**3. Methodology**

This study was conducted in the Emirate of Abu Dhabi in the United Arab Emirates (UAE). A review of relevant past literature was undertaken to understand the state of research in the area of traffic sign comprehension levels among motorists. Traffic comprehension studies at the global, regional and country level were explored. The UAE is a melting pot of people from across the globe working or visiting for tourism. Despite the efficient public transportation, residents often opt to own and drive their own cars, resulting in a highly diverse respondent population within the survey sample. A Likert-based questionnaire was designed based on similar ones used for past studies by Al-Madani and Al-Janahi [11]. The questionnaire was composed of two parts. Part one intended to collect information about drivers' age, gender, nationality, marital status, education level, number of years having a driving license and whether they have multiple ones, the number of fines and number of accidents they have had. Part two of the questionnaire was designed to present to the respondent a list of 18 traffic signs (including regulatory, warning, and guide sign types) with a multiple-choice query so they could pick the meaning of the traffic sign from the offered choices.

Purposive sampling technique [24] was adopted for the administration of the questionnaires. The respondents were chosen among visitors to three major shopping malls within Abu Dhabi, including Marina mall, Al Wahda mall and Abu Dhabi Mall. Due to the often high temperatures, these are the places many residents like to visit and stay during the day, engaging in shopping, dining and amusement activities for children. The parking managers at the three malls were used to distribute the questionnaires to drivers on their way in and they dropped off the questionnaires on their way out. About 203 questionnaires were returned out of the 300 that were distributed. However, only 200 were used for the analysis; three questionnaires were removed for incomplete information. The number of respondents was comparable with previous studies that reported 32 respondents, 59 respondents and 53 respondents used by Alnuaimi and Al Mohsin [25], Ruqaishi and Bashir [26] and Oyegoke and Al Kiyumi [27], respectively. The response rate of about 67% can be regarded as good considering the low responses to surveys within the GCC, knowing that the average response rate for survey data collection from organizations was 35.7% [28].

#### **4. Results**

The analyses involved 200 respondents who owned a UAE drivers' license as a minimum requirement to be eligible to complete the questionnaire. The respondent's educational profile showed that there were five respondents with a PhD., 10 respondents with a Master's degree, 85 respondents with a B.Sc. degree, 36 with a Diploma certificate and 64 respondents with a high school certificate. In terms of marriage status, there were 88 married and 112 single respondents within the sample. The respondents were a rich reflection of the basic profile of Abu Dhabi's residents, with 20 respondents from Europe, 10 Americans, 42 Asians, 19 Africans and 109 Middle Easterners, as shown in Table 1 below.

**Table 1.** Crosstab marital status vs. region of respondents.


A further analysis of the respondents' profile, shown in Table 2, revealed that respondents with only one country's drivers' license constituted 50.5%, while respondents with over three countries' drivers' license constituted only 4% of the total respondents. The respondents' profile also showed that 170 respondents drove their private cars, 12 were taxi drivers, 7 drove a bus or a lorry, while 11 rode motorcycles.


**Table 2.** Crosstab of vehicle type vs. number of drivers' licenses.

A normality test was conducted on the data to determine the distribution and decide whether parametric or non-parametric procedures would be used in the analysis. Both Kolmogorov-Smirnov and Shapiro-Wilk tests, presented in Table 3, revealed that the data violates normality assumptions, hence non-parametric procedures were employed in analyzing the data [29].

#### **Table 3.** Tests of normality.


<sup>a</sup> Lilliefors Significance Correction.

There are two variables within the data set that qualify for analysis using the Mann-Whitney U test for independent samples: they are the gender and marital status variables, while the Kruskal-Wallis H test was used for independent samples with more than two categories. The Mann-Whitney U test to determine whether there exists a statistically significant difference in responses between the genders (males and females) and respondents' marital status returned non-significant results, indicating that these independent variables did not affect road sign comprehension in any way, as shown in Table 4 below.



Kruskal-Wallis H test was conducted on the independent variables of age, academic qualifications, type of vehicle driven, and the region of origin of the respondents. The results shown in Table 5 reveal that there was no statistically significant difference across the respondents based on age of respondents (*p*-value = 0.184); academic qualification (*p*-value = 0.217); type of vehicle driven (*p*-value = 0.541); and region of respondents (*p*-value = 0.390). The implication of this result is that road sign comprehension capacity is not affected by a driver's age, academic qualification, type of vehicle or the geographical region from which the driver originates. This last result is especially important given the fact that non-citizen residents in the UAE are treated based on their country of origin when seeking to convert their home-country drivers' licenses to a UAE drivers' license. However, the test revealed that there was a statistically significant difference based on respondents' number of licenses.

**Table 5.** Kruskal Wallis H test results.


To investigate further and determine which group was responsible for the difference, a Kruskal-Wallis one-way ANOVA was conducted, and the results are shown in Table 6 below. This test compares the median of the distribution to determine which group differs from the others [29]. The group with more than three licenses was responsible for the difference, even though the number of cases in this group was very small (8). However, due to their very small number, it would be unwise to generalize the results of this research endeavor.

**Table 6.** Kruskal-Wallis One-Way ANOVA test.


The overall average performance of the study sample was 77% across all 200 respondents in terms of their ability to correctly recognize the road traffic signs presented to them. This is an improvement on the earlier study which found a 56% average performance in the comprehension of traffic signs [11]. The performance of the sample is shown in Table 7. The performance of the respondents was below 90% on 8 out of the 18 road signs that were presented. Good comprehension of road signs reduces driving mistakes on the road, and consequently a reduction of accidents on the roads. However, the government relies on punitive measures such as fines to ensure adherence and compliance. As the results above reveal, there are a couple of areas where more sensitizations will be required, especially because all the signs used in the questionnaire were taken from the UAE's Emirates Driving Institutes' (EDI) Guidance chart for road signs, road markings and traffic signals.

A further breakdown of the performance of the respondents according to gender (see Table 8) reveals a very interesting trend. A higher percentage of males failed to correctly name each of the "advance warning signs" they were presented compared to their female counterparts. This finding is contrary to the earlier GCC-wide study by Al-Madani and Al-Janahi [11] who found that male drivers performed better than females.

Interestingly, all the road signs where the respondents failed to pick the correct answer were contained in the "advance warning signs" section of the EDI Guidance chart. Therefore, it has become pertinent that training instructors and police driving assessment officers pay more attention to these areas when training or assessing potential drivers, respectively. Furthermore, the government should consider a traffic sign retest exam for all drivers when they apply for their driving license renewal. While the importance of driver education is recommended [20] and aggressive campaigns have been suggested [21], it has also been empirically proven that driver training [3] holds the most potent potential for enhancing safety on our roads.


**Table 7.** Road sign comprehension performance.


**Table 8.** Percentage of the seven road sign comprehension failures by gender.

#### **5. Discussion**

As indicated earlier, the respondents were a rich reflection of the basic profile of Abu Dhabi's residents, which makes the study results more realistic and reflects the real prevailing situation of sign comprehension of drivers in the Emirate of Abu Dhabi.

The results revealed that there was no statistically significant difference in responses between the genders (males and females) and respondents' marital status, indicating that these independent variables did not affect road sign comprehension in any way. Both results tend to be an improvement on an earlier study conducted by Al-Madani and Al-Janahi [11] across the GCC countries. However, the majority of earlier studies have consistently found that differences exist between the genders when it comes to trafficrelated issues and accidents [30–33]. Cordellieri et al. [30] and McKenna et al. [32], however, found that these differences do not exist in terms of driving risk perception, while the SIRC [33] argued that these differences can best be explained by virture of fundamental evolutionary differences in behavior and psychology rather than levels of competence and driving skills. In a related study, Massie and Campbell [31] found that the differences witnessed between male and female drivers disappear by the age of 60 and onwards. The results from this study may, however, have been impacted by the low number of female respondents, which was about 27%. Another possible explanation could be the large number of under 40 respondents (76%), since it has been found that age tends to mediate the observed difference [31,34].

Moreover, several earlier studies on gender revealed that males were in general better than females regarding traffic sign comprehension [19,35]. Al-Madani and Al-Janahi [36] indicated that posted signs were comprehended better by male drivers than females in Arab, European and American countries. Similarly, Hawkins Jr et al. [37] reported that men better recognized one fifth of the warning signs correctly. When education background is controlled for, the differences between males and females may become less remarkable [35]. Gender was found to have no statistically significant impact in a study that considered male and female European and American drivers having at least undergraduate degrees [11]. Likewise, Ng and Chan [38] indicated that there was no significant difference for traffic sign guessability between males and females of similar education levels. In regards to marital status, the findings of this study were in line with the finding of Ismail [19], who also found that marital status had no significant effect on the comprehension level of drivers.

The results revealed that there was no statistically significant difference across the respondents based on age, academic qualifications, type of vehicle driven and region of respondents. The implication of this result is that road sign comprehension capacity is not affected by a driver's age, academic qualification, type of vehicle and the geographical region from which the driver originates. These results disagree with Al-Madani and Al-Janahi [11], who found that the driver's education level, income range and nationality impacted ability to correctly identify traffic signs in countries within the GCC. However, this could be interpreted as an improvement in the training and testing methods in Abu Dhabi. Previous studies in general indicated that drivers' comprehension of traffic signs increased with increasing education levels [19,35,38].

As presented in the Results section, the overall average performance of the study sample was 77%, reflecting respondents' ability to correctly recognize the road traffic signs presented to them. This result can be considered as an improvement on the earlier study which found a 56% average performance in the comprehension of traffic signs, and 62% for the UAE in particular [11]. Similar studies showed that drivers could not comprehend a large percentage of the road signs presented to them [12]. In Turkey, participants correctly identified 81% of the signs presented [13]. In Kano city, Nigeria, it was found that the average traffic sign comprehension level was as high as 79% [14]. A study of drivers' sign comprehension across four countries (Poland, Finland, Canada and Israel) found that on average the drivers only identified 58.5% of the signs correctly [5]. This goes to show that the problem is not limited to developing countries only, as developed countries are also struggling.

The timeliness and the robustness of traffic sign comprehension while driving is very important to drivers. Safe driving and accident avoidance can be ensured when a timely visual recognition of traffic signs takes place [21,39,40]. Zhang et al. [41] clustered the factors that affect a traffic sign's visual recognizability into three groups: (1) geometric factors of traffic sign, such as sign's size, placement, mounting height, aim, depression angle, occlusion angle, curvature of road and changing road surface [41]; (2) movement factors of vehicle, such as speed, Geometric Field Of View (GFOV), which decreases with speed being increased [42], and height and direction of line of sight which governs whether the traffic sign is contained by the GFOV; and (3) other factors, for example, weather conditions [43], lighting conditions [44], drivers' view reaction time which is affected by drivers' age and sight capabilities [44], and traffic density cognitive burden [45]. Moreover, occlusion may cause interruptions to visual continuity, thus affecting the recognition of the traffic signs [46]. Considering the abovementioned factors that affect the recognizability of the traffic sign in real-life driving practice, it is expected that the level of comprehension of drivers will reduce to an extent that might be significant if the above factors are not considered in the design and placement of these signs.

Traffic Sign Recognition (TSR) systems have recently become common in car manufacturing industries. Road safety enhancement can be achieved using TSR technology by providing timely warnings and dynamically assisting drivers in adhering to posted speed limits [47]. The most basic TSR systems employ vehicle-mounted cameras for identification of traffic signs. Their main function is to notify the driver of near traffic signs in a process that starts with scanning the roadside for signs, then using real-time image processing software to identify and inform the driver by a sound or displaying a symbol on the vehicle's dashboard panel [48]. Zhang et al. [41] indicated that the most current traffic sign detection and classification methods are based on color and shape information within images or videos. These methods use color information to segment the sign candidate area, and then to extract the traffic sign. Challenges to the vision-based TSR systems include variations in lighting and weather conditions, occluded signs and image distortion due to vehicle motion [41,48]. Zhang et al. [41] pointed out that there are different methods for recognition evaluation (simulator and image methods and naturalistic driving experimentation), but each has its own limitations. As indicated earlier, deployment of TSR systems will bring great enhancement to road safety and reduce accident involvement. Meanwhile, limitations to TSR deployment should be addressed, which include fixing conflicts in signage design, cooperation between the vehicle industry and road operators to achieve reliability of the TSR system, and adaptation with continuous development of the TSR technology which is expected to become more accurate and more reliable [48].

The study outcomes clearly show that traffic safety can be enhanced by means of providing well maintained, readable and well-placed signs; driver education and training; inclusion of TSR technology into the transportation system; and consideration of traffic sign retest exams for all drivers when they apply for their driving license renewal.

#### **6. Conclusions**

Traffic signs are components of road infrastructures that are meant to assist road users in safely navigating to their destinations. This study sought to assess the comprehension level of drivers in the Emirate of Abu Dhabi (UAE). The respondents' overall performance was an improvement on earlier studies, however, there is still much to be improved upon. Although earlier studies found that driver's age, marital status, gender, nationality, level of education and type of vehicle all had a statistically significant impact on road sign comprehension, the data analysis for this study did not support such a conclusion. Furthermore, contrary to earlier studies, a higher percentage of males failed to correctly identify the "advance warning signs" compared to female respondents. One of the most critical findings from this study is the ability to identify the category of road signs that are problematic to drivers in Abu Dhabi. This finding should assist driving instructors and police driving assessment officers in redesigning their teaching materials and assessments, respectively. The fact that despite increased fines driving violations persist points to the need to shine a spotlight on the training of drivers and the assessment methods for issuing licenses, as well as requirements for license renewal.

**Author Contributions:** Conceptualization, T.M.A.-R.; methodology, T.M.A.-R.; formal analysis, T.M.A.-R., A.A.U.; resources, T.M.A.-R., A.A.U.; data curation, T.M.A.-R., A.A.U.; writing—original draft, T.M.A.-R., A.A.U.; writing—review and editing, T.M.A.-R., A.A.U.; supervision, T.M.A.-R.; project administration, T.M.A.-R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research did not receive any specific grants from any funding agencies.

**Data Availability Statement:** All data of this research were presented in the article.

**Acknowledgments:** The authors would like to thank the following individuals for their help in distributing and collecting the survey questionnaires: Samir Aldraidi, Mahmoud M. Ayman, Abdullah S. Abueidda and Mohammed S. Almansoori.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

