Study Aim

The aim of this study is to check whether certain personality characteristics and patterns of alcohol consumption may enhance the propensity for law-breaking and/or relapsing, leading to disqualification from driving. Three main objectives are set:

Table 1 Demographic Factors and Accident Involvement 

Note. Group A = non-offenders; group B = offenders banned from driving by court order; group C = offenders banned from driving by accumulation of penalty points.

Participants

A sample of 358 drivers participating in the study was divided into two groups: a target group of 232 traffic offenders taking a re-education driving course after revocation of their driver’s license, and a reference group of 126 habitual drivers with a valid driver’s license, randomly selected from the normal population. Participants were holding a driver’s license for an average of 15 years and driving regularly or on a daily basis. Table 1 displays the distribution of men and women, demographic characteristics, and accident involvement in the driver sample.

Concerning gender, 88% of them were male (n = 316), and only 12% were women (n = 42): there were 109 men and 17 women in the group of non-offenders, 107 men and 20 women in the group of offenders banned from driving by court order, and 100 men and 5 women in the group of offenders banned from driving by accumulation of penalty points.

Concerning age, participants were 34 years old on average, with values of mean age and standard deviation of 34.7 (10.5) in the group of non-offenders, 34.6 (10) in the group of offenders banned from driving by court order, and 33.7 (11.1) in the group of offenders banned from driving by accumulation of penalty points.

The statistical analysis did not reveal significant differences in the gender variable, comparing the groups of offenders and non-offenders, χ²(1, N = 358) = 0.58, p = .44. Regarding age, neither age, t(355) = 0.20, p = .84, nor age ranges, χ²(4, N = 358) = 0.21, p = .995, revealed differences between the groups of offenders (M = 34.23, SD = 10.50), and non-offenders (M = 34.47, SD = 10.51).

There were no significant differences in driving experience (namely, how long they had had their driver’s license) between offenders (M = 13.98, SD = 10.23), and non-offenders (M = 15.51, SD = 10.66). There were no differences in driving frequency either, χ²(3, N = 358) =17.733, p < .001, except a percentage of daily drivers slightly higher than expected in the group of offenders (89.5%) compared to the group of non-offenders (73.8%).

The group of offenders was divided into two categories or subgroups, according to the reason their driving privileges had been suspended or revoked: a group of 127 traffic offenders disqualified by a court conviction (54.74%) and a second group of 105 traffic offenders disqualified without court conviction (45.26%).

It should be noted that in the Spanish legal system traffic offenses can be criminal or civil offenses. Civil offenses or infractions are generally punishable only by fines or administrative action such as demerits (penalty points), whereas a criminal offense is usually punishable with jail time, a fine, or both. For instance, drinking and driving can be either a misdemeanor or a crime, sanctioned or punished depending on the blood alcohol content (BAC) level when operating a vehicle (over 0.5 g/l means 4 penalty points, over 1.0 g/l means 6 penalty points, over 1.2 g/l means the driver is legally impaired to drive). Since revocation of a driver’s license may be the result of any of these two causes, administrative or judicial cause, both subgroups of traffic offenders were differentiated and compared. Habitual traffic offenders (HTO), i.e., drivers whose driving privileges have been suspended or revoked and get caught driving while disqualified (during the period of suspension or revocation), were excluded. Professional drivers were excluded, too.

Procedure and Research Design

We used a cross-sectional study design. The study was conducted in six qualified driving schools where driver re-education and awareness-raising courses usually take place. As explained above, drivers are disqualified when they get twelve or more demerit points within a three-year period, or when they are convicted of a traffic offense and lose their license. In both cases, disqualified drivers take these intensive, mandatory courses and have to pass a new driver’s test.

Traffic offenders were asked to participate in the study on a voluntary basis before beginning the course. Anonymity and confidentiality were ensured. Those who agreed were informed that the research was a study designed at the university and independent from the re-education course, with no influence on their driving license suspension or test to avoid any tendency towards social approval.

Once they knew the aim of the study and after signing their informed consent, they were given the battery of tests by a team of qualified surveyors, and all the research was supervised by university professors.

Non-offenders used as the reference group were common drivers, randomly selected from the general population, using the latest data provided by the General Directorate of Traffic (DGT), the Spanish agency responsible for driver’s licenses and mobility and transport policies. Thus, the same study was conducted on a random sample of 126 Spanish habitual drivers with a valid driving license, as similar as possible in age, gender, driving frequency, and driving experience. Like traffic offenders, the group of non-offenders participated on a voluntary basis and gave their informed consent, too. Ethics and procedure were similar, except they did not have to attend classes, and an appointment was scheduled to interview them for the survey.

Ethics statement: The research was conducted according to the ethical principles for medical research involving human subjects, including research on identifiable material and data of the Declaration of Helsinki adopted by the World Medical Association (WMA). All participants were invited to take part in the study on a voluntary basis and gave their informed consent. The methods proposed for research were reviewed by an institutional review board (an independent ethics committee at the University Research Institute on Traffic and Road Safety (INTRAS) to ensure that they were ethical.

Measurement Instruments

Social desirability of respondents was assessed with the Marlowe-Crowne Social Desirability Scale (MC-SDS), translated into Spanish (^{Ferrando & Chico, 2000}). The scale consists of 33 items that assess whether respondents are concerned with social approval. A high number of socially desirable responses indicates that the respondent is generally concerned with social approval and conforming to societal conventions, while a low score indicates that the respondent is less concerned with such things and is more willing to answer a survey truthfully and represent himself or herself accurately (^{Crowne & Marlowe, 1964}). To prevent any possible bias, we performed statistical analysis comparing responses of each group of drivers (non-offenders, convicted, and penalized offenders) in negation (15 items) and attribution (18 items). There were no significant differences in attribution F(2.35) = 0.96, p = .38, ƞ² = .00, or negation F(2.35) = 0.00, p = .99, ƞ² = .00, meaning that our results would not be better explained by a possible bias of social desirability.

Personality and alcohol consumption were assessed using three different tests: the Driving Anger Scale (DAS), the Sensitivity to Punishment and the Sensitivity to Reward Questionnaire (SPSRQ-20) and the Zuckerman-Kuhlman-Aluja Personality Questionnaire (ZKA-PQ). As the scales were already established, we performed confirmatory factorial analysis for validating our constructs. With DAS and SPSRQ-20, we used the direct oblimin rotation method to test whether our data fit the hypothesized measurement model. With ZKA-PQ, we selected the method of extraction using orthogonal rotation varimax as the authors did in their original research. The results were similar to those obtained in original and previous studies. Table 3 displays the means and standard deviations in measured variables of personality and alcohol consumption for each group, and Table 4 shows the internal validity of measures.

Table 3 Variables of Personality and Alcohol Consumption. Means and Standard 

Deviations*Significant differences.

Table 4 Internal Validity of Measures of Personality and Alcohol Consumption  

Although driving anger cannot be considered as a personality trait, it was included because of its potential value for research on accident prevention and health psychology. Responses to different potentially angering driving-related situations involving hostile gestures, illegal driving, police presence, slow driving, discourtesy, and traffic obstructions correlate positively, suggesting a general dimension of driving anger as well as anger linked to specific driving-related situations.

Driving anger was assessed with the 14 items short form of the Deffenbacher Driving Anger Scale (DAS) (^{Deffenbacher, Oetting, & Lynch, 1994}), translated into Spanish and validated for the Spanish driver population in two previous studies (Egea-Caparrós, Velandrino-Nicolás, Fernández-Ros, & Prieto-Martínez, 2012; ^{Herrero-Fernández, 2011}). A first principal component factorial analysis confirmed the good fit in three factors suggested by ^{Herrero-Fernández (2011)}, explaining 53.62% of variance. A second factorial analysis confirmed a four-factor model suggested by ^{Egea-Caparrós et al. (2012)}, explaining a 60.01% variance. The scale demonstrated good internal reliability using the four-factor model, with Cronbach α for each subscale ranging from .50 to .82, and total score of .85. The Kaiser-Meyer-Olkin measure of sample adequacy was meritorious (KMO = .86), as well as Barlett’s test of sphericity, χ²(91) = 1380.19, p < .001. We considered, therefore, using four subscales for the purposes of our study, concerning anger over traffic obstructions, illegal driving, hostile gestures, and the possibility of being fined: DAS-FI- Impeded Progress, DAS-FII- Reckless Driving, DAS-FIII- Direct Hostility, DAS-FIV- Sanction Hostility.

Behavioral inhibition was assessed with the 20-item short form of the Sensitivity to Punishment and Sensitivity to Reward Questionnaire (SPSRQ-20) using a Likert-type rating scale for responses (^{Torrubia, Ávila, Moltó, & Caseras, 2001}). The results of our statistical analysis were similar to those obtained in the original studies (^{Aluja & Blanch, 2011}), with two factors explaining 36.09% of variance: sensitivity to punishment and sensitivity to reward. In our analysis, Cronbach α was: .80 in the total score, M = 43.33, SD = 8.65; .79 in the sensitivity to punishment factor, M = 21.31, SD = 5.58; and .78 in the sensitivity to reward factor, M = 22.03, SD = 5.59. Both were included.

Five other personality traits were measured using the 200 items of the Zuckerman-Kuhlman-Aluja Personality Questionnaire (ZKA-PQ) (^{Aluja et al., 2013}), which includes five factors with four facets per factor and ten items per facet. Confirmatory factor analysis showed satisfactory goodness of fit indexes: the Kaiser-Meyer-Olkin measure of sample adequacy was meritorious (KMO = .80), as well as Barlett’s test of sphericity, χ²(190) = 3229.17, p < .001. As in an earlier study by ^{Aluja, Kuhlman, and Zuckerman (2010)}, our factorial analysis showed a five-factor structure agglutinating 20 facets and explaining 56.08% of variance, with Cronbach α of .80 in total score and ranging from .86 to .91 for aggression (α = .91), activity (α = .88), extraversion (α = .87), neuroticism (α = .89), and sensation seeking subscales (α = .86).

Problematic drinking was assessed using the ten item Alcohol Use Disorders Identification Test (AUDIT), which measures four levels of alcohol consumption: non-problematic, hazardous, harmful, and dependence (^{Saunders, Aasland, Babor, de la Fuente, & Grant, 1993}). The AUDIT was developed as a simple screening method for excessive drinking, and it is useful to identify persons with hazardous and harmful patterns of alcohol consumption. It also helps identify alcohol dependence and some specific consequences of harmful drinking. In our study, and based on recommendations provided in the manual (^{Babor, Higgins-Biddle, Saunders, & Monteiro, 2001}), we used the total score, taking scores over eight as indicators of problematic alcohol consumption.

Table 2 Types of Traffic Offenses 

Note. Group A = non-offenders; group B = offenders banned from driving by court order; group C = offenders banned from driving by accumulation of penalty points.

Data Analysis

Statistical analysis was performed using the SPSS Statistics 19. A database was prepared with respondent information including demographics, data about driving experience, traffic offenses, and crash involvement, and responses to selected subscales as part of the battery of tests. Data were cleaned, checking for mistakes or inconsistencies, so that some subjects were removed, decreasing the sample size. To check the distribution of the statistical data set and the assumption of normality before running the statistical test of regression, a descriptive analysis was done, with measures of central tendency, scattering, skewness, and kurtosis, for quantitative variables. Statistical analysis was performed according to the aims.

First, we described the characteristics of the target group of traffic offenders and compared them with the reference group of common drivers, including age, gender, driving frequency, and crash involvement. We considered the reason for disqualification and the offenses committed before being banned from driving and described common traffic violations that cost them demerit points and/or license suspension.

Second, to validate the measurement scales used in our study, we performed psychometric analysis. We did descriptive analysis of items, factors, and total scores of questionnaires selected for the study, with means and standard deviations for our sample. To check how suited our data were for factor analysis, we performed the KMO test for sampling adequacy for each variable of the model and the complete model, and Bartlett test of sphericity. Then, we proceed to confirmatory factorial analysis of each scale using methods of principal components and principal axis factoring, and methods of factor rotation oblimin direct or varimax, depending of the scale.

Third, to test whether or not there were significant differences between our groups in the selected set of individual characteristics (personality and alcohol consumption), we performed an analysis of variance of one factor (traffic offenses) with three levels (non-offenders, convicted offenders, penalized offenders). With this design, we obtained multiple observations in the form of scores on scales of personality and alcohol consumption, from a number of drivers belonging to the three levels of the traffic offences factor. Non-significance of the test statistic associated with this technique would imply that offending driving is not reflected in individual characteristics as measured by the scales. On the other hand, significance would imply that drinking and other individual tendencies (e.g., anger, behavioral inhibition, or aggression) afflict different groups differently.

Fourth, to determine whether there was a causal relationship and to identify predictors of an offending driving style conducive to disqualification, we performed a multinomial logistic regression, the linear regression analysis to conduct when the dependent variable is nominal with more than two levels. In this analysis, we used traffic offenses as dependent variable and those factors of personality and alcohol consumption that differed significantly among the groups of drivers in our previous analysis as independent variables. The model can predict whether changes in these explanatory variables (related to personality and alcohol consumption) can subsequently affect the driving style and the probability of being banned from driving by serious or repeat traffic offenses, either penalized or convicted. We used the method of maximum likelihood to estimate the coefficients and calculate the odds ratio with a confidence interval of 95%, and the Nagelkerke R² test of goodness of fit.

Finally, we performed two additional binary logistic regression analyses: one addressing both groups of traffic offenders jointly (to contrast offenders vs. non-offenders) and another considering them individually (to contrast convicted vs. penalized offenders). These analyses were performed using a stepwise regression and a method of maximum likelihood, with Hosmer-Lebeshow and Nagelkerke R² test of goodness of fit.

Differences in Traffic Violations and Crash Involvement

As mentioned above, all traffic offenders had been disqualified from driving, but the reasons for their disqualification differed, so that some were convicted and others were not. It was the first time for nearly all of them: only 3% of traffic offenders had already been disqualified from driving on account of offenses committed when driving. The offenses they had committed before being banned from driving were different, too.

Most common traffic violations that cost them demerit points and/or license suspension were: speeding (exceeding speed limits), driving under the influence of alcohol (blood alcohol level over 0.5 g/l while driving), cell phone ban (use of cell phone while operating a vehicle), seatbelt violations (failure to fasten seatbelt), violation of road signs (running a red light or illegal passing and other), reckless driving (turning or changing lanes without proper signaling, driving in wrong direction on highway, tailgating or inadequate distance from vehicle in front with endangerment or damage, and other), driving while disqualified (driving without a valid license or during the period of suspension or revocation), and other miscellaneous violations. Frequencies of these traffic offenses in the sample of our study are described in Table 2.

As many as 163 drivers of the 232 traffic offenders (70%) drink and drive. A comparison of the two groups of traffic offenders shows a higher incidence of this offense on those who had been disqualified by court conviction (94%) compared to those who had been disqualified by accumulating penalty points (41%). Membership of this subgroup should be drawn primarily from drunk drivers, given that they have been arrested for driving under the influence with a high blood alcohol level over 1.2 g/l, and their driving privileges are revoked immediately.

In case of lower blood alcohol levels, driving under the influence can result in four or six penalty points for licensed drivers, so this type of offense is also found in the group of traffic drivers disqualified without a court conviction, i.e., by penalty points. However, the lower incidence of drinking and driving suggest a different pattern of alcohol consumption. We proceeded from that premise, searching for differences in alcohol consumption between these two subgroups in a further analysis.

Moreover, although the myriad traffic offenses among drivers disqualified without a court conviction (such as speeding, phone use ban, and seatbelt and road signs violations) show a recidivist behavior accumulating penalty points, all this suggests that membership of this subgroup may well be drawn primarily from repeat traffic offenders with a persistent risky driving behavior who probably are not aware of that fact, have little awareness of it, or are little concerned about it. We proceeded from that second premise, and we investigated additional differences between these two groups.

A first finding was that around 17% of them had previously and unsuccessfully completed re-education and awareness-raising courses. As explained above, any driver who commits a traffic offense can attend these classes and get privileges removing penalties (up to six points every two years). Some individuals may intentionally complete these courses to prevent being banned from driving when they are at risk of disqualification. Only ten people from the other groups (nine convicted and one non-offender) had taken courses before. Statistical differences between groups were significant, χ²(2, N = 358) = 20.15, p < .001.

Additionally, it should be taken into account that many traffic offenders had already been involved in crashes. When comparing crash involvement in both subgroups of traffic offenders, those disqualified without a court conviction are found to be responsible for a greater number of accidents involving persons and resulting in casualties. A first statistical analysis does not reveal significant differences in non-injury accident involvement among the three groups of common drivers, offenders disqualified by a court conviction and offenders disqualified by penalty points, χ²(2, N = 358) = 1.5, p = .47, but a second analysis reveals that there are significant differences in accidents causing injuries to people and fatal accidents. In other words, they are not more frequently involved in crashes, but the accidents they cause are more serious and more likely to result in death and injuries. Mann-Whitney’s U test using Bonferroni adjustments in three possible combinations reveals significant differences between traffic offenders disqualified by penalty points and traffic offenders disqualified by a court conviction, z = 2.79, p = .005 and, of course, between the former and common drivers, z = 3.45, p = .001. In short, the convicted are mostly drunk drivers, responsible for more serious offenses, and the sanctioned are mostly recidivists, responsible for less serious but repeat offenses causing, despite this, more serious accidents.

Differences in Personality and Alcohol Consumption

To know whether performance on each personality and alcohol consumption test differs significantly among three groups of drivers, i.e., whether all three levels (groups of offenders, convicted offenders, and penalized offenders) have similar average scores, scores on different scales were registered and analyzed. Means and standard deviations in each measured variable of personality and alcohol consumption for each groups are shown in Table 3.

Table 4 shows the internal validity of personality and alcohol consumption.

Table 5 shows data of group effect and statistical differences among the groups. Results reveal significant differences in eight of the variables analyzed, the one related to alcohol consumption and seven variables of personality: anger over impeded progress (DAS-FI), sanction hostility (DAS-FIV), sensitivity to reward, aggression, activity, neuroticism, and sensation seeking.

Table 5 Differences in Personality and Alcohol Consumption. Group Effect and Pairwise Comparison 

Note. Group A = non-offenders; group B = offenders banned from driving by court order; group C = offenders banned from driving by accumulation of penalty points. ns = p > .10, ns1: p = .10, *p = .05, **p = .01, ***p = .001.

Pairwise comparison of groups reveals some significant differences between the group of non-offenders and both groups of traffic offenders (A-B and A-C). Basically, non-offenders are less inclined than offenders to lose their temper in certain situations that easily cause frustration and anger (such as being stuck in a traffic jam, tailgating a cyclist, or waiting while a vehicle is being parked). Compared with offenders, non-offenders are also less likely to be hostile, aggressive drivers, or have a penchant for thrill and excitement while driving, and less likely to have alcohol issues.

Other differences can be found between the group of non-offenders and only one of the two groups of offenders. Non-offenders appear to be, in this sense, less inclined to get annoyed in situations conducive to penalties (such us being caught or detected by radar while driving too fast, or regular police traffic controls on the road) compared with the offenders disqualified by penalties, and less inclined to activity and anxiety compared with those disqualified by conviction.

Pairwise comparisons of groups also reveal some subtle yet significant differences between the group of penalized offenders and the group of convicted offenders (B-C). Basically, convicted offenders are less inclined than penalized offenders to get annoyed at the possibility of being fined and less inclined to be active and aggressive while driving. A more obvious difference between these two groups is that convicted offenders are more likely to have a problem with alcohol.

Predictors of Disqualification from Driving

A regression model is used to determine whether there is a causal relationship. In our analysis, we selected disqualification from driving (Y) as dependent variable and used explanatory factors that highlighted significant differences in our previous analysis as independent variables (X): anger over impeded progress, sanction hostility, sensitivity to reward, aggression, activity, neuroticism, sensation seeking, and problematic alcohol consumption.

We found a moderate coefficient of determination, Nagelkerke R² (.57). The highest coefficient value, and consequently the highest predictive power, was found, as expected, in alcohol consumption, with lower values in sanction hostility and activity. Table 6 displays the coefficients resulting from this regression analysis.

Table 6 Multinomial Regression Analysis. Coefficients 

Note. Group A = non-offenders; group B = offenders banned from driving by court order; group C = offenders banned from driving by accumulation of penalty points.

On the basis of a cut-off estimated probability of p (Y = 1), the model enables adequate estimates in as much as 67.1% of cases. The model, nevertheless, fits well with non-offenders (classifying 87.9% of subjects in this group), but it does not fit as neatly with offenders (classifying 60.8% of convicted and 45.8% of penalized). Such low percentages suggest that both groups of traffic offenders should be addressed as a single group.

Table 7 Binary Regression Analysis-1. Omnibus and Goodness of Fit Tests 

Note. Dichotomous dependent variable = disqualified (1), non-disqualified (0).

Table 8 Binary Regression Analysis-1. Coefficients 

Note. Dichotomous dependent variable = disqualified (1), non-disqualified (0).

To test it, we performed two additional binary regression analysis with a dichotomous dependent variable (Y): disqualified (1), non-disqualified (0), and disqualified with conviction (1), disqualified without conviction (0). In both cases, the explanatory factors (X) were the same as before.

In the first binary regression analysis (disqualified-non-disqualified), we found a better coefficient of determination, Nagelkerke R² (.62). On the basis of a cut-off estimated probability of p (Y = 1), the model enables adequate estimates of as much as 84.7% of cases. It has high specificity and sensitivity, classifying correctly 82.8% of non-offenders and 85.9% of offenders. Tables 7 and 8 display the results of this analysis. The highest coefficient value, and consequently the highest predictive power, was found in alcohol consumption, with lower values in activity.

In the second binary regression analysis (disqualified with court conviction-disqualified without court conviction), we found moderate goodness of fit coefficients, indicating that 20% of variance in our dependent variable can be explained. On the basis of a cut-off estimated probability of p (Y = 1), the model enables adequate estimates of 68.6% of cases. It has medium specificity and high sensitivity, classifying correctly 56.6% of non-convicted and 78.4% of convicted. Tables 9 and 10 display the results of this second binary analysis. The highest coefficient value, and consequently the highest predictive power, was found, again, in alcohol consumption, with lower values in sanction hostility and sensation seeking.

Table 9 Binary Regression Analysis-2. Omnibus and Goodness of Fit Tests 

Note. Dichotomous dependent variable = disqualified with court mandate (1), disqualified without court mandate (0).

Table 10 Binary Regression Analysis-2. Coefficients 

Note. Dichotomous dependent variable = disqualified with court mandate (1), disqualified without court mandate (0).

Results reveal that personality characteristics are not as good predictors of driving offenses as drinking is. Alcohol disorders are depicted in a showy way as the best predictor of disqualification from driving by serious or repeat traffic offenses, either penalized or convicted.

According to the first binary regression analysis, this probability is 2.02 times greater for drivers with high levels of alcohol consumption as measured by the Alcohol Use Disorders Identification Test (AUDIT), and 1.04 times greater for drivers with high scores in “activity” as measured by the ZKA-PQ.

According to second binary regression analysis, the probability of disqualification from driving with a court conviction would be 1.11 higher for drivers with high levels of alcohol consumption as measured by the Alcohol Use Disorders Identification test (AUDIT). The probability of disqualification from driving without a court conviction is 0.97 greater for drivers with high scores in sanction hostility as measured by the DAS and 0.76 times greater for drivers with high scores in sensation seeking as measured by the ZKA-PQ.

Differences in Traffic Violations and Crash Involvement

Our first testable hypothesis was that there were significant differences among traffic offenders in a number of variables, including traffic violations that lead to demerit points and/or loss of license and crash involvement.

The results show that offenders are responsible of more accidents than non-offenders, but penalized offenders are responsible of more accidents with victims than convicted offenders. Whereas convicted are mostly drunk drivers, responsible of more serious offenses, sanctioned offenders are mostly recidivists, responsible of less serious but repeat offenses. However, their recidivist behavior and the countless traffic offenses they commit show this group of offenders as less concerned with other individuals, i.e., drivers at higher risk of accident and death. ^{Novoa et al. (2010)} and other similar studies (^{Castillo-Manzano et al., 2015}; ^{Pulido et al., 2010}; ^{Watson, Watson, Siskind, Fleiter, & Soole, 2015)} show this relationship between frequent penalties and disregard for public safety or causing death and harm and find correlation between a decrease in the number of sanctions and a reduction of the number of accidents and deaths. Our study supports this finding.

Differences in Personality and Alcohol Consumption

The second hypothesis was that there were significant differences between traffic offenders and non-offenders and between both subgroups of offenders in a set of explanatory factors of personality and alcohol consumption, as well as specific differences among offenders based on the reason for disqualification.

Our study also supports the belief that certain personality factors may influence a person’s proclivity to commit offenses. We find significant differences concerning personality and alcohol consumption between traffic offenders and non-offenders, as well as specific differences among offenders based on the reason for disqualification, a court mandate or demerit points. In that respect, penalized offenders have higher scores than non-offenders and convicted offenders on personality factors that different studies have proved to be associated with risky behaviors and negative attitudes towards traffic safety (^{Dahlen & White, 2006}; ^{Ledesma, Poó, & Peltzer, 2007}; ^{González-Iglesias & Gómez-Fraguela, 2010}; ^{Herráiz, Chamarro-Lusar, & Villamarín, 2011}; ^{Ulleberg, 2002}), such as sensation seeking, aggressiveness, anger, and alcohol abuse.

^{Aluja et al. (2010)} described sensation seekers as people who do not tolerate boredom and prefer different and exciting activities, as well as impulsive, not reflective, impatient individuals, who often make hasty decisions without considering the consequences. More than 30 years ago, ^{Weinstein (1980)} already suggested that sensation seekers have an “optimistic bias” and feel that the negative consequences of their actions are less likely to occur. These personal characteristics may lead to committing offenses while driving, repeatedly, without considering the legal consequences.

Regarding aggressiveness, ^{Aluja et al., (2010)} described aggressive individuals as prone to anger, having strong temperaments, who easily feel annoyed or disturbed and things get on their nerves. This trait may lead to unsafe behaviors while driving, such as speeding, reckless driving, and other actions involving endangerment, that are usually penalized. In our study, penalized offenders get higher scores on aggressiveness than convicted offenders, whose unsafe behavior is mainly due to their drinking problems rather than personality-related variables.

Anger has been described either as a personality factor related to aggressiveness, or a person’s emotion that elicits aggression or motivates aggressive behaviors (^{Egea-Caparrós et al., 2012}). Some authors such as ^{Underwood, Chapman, Wright, and Crundall (1999)} and ^{Deffenbacher et al., (1994)} consider aggressiveness as the behavioral expression of choleric states. The questionnaire used in this research evaluates anger in driving as one person’s proclivity to experience this emotion while driving in specific traffic situations (^{Herrero-Fernández, 2011}). The study shows that non-offenders display lower anger than offenders, and penalized offenders display the highest anger among offenders. By analyzing the specific factors one by one, the highest differences are registered in impeded progress and sanction hostility, meaning than penalized offenders are prone to feel angry and to show hostility over the possibility of being fined and while driving in heavy traffic conditions or other frustrating situations. Similar conclusions by comparing offenders and non-offenders are drawn in studies by ^{Sullman (2006)}, ^{Gómez-Fraguela, and González-Iglesias (2010)}, or ^{Egea-Caparrós et al. (2012)}.

Many studies have confirmed the relationship between traffic offenses, recidivism and drinking (^{Cavaiola, Strohmetz, & Abreo, 2007}; ^{Hubicka et al., 2008}; ^{Nochajski & Stasiewicz, 2006}). This study openly and clearly indicates the salience of alcohol problems among traffic offenders, especially among those disqualified by a court conviction, who are mostly drunk drivers. ^{Failde-Garrido et al. (2016)} report similar findings using the same instrument in a group of convicted traffic offenders, when compared with drivers with no criminal record. ^{Hubicka et al. (2008)} and ^{Cavaiola et al., (2007)} have also reported the relationship between alcohol dependence and the manifest proclivity to committing an offense.

Recent studies by ^{Useche, Ortiz, and Cendales (2017)} and ^{Montoro, Useche, Alonso, and Cendales (2018)} have shown an association between work-related psychosocial factors and individual characteristics of public transport drivers and the rate of traffic sanctions they are subject to, assessing the mediation of driving anger in this relationship. The former study examines the association between stress-related work conditions of professional drivers and risky driving behaviors, whereas the latter supports the idea that traffic penalties reported by public transport rates are preceded by work-related, personality, and other individual factors that, when combined with driving anger, enhance the occurrence of road misbehavior that may affect overall road safety.

Predictors of Disqualification from Driving

Finally, we conducted a third analysis to test whether changes in these explanatory factors would subsequently affect the probability of being banned from driving by serious or repeat traffic offenses, either with or without court mandate. The third hypothesis was that some of these factors, mainly alcohol consumption, will be a better predictor of disqualification from driving.

In fact, our regression analysis shows that high levels of activity, a penchant for thrill or sensation seeking, and some propensity to hostility while driving can satisfactorily predict group membership. When considering both subgroups of traffic offenders together, the probability of “disqualification from driving” is higher for drivers with alcohol problems and highly active drivers. When considering both subgroups individually, the probability of “disqualification from driving with a court conviction” is higher for drivers with alcohol problems, and the probability of “disqualification from driving without court conviction” is higher for drivers who get angry over the possibility of being fined and for sensation seekers. In both cases, alcohol issues are depicted as the best predictor of disqualification from driving by serious or repeat traffic offenses, either penalized or convicted.