Meta-analysis of the correlation between personality characteristics and risky driving behaviors

Abstract: Background: A systematic review and meta-analysis was performed to determine the relationships risky driving behaviors (RDBs) have with the big five personality factors, sensation seeking and driving anger. Methods: The PubMed, EMBASE, Web of Science, Scopus, Psychinfo, and the Cochrane Library databases were systematically searched. All original studies were retrieved that assessed the relationships RDBs had with the big five personality factors, sensation seeking, and driving anger. Heterogeneity between studies was examined using the Cochran Q statistic and I2 tests. After applying Fisher’s r-to-z transformation, the correlation coefficients (r) were summarized from each study and 95% confidence intervals (CIs) were estimated. Results: Overall, 22 studies were included in the meta-analysis, which included 11211 participants. The results showed that RDBs had a significant negative relationship with agreeableness (r -0.27; 95% CI, -0.36, -0.19; P less than 0.0001), but significant positive relationships with neuroticism (r 0.16; 95% CI, 0.03, 0.29; P=0.584), sensation seeking (r 0.28; 95% CI: 0.23, 0.33; P less than 0.0001) and driving anger (r 0.39; 95% CI: 0.14, 0.64; P=0.002). Conversely, RDBs were not significantly related to extraversion (r -0.01; 95% CI, -0.08, 0.05; P=0.705), conscientiousness (r -0.05; 95% CI, -0.21, 0.12; P=0.584), or openness (r -0.06; 95% CI, -0.12, 0.00; P=0.065). Conclusions: Therefore, it appears that individuals most likely to engage in risky driving behaviors would be low in agreeableness, but high in neuroticism, sensation seeking and driving anger.


Introduction
ing someone to give way, and weaving in/out of traffic. [1][2][3][4][5] Previous research has shown that risky driving behaviors are related to collision involvement and greatly increase the chances of injury or death. [6][7][8] It has been shown that approximately 40 to 95 percent of road traffic injuries are as a result of risky driving behaviors. [4][5][8][9][10] RDBs are a multidimensional in nature and there are many factors that influence engagement in these types of behaviors. 2,[11][12][13] A driver's personality is one of the most important underlying causes and a variable strongly relevant to RDBs. [14][15] For example, one study showed that personality characteristics (PCs) explained more than 35 % of the variance in risky driving behavior. 15 Of the many PCs identified as potential predictors of RDBs, the big five personality factors (including extraversion, agreeableness, conscientiousness, neuroticism, and openness), sensation seeking, and driving anger have garnered the most support to date. 16 However, despite the fact that several studies have reported significant correlations (positive/negative) between PCs and RDBs, [16][17][18][19] a number of other studies did not find RBDs were significantly related to PCs, such as: agreeableness, 20 neuroticism, 6 extraversion, 20 openness, 21 sensation seeking, 22 and driving anger. 23 In a meta-analysis conducted by Demir et al. 24 they reported that driving anger had significant associations with aberrant driving behaviors (based on Driver Behavior Questionnaire -DBQ) factors. In another metaanalysis, using Iranian populations, it was found that more than 50 percent of drivers have sleep quality disorders. 25 Therefore, the relationships RDBs have with personality factors remains, to some degree, controversial and at the same time are very important.
Although several studies have examined the influence of personality on RDBs, we are aware of no other systematic review and meta-analysis which attempts to integrate and combine the results and draw conclusions about the effect of PCs (all of the big five personality factors, sensation seeking, and driving anger) on RDBs. The current study was performed to summarize the available evidence to establish the relationships that RDBs have with the big five personality factors, sensation seeking and driving anger.

Search strategy and selection studies
Eligible studies were identified using PubMed, Embase, Web of Science (WOS), Scopus, Psychinfo, and the Cochrane Library databases for published articles from inception up to December 2017, with two authors (MA & RT) independently searching each database. Also, the reference lists of identified studies were manually searched to increase sensitivity in the search strategy. The search was limited to publications in the English language. The databases were searched using the following keyword search terms: personality ["personality characteristic" OR "personality traits" OR "personality factors" OR "personality variables" OR "extraversion variable" OR "agreeableness variable" OR "conscientiousness variable" OR "neuroticism variable" OR "openness variable" OR "sensation seeking" OR "driving anger"] AND risky driving ["risky driving (RD)" OR "risky driving behaviors (RDBs)"].

Inclusion and exclusion criteria
Two authors (RT & MA) independently selected all relevant articles, if they met the following inclusion criteria:1) study was original research in the English language; 2) study investigated the correlation between PCs (including extraversion, agreeableness, conscientiousness, neuroticism, openness, sensation seeking, and driving anger) and RDBs; 3) study contained adequate data to calculate the correlation coefficients (r); 4) study used a standard questionnaire for measure personality traits; and 5) the study measured an aspect of risky driving.
Studies were excluded if they were not published in peer-reviewed journals or did not meet the lowest acceptable quality assessment score. Any disagreements were resolved by discussion and, if required, consensus was reached by consultation with the 3 rd author (K B.L). Figure 1 illustrates the details of the study selection process and results according to the preferred reporting items for systematic reviews and meta-analyses guidelines (PRISMA). Finally, after screening, 22 out of the 2261 published studies were included in our metaanalysis. 6,[15][16][17][18][19][20][21][22][23][32][33][34][35][36][37][38][39][40][41][42][43] The studies were published between 1994 and 2017. These 22 selected studies included 11211 participants; with a median number of 260 (range: 40-2604) participants per study. Ten studies reported correlation on extraversion, 14 on agreeableness, 17 on conscientiousness, 16 on neuroticism, 8 on openness, 13 on sensation seeking, and 7 on driving anger. The regions of the studies were: 11 from the European continent, 7 from the American continent and 4 from other parts of the world. The personality measures among study participants were the: IPIP in 11 studies, NEO-PI in 4 studies, and other measure in 7 studies. The key characteristics of the studies are presented in Table  1.

Main outcomes
Pooled estimates of the r between PCs (big five personality factors, sensation seeking, and driving anger) and RDBs are shown in Figure 2 and Figure 3. The correlations between PCs and RDBs, based on the subgroup and sensitivity analysis, are shown in Tables 2  and 3. The results of the subgroup and sensitivity analyses showed that the correlations were different in some of the specific subgroups for the measured variables and in each study.
In the sensitivity analysis, to determine the effect of each study on the strength of the relationship between big five personality factors and RDBs, the pooled r were estimated after excluding each study from the analysis. This meta-analysis showed no significant difference between the pre-and post-sensitivity pooled rs, but for openness there were significant differences between pre -0.06 (95% CI: -0.12, 0.00) and post-sensitivity pooled r-0.09 (95% CI: -0.13, -0.04), after omitting the Seibokaite et al. article 21 (Table 3).

Correlation between sensation seeking and driving anger with RDBs
Similar findings were observed for sensation seeking and driving anger. A significant positive correlation of r 0.28 (95% CI: 0.23, 0.33; P<0.0001) was found between RDBs and sensation seeking, while the relationship between RDBs and driving anger found a significant positive relationship of r 0.39 (95% CI: 0.14, 0.64; P=0.002) (Fig. 3). In the subgroup analyses, we found that the significant positive relationships that sensation seeking and driving anger had with RDBs were not influenced by continent, personality measures, age groups, or gender (Table 2). Sensitivity analyses were conducted, and the findings for sensation seeking and driving anger remained consistent with the pooled r. The lower and higher pooled r in the sensitivity analysis for sensation seeking were 0.26 (95% CI: 0.21, 0.31), after omitting the Hartos et al. 38 and 0.29 (95% CI: 0.25, 0.34) after omitting Falco et al., 23 respectively. For driving anger, a lower pooled r was found in the sensitivity analysis of 0.26 (95% CI: 0.12, 0.39) after omitting the Chraif et al. 19 and a higher pooled r of 0.45 (95% CI: 0.17, 0.73) after omitting Falco et al. 23 ( Table 3).   Because there was evidence of publication bias for conscientiousness (B = -9.95, P = 0.004) and openness (B= 2.78, P = 0.030), non-parametric analyses were performed (Duval and Tweedie). The meta-analysis based on the censored studies indicated that the pooled r on conscientiousness was not significantly changed before -0.05(95% CI, -0.21, 0.12) or -0.05 (95% CI, -0.21, 0.12) after the censored studies were included in the meta-analysis. For openness, the analysis indicated that the pooled r on openness significantly changed before -0.06 (95% CI, -0.12, 0.00) and after -0.09 (95% CI, -0.16, -0.03) when the censored studies were included in the meta-analysis. The finding for openness approximately agreed with what we expected. Therefore, according to the results of the Egger's regression tests the current findings were supported.

Discussion
As far as the authors are aware, this systematic review and meta-analysis is the first to study the relationships the big five personality factors (including extraversion, agreeableness, conscientiousness, neuroticism, and openness), sensation seeking, and driving anger have with RDBs among drivers from around the world. Our meta-analysis indicated that the relationship RDBs have with agreeableness was negative, and with neuroticism, sensation seeking, and driving anger there were positive relationships. In contrast, RDBs were not significantly related to extraversion, conscientiousness, or openness.
The finding of a negative relationship with agreeableness has been supported by previous researches, which have demonstrated that risky driving behavior can be predicted by agreeableness. These studies have reported that low levels of agreeableness can predict high levels of risky driving outcomes, including crashes. 15-17, 19, 21, 32, 35-37, 40, 42-43     This meta-analysis of primary studies also showed that neuroticism was positively related to RDBs, meaning that higher levels of neuroticism were related to higher level of RDBs and vice versa. This finding has been supported by several studies which have shown that those high in neuroticism show more risky driving behaviours. 17, 19, 21, 34-35, 37, 39, 42 In contrast, two studies reported that neuroticism had no relationship with RDBs. 6,41 The most likely reason for this discrepancy may be that these studies did not have sufficient sample sizes to answer the questions being studied. Furthermore, a small number of studies have reported that neuroticism had a significant negative relationship with RDBs. 16,40,43 Perhaps this dissimilar finding was due to the mean age of participants in these studies, which mainly consisted of young adults. Our subgroup analysis indicated that in Similar to previous studies, our meta-analysis confirmed that sensation-seeking and driving anger had significant positive relationships with risky driving behaviours. 15-16, 34, 36 Johan et al. also found that persons with high levels of sensation-seeking perform risky driving behaviors to satisfy their need for strong emotions, despite perceiving the risk associate with some risky behaviours. 14 Consequently, individuals with high levels of sensation-seeking are exposed to an elevated driving risk, so effective interventions need to be investigated in future studies.
Previous research on driving anger has found significant positive relationships with risky driving behaviors, including losing control of one's vehicle. 16 Previous research has also shown that driving anger is common. 44 However, we need to investigate relevant interventions to deal with driving related anger. In research conducted by Deffenbacher et al. cognitive and physical relaxation interventions were found to significantly reduce risky driving behaviours. 45

Strengths and limitations
This study, like other study, has some potential strengths and limitations. Firstly, although the number of participants included in our meta-analysis was large, according to the subgroup analysis there were limitations with certain subgroups, which may limit our ability to generalize our findings. Secondly, the present study only included published articles that reported data we could use to estimate the pooled r, which resulted in the exclusion of many other studies. However, Egger's test indicated no evidence of publication bias among the included studies and a random effects model was used to decrease the heterogeneity effects on the pooled r. Thus, the present study's findings appear to be reliable.

Conclusions
Overall, despite these limitations the current metaanalysis demonstrated that the relationship between RDBs and agreeableness was negative, and with neuroticism, sensation seeking, and driving anger there were positive correlations. Given these findings, we need to pay more attention to the importance of traffic psychology in order to reduce and control risky driving behaviors. An additional prospective study with a larg- Abbreviation: r ; correlation coefficient.