re s o u rces devoted to increasing this probability would re d u c e detection to only about one in five hundred impaired driving trips. In the face of such small probabilities severe sanctions a re still likely to be ineffective. As Ross (1985b: 207) con- cludes: “What has been achieved through the deterrence pro- grams is bluff and the bluff has been discovered by the public.”

In sum, public policies primarily based on the assumptions of deterrence theory are likely to have limited success re d u c- ing the frequency drinking-and-driving or sinful behavior in general. This is especially the case for the hard c o re sinner who often is the explicit target of more severe sanctions. Instead, a strategy to combat undesired behavior must recognize the diversity of the target populations at risk for sinful behavior and utilize a wider variety of policy tools. Additional re s e a rc h is needed to test the applicability of this thesis to other issues characterized by morality politics and other contexts where d e t e rrence is used to modify behavior. Doing so will con- tribute to the development of a theory of policy design that will assist in crafting more effective public policy.

REFERENCES

Baum, Scott. 1999. “Drink Driving as a Social Problem: Compar- ing the Attitudes and Knowledge of Drink Driving Offenders and the General Community.” Accident Analysis and Prevention 32: 689-94.

Berk, Richard A., Alec Campbell, Ruth Klap, and Bruce Western. 1992. “The Deterrent Effect of Arrest in Incidents of Domestic Violence: A Bayesian Analysis of Four Field Experiments.” American Sociological Review 57: 698-708.

Chaloupka, Frank J., Henry Saffer, and Michael Grossman. 1993. “Alcohol-Control Policies and Motor-Vehicle Fatalities.” Journal of Legal Studies 22: 161-86.

Delli Carpini, Michael X., and Scott Keeter. 1996. What Americans Know About Politics and Why It Matters. New Haven, CT: Yale University Press.

Evans, Leonard. 1991. Traffic Safety and the Driver. New York: Van Nostrand Reinhold.

Evans, William N., Doreen Neville, and John D. Graham. 1991. “General Deterrence of Drunk Driving: Evaluation of Recent American Policies.” Risk Analysis 11: 279-89.

Gibbs, Jack P. 1975. Crime, Punishment and Deterrence. New York: Elsevier.

Gordis, Enoch. 1991. “From Science to Social Policy: An Uncertain Road.” Journal of Studies on Alcohol 52: 101-09.

Gusfield, Joseph R. 1981. The Culture of Public Problems: Drinking-Dri- ving and the Symbolic Ord e r. Chicago: University of Chicago Pre s s.

Haight, Frank. 1985. “Current Problems in Drinking-Driving: Research and Intervention.” Journal of Studies on Alcohol, Sup- plement No. 10: 13-18.

Homel, Ross. 1988. Policing and Punishing the Drinking Driver: A Study of General and Specific Deterre n c e. New York: Springer- Ve r l a g .

Huckfeldt, Robert, Paul Allen Beck, Russell J. Dalton, and Jeffrey Levine. 1995. “Political Environments, Cohesive Social Groups, and the Communication of Public Opinion.” American Journal of Political Science 39:1025-1094.

Huckfeldt, Robert, and John Sprague. 1987. “Networks in Con- text: The Social Flow of Political Information.” American Politi- cal Science Review 81: 1197-1216.

DRINKING-AND-DRIVING IN AMERICA 63

Jacobs, James B. 1989. Drunk Driving: An American Dilemma. Chicago: University of Chicago Press.

Kenkel, Donald S. 1993. “Drinking, Driving, and Deterrence: The Effectiveness and Social Costs of Alternative Policies.” Journal of Law and Economics 36: 887-911.

Laurell, H. 1991. “The Swedish Experience: Changes in BAC Leg- islation.” Alcohol, Drugs and Driving 7: 261-65.

Levitt, Steven D., and Jack Porter. 2001. “How Dangerous Are Drinking Drivers?” Journal of Political Economy 109: 1198-1237.

Lukes, Steven, and Andrew Scull, eds. 1983. Durkheim and the Law. New York: St. Martin’s.

Lupia, Arthur, and Mathew McCubbins. 1998. The Democratic Dilemma: Can Citizens Learn What They Need to Know? New York: Cambridge University Press.

Macdonald, S., and S. Dooley. 1993. “A Case-Control Study of Dri- ving-While-Impaired Offenders.” Drug and Alcohol Dependency 33: 61-71.

Meier, Kenneth J. 1994. The Politics of Sin: Drugs, Alcohol, and Public Policy. Armonk: Sharpe.

_____. 1999. “Drugs, Sex, Rock, and Roll: A Theory of Morality Politics.” Policy Studies Journal 27: 681-95.

Meier, Kenneth J., and Michael J. Licari. 1997a. “Cigarette Taxes, Smoking, and Public Health.” American Journal of Public Health 87: 1126-30.

_____. 1997b. “Public Policy Design: Combining Policy Instru- ments.” Paper presented at the APSA annual meeting, Wash- ington, DC. As cited in Meier (1999).

Mooney, Christopher Z., and Mei-Hsien Lee. 1999. “The Temporal Diffusion of Morality Policy.” Policy Studies Journal 27: 766-80.

Moskowitz, Joel M. 1989. “The Primary Prevention of Alcohol Problems: A Critical Review of the Research Literature.” Journal of Studies on Alcohol 50: 54-88.

National Highway Tr a ffic Safety Administration. 1996. N a t i o n a l S u rvey of Drinking and Driving Attitudes and Behavior: 1995. Wa s h- ington, DC: Department of Tr a n s p o rtation, DOT HS 808 438.

Neustrom, Michael W., and William M. Norton. 1993. “The Impact of Drunk Driving Legislation in Louisiana.” Journal of Safety Research 24: 107-21.

Pate, Anthony M., and Edwin E. Hamilton. 1992. Formal and Infor- mal Deterrents to Domestic Violence: The Dade County Spouse Assault Experience. American Sociological Review 57: 691-97.

Roch, Christine H., John T. Scholz, and Kathleen M. McGraw. 2000.” Social Networks and Citizen Response to Legal Change”. American Journal of Political Science 44: 777-91.

Roebuck, Julian B., and Komanduri S. Murty. 1996. The Southern Subculture of Drinking and Driving: A Generalized Model for the Southern White Male. New York: Garland.

Ross, H. Laurence. 1985a. “Deterring Drunken Drivers: An Analy- sis of Current Efforts.” Journal of Studies on Alcohol Supplement No. 10:122-128.

_____. 1985b. Summary of Topic C—Countermeasures. Journal of Studies on Alcohol Supplement No. 10: 207-09.

_____. 1992. Confronting the Drunk Driver: Social Policy for Saving Lives. New Haven, CT: Yale University Press.

Ross, H. Laurence, and Hans Klette. 1995. “Abandonment of Mandatory Jail for Impaired Drivers in Norway and Sweden.” Accident Analysis and Prevention 27: 151-57.

Royal, Dawn. 2000. “Volume I: Findings: Racial and Ethnic Group Comparisons: National Surveys of Drinking and Driving Atti- tudes and Behavior.” August 2000. Washington, DC: U.S. Department of Transportation, NHTSA, DOT HS 809 071.

Ruhm, Christopher J. 1996. “Alcohol Policies and Highway Safety

Fatalities.” Journal of Health Economics 15: 435-54. Schneider, Anne, and Helen Ingram. 1990. “Behavioral Assump-

tions of Policy Tools.” Journal of Politics 52: 510-29. _____. 1993. “Social Construction of Target Populations: Implica-

tions for Policy and Theory.” American Political Science Review 87: 334-47.

Schneider, Mark, Paul Teske, and Melissa Marschall. 2000. Choos- ing Schools: Consumer Choice and the Quality of American Schools. Princeton, NJ: Princeton University Press.

Sherman, Lawrence W. 1984. “Experiments in Police Discretion: Scientific Boon or Dangerous Knowledge?” Law and Contempo- rary Problems 47: 61-81.

Sherman, Lawrence W., Douglas A. Smith, Janell D. Schmidt, and Dennis P. Rogan. 1992. “Crime, Punishment, and Stake in Con- formity: Legal and Informal Control of Domestic Violence.” American Sociological Review 57: 680-90.

Sloan, Frank A., and Penny M. Githens. 1994. “Drinking, Driving and the Price of Automobile Insurance.” Journal of Risk Insur- ance 61: 33-58.

Sniderman, Paul M. 1993. “The New Look in Public Opinion Research.” In Ada W. Finifter, ed., Political Science: The State of the Discipline II. Washington, DC: APSA.

Sniderman, Paul M., Richard A. Brody, and Phillip E. Tetlock. 1991. Reasoning and Choice: Explorations in Political Psychology. New York: Cambridge University Press.

Ungerleider, Steven, and Steven A. Bloch. 1988. “Perceived Effec- tiveness of Drinking-Driving Countermeasures: An Evaluation of MADD.” Journal of Studies on Alcohol 49: 191-95.

Vingilis, Evelyn R. 1990. “A New Look at Deterrence.” In R. Jean Wilson and Robert E. Mann, eds., Drinking and Driving: Advances in Research and Prevention. New York: Guilford Press.

Voas, Robert B., and J. Hause. 1987. “Deterring the Drinking Driver: The Stockton Experience.” Accident Analysis and Preven- tion 19: 81-90.

Voas, Robert B., A. Scott Tippetts, and James Fell. 2000. “The Rela- tionship of Alcohol Safety Laws to Drinking Drivers in Fatal Crashes.” Accident Analysis and Prevention 32: 483-92.

Wagenaar, Alexander C., Terry S. Zobeck, Gerald D. Williams, and Ralph Hingson. 1995. “Methods Used in Studies of Drink- Drive Control Efforts: A Meta-Analysis of the Literature From 1960 to 1991.” Accident Analysis and Prevention 27: 307-16.

Wasserman, Stanley, and Joseph Galaskiewicz, eds. 1994. Advances in Social Network Analysis: Research in the Social and Behavioral Sciences. Thousand Oaks, CA: Sage.

Whetten-Goldstein, Kathryn, Frank A. Sloan, Emily Stout, and Lan Liang. 2000. “Civil Liability, Criminal Law, and Other Policies and Alcohol-Related Motor Vehicle Fatalities in the United States: 1984-1995.” Accident Analysis and Pre v e n t i o n 32: 723-33.

Wiliszowski, C., P. Murphy, R. Jones, and J. Lacey. 1996. “Deter- mine Reasons for Repeat Drinking and Driving.” Washington, DC: U.S. Department of Transportation, National Highway Traffic Safety Administration, DOT HS 808 401.

Zador, Paul L. 1991. “Alcohol-Related Relative Risk of Fatal Driver Injuries in Relation to Driver Age and Sex.” Journal of Studies on Alcohol 52: 302-10.

Received January 13, 2003. Accepted for Publication April 28, 2003. [email protected] [email protected]

64 POLITICAL RESEARCH QUARTERLY