THE FACT that human beings make erroneous estimates of their own ability to drive motor vehicles may have important safety implications (Horrey, Lesch, Mitsopoulos-Rubens, & Lee, 2015). Thus, assessment of own driving skills is an important dimension of road risk management (Boceara, Delhomme, Vidal-Gomel, & Rogalski, 2011) and can be of fundamental importance in both, reducing the likelihood of an accident and increasing safety in driving (Amado, Arikan, Kaça, Koyuncu, & Turkan, 2014).
The research by Kruger and Dunning (1999) showed results that became known as the Dunning-Kruger effect, a phenomenon that leads people with low academic performance to believe they know more than others. The authors state that participants who scored in the lower quartile in tests as grammar and logic overestimated their performance and thought they were above average. In contrast, participants who were in the upper quartile underestimated their skills when compared to their colleagues, not realizing that high skill is not necessarily an attribute shared with their peers.
Positive self-assessment can have diverse consequences. For example, there is a relationship between self-efficacy beliefs and academic performance (Cupani & Zalazar-Jaime, 2014). However, the overestimation of one's driving ability is probably more dangerous than its underestimation (Nakai & Usui, 2012), because it may increase a risky driving behavior (Isler, Starkey, & Drew, 2008; White, Cunningham, & Titchener, 2011), and it could cause non-assertive decision-making (Horrey et al., 2015; Kruger & Dunning, 1999).
Furthermore, overconfidence among drivers makes them less likely to adapt their behavior to traffic situations, leading to even more violations of traffic rules (De Craen, 2010; Isler et al., 2008). The question of overestimated driver's self-appreciation also reflects the impact of traffic safety campaigns: why people who overestimate their skills should pay attention to the instructions given to drivers in general since they are safer and drive better than these drivers? (Svenson, 1981). Hence, such drivers do not seem to identify themselves as requiring instruction (Svenson, 1981), because they believe that campaigns are aimed at drivers who are not so skillful and who are not as protected as they think they are (Horswill, Waylen, & Tofield, 2004). Accordingly, one study showed that interest in post-license driver training was greater in the case of less confident participants (Molina, Sanmartin, & Keskinen, 2013).
Thus, it is essential that people know about their actual skills (Zell & Krizan, 2014). The ability to drive is related to the level of knowledge and skills in traffic, as well as processing information and performance in different driving tasks, as maneuvers, detection of risks, etc. (Elander, West, & French, 1993).
In the second half of the last century, several authors researched people's self-assessment of their ability to drive automotive vehicles. All of these authors showed that drivers overestimated their ability when compared to the average, or with the "common driver" (Delhomme, 1991; Goszczynska & Roslan, 1989; Groeger & Grande, 1996; Mccormick, Walkey & Green, 1986; McKenna, Stanier, & Lewis, 1991; Svenson, 1981).
However, after the turn of the century, the idea that people overestimate their driving skills was refuted. For example, Sundstrom (2011) argued that the method of previous studies was not ideal for measuring such issue. In addition, De Craen, Twisk, Hagenzieker, Elffers, and Brookhuis (2011) point out that the method chosen to measure the self-assessment of driving skills has a great impact on conclusions that study reaches. Thus, the word "average" has a tendency to be understood negatively and thus encourages people to evaluate themselves more positively (Groeger, 2001).
According to Sundstrom (2008), the most common way of assessing the subjective skills of drivers has been to ask them to compare their own skill with that of "other drivers" or that of a "common driver". In the same way, according to the author, comparison with the "common driver" can be considered as an ambiguous and imprecise reference, generating several methodological problems as it results in biased evaluations, since there is a tendency for people to classify themselves in this category. Due to these methodological weaknesses, Sundstrom emphasizes the importance of considering another approach to measure the subjective ability of driving (Sundstrom, 2008). Therefore, the subjective measure of general driving ability can be related to the measure of the real driving capacity (Sundstrom, 2008).
Martinussen, M0ller and Prato (2017) compared the self-assessment with the measurement of a simulator and found that young male drivers' self-evaluations were inconsistent with their driving performance. Amado et al. (2014) also observed that people rated their ability better than the score assigned by an expert, using a naturalistic observation method.
Another important factor to be analyzed is that there is no consensus in literature if age and sex are related to how well people evaluate their driving skills. Studies have shown that young drivers have an even greater tendency to evaluate themselves as more skilled and less risk-prone than their peers (Harre, Foster, & O'Neill, 2005; Horswill et al., 2004; McKenna, 1993; Svenson, 1981; White et al., 2011). Besides, Mynttinen et al. (2009) say that inexperienced drivers have a realistic or underestimated assessment of their driving skills.
Difference between the responses of both sexes was found, concluding that males tend to evaluate their driving skills more optimistically than females (Boccara, Delhomme, Vidal-Gomel, Dommes, & Rogalski, 2010; Delhomme, 1991; McKenna et al., 1991). Other studies indicate that despite this fact, the differences between sexes are not statistically significant (Boccara et al., 2011; Nakai & Usui, 2012) or there are no significant differences at all (Sundstrom, 2011). Contrary to all these results, the research by Mynttinen et al. (2009) demonstrated that women overestimate abilities of "vehicle maneuver" and "economic driving" in relation to men.
As can be noticed, authors have different views about self-perception of driving ability: for some of them, sex (Boccara et al., 2010; Delhomme, 1991; McKenna et al., 1991) and age (Horswill et al., 2004; Svenson, 1981; White et al., 2011) is an important issue, for others it is not (Sundstrom, 2011). Some have studied people's beliefs about their ability (Delhomme, 1991; Freund, Colgrove, Burke, & McLeod, 2005; Goszczynska & Roslan, 1989; Mccormick et al., 1986; McKenna et al., 1991; Svenson, 1981) while others focused on testing the difference between the believed ability and the actual ability with practical tests (Mynttinen et al., 2009; Sundstrom, 2011).
In light of these considerations, the need for more research on the subject to solve these impasses and to test other methods suggested by authors who criticize the use of comparison with the "common driver" becomes evident. Cases in literature also supports the idea that further studies needed to be carried out on the self-assessment capability to improve road safety (Amado et al., 2014).
Despite an extensive body of literature on Traffic Psychology throughout the world, to the best of the authors' knowledge, there is yet no research on the subject that has been done in Brazil. This disagreement problem and lack of Brazilian research is intended to be solved with the present research. The vast majority of articles cited in the present study have been done in European countries as France, the United Kingdom, Sweden, Finland, the Netherlands and Spain, and other countries as the United States, Australia and Japan.
It is important to note that the countries mentioned have a different socioeconomic performance compared to Brazil. The Human Development Index of this country is well below the others, occupying the 79th position in the ranking (United Nations Development Programme, 2019). Another problem in Brazil is social inequality: the richest 10% accumulate around 42% of the country's total income (United Nations Development Programme, 2019). In terms of traffic mortality, Brazil is also the country with the worst rates among the countries mentioned. While Brazil has a rate of 19.7 deaths per 100,000 inhabitants, the rates for some of the countries mentioned are: 5.5 in France, 2.8 in Sweden, 4.7 in Finland, 5.6 in Australia, 4.1 in Japan, and 12.4 in the United States (World Health Organization, 2018).
Also, in Brazil, the car is not perceived only as an instrument for locomotion (Maoski, 2014). The symbolic and psychological factor attributed to the car may have an influence on how the driver acts and may be related to the high number of traffic accidents that occur in Brazil every year (Maoski, 2014).
Hence, understanding overestimation and influencing factors is of great importance to understand risk behaviors, and thus clarify the intervention needed to reduce them. This study aims to evaluate the assessment that people make about their own ability to drive, as well as to investigate if age, sex, and the amount of time people drive per week influence their perception of their skills as drivers. Also, a comparison is made between self-assessment and people's assessment of their friends' driving skills.
Method
Participants
The sample consisted of 151 Brazilians (50.3% male), between 18 and 58 years old (N=25.32, SD=1.66), from Curitiba (city located in the southern region of Brazil). For some analyses, the ages were divided in two groups: (a) group of young people, between 18 and 24 years old (N=102, M=21.41, SD=1.66); (b) group of adults, from 25 to 58 years old (N=49, M=33.47, SD=8.57). The criterion for participating was to have a driver's license. Regarding the schooling of the participants, 67.5% were undergraduates and the remaining ones were graduates (32.5%).
As for driving experience, participants have driving time ranging from 1 to 420 months (M=68.14, SD=85), with 7.9% driving for less than 12 months. In addition, they drive between 0 and 22 hours per week (M=7.42, SD=5.52). It is important to note that 9.9% of the sample drove less than one hour per week.
Whereas it was necessary to have Wi-Fi to apply the instruments, the convenience sample was chosen, that is, no procedure was adopted for data collection to be random. Thus, most data were collected on the campuses of the Federal University of Parana and a part was collected out of these. Participants were approached by being invited to participate of a study if they had a driver's license and then they were informed of the average time to answer the questionnaire.
Instruments
Questionnaire about driving abilities. The questionnaire developed was based on Hatakka, Keskinen, Laapotti, Katila and Kiiski (1992), and asks about one's own abilities in different aspects of driving ability. In addition, aspects that Mynttinen et al. (2009) have pointed out as important for the ability to drive were considered: parking maneuvers, control of traffic situations, showing consideration for pedestrians and cyclists, and avoiding risks. The author's point of not asking about the "common driver" was taken into consideration (Mynttinen et al., 2009). That is, it was not asked how the participant felt about the "common driver", but rather how he / she evaluated himself / herselfby performing each of the mentioned skills.
The questionnaire developed from these considerations asks the person to evaluate on a scale of 1 (terrible) to 5 (excellent) his / her own driving ability on each item. The option 3, more or less, means that the person is neither self-considered good nor bad in that specific activity. It is important to highlight that this option differs from the idea of "common driver", referring to self-comparing with the average of other drivers. In total, there are 38 items, divided into two groups: assessment of own abilities (19 items) and assessment of friends' abilities (same 19 items). Te questions covered skills of: parking and getting out of parking lots, overtaking, perceiving risks, adjusting the speed to the situation, driving in different weather conditions, driving in different ways, following the rules of the Brazilian Traffic Code, and paying attention to other people who compose the traffic (cyclists, pedestrians, etc.).
Driver Behavior Questionnaire (DBQ). After completing these scales, participants responded to the Driver Behavior Questionnaire (DBQ) by Lawton, Parker, Manstead and Stradling (1997), which was translated and adapted to Brazil by Bianchi and Summala (2002), containing 28 questions. Participants should indicate on a scale from 0 (never) to 5 (always) the frequency of each behavior. This questionnaire is composed of four factors present in the actions that drivers perform in traffic: Errors, Lapses, Ordinary Violations, and Aggressive Violations.
Errors can be understood as failures in actions planned to achieve a goal, which consequences have the potential to generate dangerous situations (Reason, Manstead, Stradling, Baxter, & Campbell, 1990). Lapses are behaviors that result from lack of attention or memory; these can cause embarrassment but usually do not affect safety (Reason, et al, 1990). Ordinary violations are intentional violations of practices that are essential to maintaining traffic safety, that is, violations of traffic laws (Lawton et al., 1997). Aggressive violations are more hostile in nature (Lawton et al., 1997). As the psychometric properties of the instrument, Bianchi and Sum-mala (2002) found Crombach's alpha from 0.79 for Errors, 0.82 for Ordinary Violations, 0.60 for Aggressive Violations, and 0.51 for Lapses.
Socio-demographic. Questions aiming at collecting socio-demographic data as sex, age, and schooling were asked as well as questions about the experience as a driver, how long the person had been driving and how many hours they usually drive weekly.
Procedures
Data collection procedure. Before responding to the questionnaire, people had to read and agree to the Free and Informed Commitment Term. Te whole process took about 10 minutes. Te research instruments were applied in tablets. They were previously transformed into a model of Google Forms. Tus, to apply them, it was necessary to have Wi-Fi Internet connection. Te sample was smaller than planned precisely because of this need for Wi-Fi at the time of data collection. Considering the population of Curitiba, 151 participants represent results with a 95% confidence level and an 8% margin of error.
Each participant received a tablet and mark his / her responses without the help from the researcher. Te sequence to answer was fixed: read and agree to the Free and Informed Commitment Term, answer the questionnaire about driving abilities (starting with items about their own abilities and then the same items about their friends' abilities), DBQ and, finally, answer the socio-demographic questionnaire.
Data analysis procedure. As data analysis, analysis of variance (ANOVA) was used to compare mean differences between groups to self-evaluation. To verify if there was a significant difference between the means of self-evaluation and evaluation for the friends, paired samples t-test was used. Te correlation test was performed to verify if there was a relation between the self-evaluation and other variables, as how long has the person been driving.
Results
Te instrument used was specifically developed for this study. Terefore, it was important to verify Cronbach's alpha. Tis coefficient, described by Cronbach (1951), aims to estimate the reliability of a questionnaire by measuring the correlation between answers. Values range from 0 to 1, with recommended values equal to or greater than .8 (Streiner, 2003). Cronbach's alpha of the self-assessment questionnaire was .83. Te participant was then asked to mark his / her friends' driving ability in the same way as the self-assessment, in this case, Cronbach's alpha was very high: .91.
Te mean of self-evaluation was 3.68 (SD=.48) and the evaluation for friends was 3.55 (SD=.58). Considering that the scale ranges from 1 to 5 points, these averages represent positive evaluations performed by participants. When using paired samples t-test with self-evaluation mean and evaluation for friends, it was possible to notice that there was a significant difference between the means of self-evaluation and the evaluation for friends (t(150)=2.47; p=.015). Tis difference showed that people self-rated themselves more positively in relation to the evaluation they performed for their friends.
Tere were significant correlations between self-assessment performance and evaluation for the friends, hour driven per week and age (see Table 1). Tis means that: (1) although the correlation is weak, a person who evaluates oneself well tends to evaluate well the driving skills of friends; (2) the longer the person drove a week, the more positive the self-assessment was; (3) the older the participant was, the more positive the self-assessment was. To make this type of analysis, the average of the responses attributed by the participant to their ability to perform each of the 19 traffic situations was considered. Furthermore, it was found that there is a positive correlation between how many hours people spend driving by week and age (r=.30; p<.01), showing that older people tended to drive more hours per week than the younger ones.
Self- evaluation | p | |
---|---|---|
Evaluation performed for the friends | 0.23 | p =.004 |
Hours driven per week | 0.33 | p <.001 |
Age | 0.20 | p =.017 |
Ordinary Violations | 0.11 | p =.170 |
Aggressive Violations | 0.17 | p =.043 |
Errors | -0.26 | p =.002 |
Lapses | -0.25 | p =.002 |
Regarding DBQ, the self-assessment showed a significant positive, but moderate correlation with Aggressive Violations, and negative with Errors and Lapses. That is, the more Aggressive Violations and the less Errors and Lapses the person made, in the past 12 months, the more positive self-assessment was. The values of the correlation tests performed can be seen in Table 1.
Considering the analysis between groups, there was a significant difference in the self-assessment according to the age group: people in the group over 24 years rated themselves better than those in the group of 18 to 24 years old. Another result is that males rated themselves more positively than females. Besides that, there was no significant difference between the self-assessment by the person's level of education. Table 2 shows the differences in self-assessment of the driving ability in relation to the age and sex groups mentioned.
Discussion and Conclusion
Literature of the late twentieth century shows a high prevalence of the idea that people perform a better self-assessment of their driving ability than the assessment of the average ability of other drivers (Delhomme, 1991; Freund et al., 2005; Goszczynska & Roslan, 1989; Groeger & Grande, 1996; Mccormick et al., 1986; McKenna et al., 1991; Svenson, 1981). In the present research, questions about the average of other drivers (or of the "common driver") were not asked, because of the fact that more recent studies (Groeger, 2001; Mynttinen et al., 2009; Sundstrom, 2008; Sundström, 2011) point out that this question generates methodological problems.
Thus, the developed instrument has questions about self-assessment and evaluation of specific driving skills of friends. Even with such a methodological change, the present study confirmed the phenomenon of high self-assessments of the ability to drive, previously demonstrated in Europe and the United States, also in Brazil. However, it should be noted that the present study couldn't confirm the Dunning-Kruger effect (Kruger & Dunning, 1999), since the skills are reported by the subjects themselves and are not actually measured.
The result found was a positive self-assessment: average of 3.68 (on a scale ranging from 1 to 5 points). In addition, the assessment for friends was 3.55, this difference in means is statistically significant, suggesting that people tend to self-evaluate better than evaluate their friends. This finding is in agreement with that presented by Horswill et al. (2004), in which drivers rated themselves superior to their peers in several components of driving ability (as risk perception, vehicle control, and overall driving ability). Similarly, Harre et al. (2005), showed that young drivers believe that they obey traffic rules more, drive safer and are more qualified, and have better reflexes than people of their age group.
The present study did not identify a correlation between how long the person has been driving and the self-assessment performed. However, it found a correlation between the driving time per week and the self-assessment. This result goes towards the result observed by De Craen et al. (2011), which showed that more experienced drivers tended to self-assess in a more optimistic way about their driving ability. The study of Martinussen et al. (2017) revealed that self-assessments of driving ability were more imprecise when the driver was less skilled than when driver was qualified. In addition, the result can also be explained with that presented by Mynttinen et al. (2009) who states that inexperienced drivers make underrated evaluations of their ability to drive.
However, there was a positive correlation between the self-assessment and the person's age, that is, the older the person is, the more positively he / she evaluates themselves. Also, when dividing the sample into two age groups (Group 1: 18 to 24 years; Group 2: over 24 years), there was a difference between the means of the two groups, reinforcing that age may be a problem to self-assessment. However, this result is not compatible with the ones found in the literature (Horswill et al., 2004; McKenna, 1993; Svenson, 1981; White et al., 2011). It is important to note that the older people were, the more hours they drive weekly, so this result can be attributed to the fact that the age factor is confused with the experience factor.
Another important fact is the difference between the sexes in relation to self-assessment. Several studies have suggested that males tend to evaluate their own driving skills better than females (Boccara et al., 2010; Delhomme, 1991; McKenna et al., 1991). However, this result runs counter to studies that claim that the difference between sexes is not statistically significant (Boccara et al., 2011; Nakai & Usui, 2012; Sundstrom, 2011). Therefore, the present research suggests that males tend to self-evaluate in a more optimistic way than female subjects.
In relation to DBQ, the present study suggests that people who commit fewer Errors are the ones that better self-evaluate. In addition, there was a positive correlation between self-evaluation and Ordinary Violations and Aggressive Violations. That is, the more the person positively self-assessed, the more Aggressive Violations and Ordinary Violations he / she committed. The literature shows that overconfidence among drivers causes them more difficulty in adapting their behavior to traffic situations, leading to more violations of traffic rules (De Craen, 2010; Isler et al., 2008). Therefore, it is possible to perceive that inflated self-assessment can cause concrete risks to people, because if the person believes to be a better driver, have the tendency to commit violations of traffic laws and aggressive behavior and endanger the lives of others as a result.
From the foregoing, the main objective of the study was reached. In other words, it was verified the self-assessment performed by people regarding their own ability to drive. In addition, other goals were also achieved, as comparing self-assessment with one's assessment of a friend's ability to drive, and understanding whether gender, age, schooling, and experience are related to self-assessment.
One limitation of the present study is the sample size and that only people from two categories of level of education (undergraduate and graduate) participated in the sample (not representative of the population). This limitation occurred due to the need of the Wi-Fi Internet for data collection, making it impossible to collect such data in the streets-which would further diversify the sample's schooling.
Because of this, it is not possible to say that schooling is not related to self-assessment, but only that there is no difference between self-assessment of the ability to drive between graduate and undergraduate people. Another limitation of the study is that all instruments of data collection depended on the self-report, that is, it was not verified whether the participants actually commit violations, but rather they were only asked about it.
The presented results and the literature suggest that people who self-evaluate very positively tend to have more traffic violations. Therefore, all the results found on the population that is most positively evaluated can be used when thinking about an intervention to remediate such evaluation and reduce risks in traffic. Thus, it is recommended that interventions in this direction be carried out. In addition, new studies on the theme need to be done, mainly in Brazil, to give more visibility and support to the theme.
Due to traffic safety, the subject of self-assessment is important to be discussed. From this research it is possible to think about the target audience that would most benefit from an intervention in the sense of reducing self-assessment: males over 24 and who spend more time driving per week would be the most needed audience to make a realistic perception about their driving skills. Interventions in this sense would be important because the data presented indicate that self-assessment, when optimistic, can generate risks in traffic. That being said, Martinussen et al. (2017) agree that the male group may benefit from intervention measures that take into account the accuracy of self-rated driving ability. And, knowing that road deaths are already considered a public health problem, this issue is of the utmost importance.