Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 47

T A B L E

O F

C O N T E N T S

ABSTRACT............................................................................................. INTRODUCTION................................................................................... 1. 2. 3. LITERATURE REVIEW... METHODOLOGY.................................. RESULTS................................................................................................ CONCLUSIONS AND 4. RECOMMENDATIONS......................................................................

2 3 4 7 12

18

5.

REFERENCES.....................................................................................

21

6.

BIBLIOGRAPHY................................................................................... APPENDIX 1: SAMPLE SIZE CASLCULATOR.............................. APPENDIX 2: SURVEY........................................................

24 25 26

APPENDIX 3: EXPERIMENT.............................................................. 31 APPENDIX 4: RESULTS ANALYSIS (first 2 objectives)................. APPENDIX 5: RESULTS ANALYSIS (3rd and 4th objectives)...... 35 47

ABSTRACT Failure to heed traffic signs is one of the most common causes of road accidents. The success of effective communication of traffic sign messages to road users depends not only on driver characteristics but also on the signs themselves. This paper addresses the effects of driver characteristics and sign features on the understandability of traffic signs. Driver characteristics considered here include: income, education, age, marital status and gender. Sign features examined here contain: presentation model, familiarity, shape and color. The population was sampled from different districts of Tashkent city. The study was divided into two stages: survey and experiment. The main purpose of the survey was to obtain information regarding driver characteristics, whereas experiment was primarily employed to test the sign comprehension level of the drivers. The results indicated that income, education and gender have a significant effect on comprehension of traffic signs, while marital status has only paltry effect. In addition, the differences in comprehension were revealed between two categories of drivers. Namely older drivers with high income and level of education comprehend signs better than their young counterpart with low level of income and education. Furthermore, familiarity, color and shape of the traffic sign were found to be highly correlated with sign understandability. Finally, the traffic sign system proved to be effective because it is based mainly on symbolical models of the signs which are found to be more eye-catching than verbal ones. These findings are believed to be important for the road sign designers. In addition, they might be useful for relevant organizations aimed at increasing the effectiveness of traffic education.

INTRODUCTION

RESEARCH QUESTION: How do driver factors and design features affect comprehension of traffic signs? RESEARCH OBJECTIVES 1. 2. 3. 4. to identify general factors that influence comprehension of traffic signs based on the to compare these findings with those indicated by a sample from Tashkent to reveal differences in comprehension among several categories of drivers to evaluate the effectiveness of traffic sign system in Tashkent studies of researchers from US, Europe, Asia and Arabic countries

Traffic signs serve as one of the most common tools for traffic control. Their main purpose is to regulate, warn and guide road users in a traffic system (Dewar and Olson, cited in Ng, 2007). A 'proper' traffic sign posted in the right place allows the road users to avoid problems on the road, which could be as simple as traffic slow down or as bad as fatal accidents (Kurniawan and Zaphiris, 2001). Despite their importance traffic signs are not always applied effectively. Indeed, seventy percent of traffic signs are ignored by drivers (Shulz, 2006). This evidence shows an importance of a detailed analysis of key factors influencing the effectiveness of traffic signs. Conspicuity, reaction time, legibility distance, glance legibility and comprehensibility1 are considered to be vital in traffic sign design. Among all these factors engineers from Australia, New Zealand, Canada, and the USA rated comprehensibility as the most important criteria (Dewar et al., cited in Shinar et al., 2003). However, in a multi-country research, which included five Arabic countries, it was found that drivers comprehended only about 56% of the 28 signs presented to them (Al-Madani and Al-Janahi, 2002). This demonstrates an urgent need for a deep investigation of factors affecting comprehension performance of road users.

Comprehensibility of a sign is a measure of how readily an observer can understand the message intended to be conveyed by the sign. (CIE, cited in Ng, 2007)

LITERATURE REVIEW The importance of setting the research project within a conceptual context should not be underestimated. A lot of previous related studies were found to ensure the credibility of the current investigation.

Indeed, such driver factors as age, marital status, income, education, driving experience and nationality were analyzed by prominent researchers from the US, Europe, Asia and Arabic countries to find how they impact on comprehension performance. There are a lot of similarities as well as differences in the results of surveys conducted. Researches on effect of income, nationality, education, marital status and driving experience demonstrated similar results. For example, Al-Madani and Al-Janahi (2002) found that:
Western drivers comprehend the signs significantly better than drivers of other nationalities. Drivers, in the various experience categories, holding low educational qualifications and in low income categories comprehend signs significantly less well than those who are holding high level of education and with high income. Excluding drivers between 3544 years of age, drivers comprehension of signs is not related with experience for any other age group. Understanding traffic signs does not change significantly with years of driving experience in female drivers. Male drivers with over ten years of driving experience are significantly better than less experienced male drivers. Single and married drivers understand the signs equally well.

These results comply with the findings of Al Gadhi (1994) who also found that there is a positive relationship between education and income and comprehension performance, while there is no correlation between single and married drivers. Ng and Chan (2008) agree with Al Madani and AlJanahi on that the driving experience has no effect on sign comprehension. In some cases the results of the studies contradict rather than comply with each other. For example, there are continuing debates on how age affects the comprehension performance of the drivers. Some researchers found that older drivers have poorer understanding of traffic symbols than younger drivers do. Dewar et al. (1994) studied the comprehension level of 85 US traffic signs for drivers from Texas, Idaho, Alberta and Canada. It was found that there was no difference for 52 4

signs analyzed, meanwhile drivers in the older age group (60 and over) understood less well than the younger ones for the remaining 33 signs. The supporting evidence was provided by another survey (Shinar et al., 2003) conducted mainly in European countries. Nevertheless, if Arabian countries are considered, the results are absolutely different there. Indeed, Al-Madani and Al-Janahi (2002) found that comprehension of traffic signs is positively correlated with drivers age, which contradicts the previously mentioned findings. Interestingly, there is also a third approach to evaluation of the correlation between age and comprehension. To be more precise, Ng and Chan (2008) found that driver factors of age group had no effect on comprehension performance. The most rational explanation for such a split of opinions is the difference in nationalities sampled, time during which the surveys were conducted, distinctions in design features of traffic signs and some other specific factors. Not only driver factors but also the design features of the signs affect the success of effective communication of traffic sign messages to road users. Some researchers investigated the relationship between compliance with ergonomic principles2 and comprehensibility of traffic signs. In 2004, Shinar and Ben-Bassat in their study Are ergonomically designed traffic signs more comprehensible? tested directly the the relationship between sign comprehension and the extent that the sign complies with three ergonomic principles: sign-content compatibility, familiarity, and standardization. The results of this study illustrated strong correlation between sign compliance with each of three ergonomic principles and signs comprehension probability. A positive relationship between familiarity and comprehensibility was also identified by Dewar et al. (1994) as well as Ng and Chan (2008). Other researchers emphasized the importance of sign location, color and shape in comprehensibility of traffic signs. According to Borowsky et al. (2008) drivers were less likely to identify the traffic sign when it was located in an unexpected location. The results of the study led to the following conclusion: to increase their timely probability identification, traffic signs should be posted in
2

There are five main ergonomic principles relevant to traffic signs design (Sanders and McCormick, cited in Ben-Bassat, 2003): 1) Spatial compatibility the physical arrangement in space, relative to the position of information and directions 2) Conceptual compatibility the extent to which symbols and codes conform to peoples associations 3) Physical representation the similarity between the content of the sign and the reality it represents 4) Familiarity the extent to which the driver is familiar with the sign from his driving experience 5) Standardization the extent to which the codes used for different dimensions like color

and shape are consistent for all signs

expected locations (Borowsky et al., 2008). However, proper location alone is not sufficient unless it is accompanied by appropriate color and shape. Thus, Gao, Podladchikova and Shaposhnikov (2003) applying specific computational visual models, found that traffic signs are better recognized when their color and shape contrast with that of background (e.g. billboards, trees). For example, commonly used green colored direction sign will be hard to recognize if it is posted near trees. On the other hand, the findings of Zakowska (2001) demonstrate that adjustment of traffic signs to the background may lead to misunderstanding, since in this case the same message will be displayed by different signs Before applying the above mentioned sources in our research the strengths and weaknesses should be critically assessed. The studies of Al-Madani and Al-Janahi(2003), Zakowska (2001), Ng and Chan (2008), Ng (2007) appear to be the basis for our future research, since they are relevant, upto-date and based on large samples from a wide range of countries. These studies helped us to identify the main factors, which affect comprehension and hence effectiveness of traffic signs. Furthermore, a lot of researchers, e.g. Ben-Bassat (2003) and Shinar et al.(2003), use findings of the above mentioned investigators in their studies. However, the results of some of the reviewed sources can not be generalized due to the following limitations. For example, the findings of Dewar et al.(1994) and Al-Gadhi et al.(1994) obtained 15 years ago could be obsolete and hence not relevant in our research. Some results (e.g. Al-Madani, Al-Janahi, Ng ) of the investigations were distorted as there were a limited number of females in the experiments conducted. In addition, the questionnaires were not completely filled in by respondents. For example, only 70% of the questions were answered in Hong Kong (Ng and Chan, 2008). All in all, a thorough research concerning influence of different factors on traffic sign comprehension was made in developed countries as well as in developing ones. However, the main emphasis was placed on American, European and Arabic countries with practically no study of Central Asia. In the following report, a detailed research is conducted in the capital of Uzbekistan Tashkent - with the aim of identification the degree of dependence of traffic signs comprehension on specific driver factors and design features. Furthermore, the weaknesses of the current traffic signs system will be identified and ways of improvement will be suggested.

METHODOLOGY
APPROACH AND HYPOTHESES The research is carried out using deductive approach for several reasons. Firstly, a cause-effect link between selected variables will be investigated without explaining the nature of such relationship. For example, it was found that income of the driver is positively correlated with ones comprehensibility of traffic sign. However, the reasons, why rich people better recognize the signs will not be mentioned. Secondly, a lot of theoretical concepts, describing the impact of different factors on traffic signs, have already been developed. Therefore, in our research we are not going to formulate a theory. Instead, we are planning to test established theories through corresponding hypotheses with a purpose of confirming, rejecting or modifying the theories depending on their applicability to Tashkent. The following hypotheses are formulated: 1. Monthly income, education, and age have a significant effect on comprehension of traffic signs, while marital status and gender have no or paltry effect. 2. Pictorial sings are more eye-catching than verbal ones ceteris paribus. 3. Familiarity, color and shape of traffic signs are strongly correlated with their understandability. 4. Old drivers with high income and level of education recognize sings better than young ones with low income and level of education.

SAMPLING In order to test the above stated hypotheses, first of all, the suitable sample size as well as technique should be identified. Particularly, sample size of 384 was estimated using Raosoft calculator (Raosoft sample size calculator, 2004) (Appendix 1). This sample was marked as a recommended one given 5% margin of error (taken by researchers in most cases), nearly 440 thousand population3, and normal probability distribution.

Jalilov, J.R., personal communication with the head of Uchtepa district GIBDD, (01 April 2009).

Turning to the sampling technique, stratified random sampling seems to be the most appropriate for the research. The stratification was done proportionally according to the number of cars registered per each district (Table 1 and Graph 1). This was done to ensure that sample is as close to population in terms of driver characteristics as possible (choosing only one district may cause distortion of results as the drivers living in this district may not have the same driver characteristics as all drivers of Tashkent) Later, the desired number of respondents is going to be picked randomly at the randomly chosen car parks in each of the districts so as to avoid bias in the selection of samples, thus ensuring reliability and validity of sampling technique. Table 1: Proportional stratification according to the number of cars registered per district Name of the district Mirobad Yakkasaray Mirzo Ulugbek Shayhantour Yunusabad Chilanzar Hamza Sergeli Sabir Rahimov Uchtepa Bektemir TOTAL Number of registered cars per district as at 1 January 20094 (in thousands) 57.2 30.8 74.8 17.6 39.6 61.6 39.6 26.4 48.4 30.8 13.2 440
Graph 1: Sam size in each of the districts ple 27 11 50 42 27 23 35 54 35 15
Yakkasaray Shayhantour Chilanzar Sergeli Uchtepa

Percentage out of total (%) 13 7 17 4 9 14 9 6 11 7 3 100

Sample size

50 27 65 15 35 54 35 23 42 27 11 384

65

Mirobad Mirzo Ulugbek Yunusabad Hamza Sabir Rahimov Bektemir

Jalilov, J.R., personal communication with the head of Uchtepa district GIBDD, (01 April 2009).

STRATEGIES In our cross-sectional study, we will employ a combination of survey and experiment, because our enquiry has several different aims, which will be achieved through the use of corresponding methods. Survey will enable us to gather a considerable amount of data in a timely and efficient manner. Furthermore, due to the simplicity and familiarity of the survey, a high respondent rate is expected. On the other hand, experimental strategy will be adopted to test drivers ability to recognize a wide range of traffic signs within certain time constraints. Use of multi-methods will enables us to triangulate the data, in other words, it will ensure that the data obtained will be interpreted in a right way. For instance, the findings obtained through questionnaires will be supported by data from an experiment. Besides, since data are affected by the method used, combination of several methods will minimize so called method effect. RESEARCH INSTRUMENTS Strategy: Survey Research instrument: Self-administered questionnaire. The primary purpose of the survey is to collect standardized data necessary for testing the first two hypotheses. For this reason, self-administered delivery and collection questionnaire appears to be the most suitable technique. In addition, compared to other research instruments, such as telephone interview, questionnaires are less expensive in terms of time and money as well as easier to analyze. The questionnaires mainly concerned with respondents biographical information would be distributed among sampled at randomly chosen car parks in each of the districts. Respondents are expected to answer multiple-choice, numerical, open-ended and ranking questions to identify their driver characteristics (age, gender, etc.), to evaluate their comprehension of signs and to assess the importance of some features of traffic signs (Appendix 2). To ensure that the data collected via questionnaires will enable a research question to be answered, a pilot test with 15 volunteers will be conducted. Thereby, questions validity and suitability, reliability of the data obtained as well as clarity of instructions will be assessed. Strategy: Experiment Research instruments: Sign comprehension test In the next stage of our study, experimental strategy will be employed to test drivers ability for proper sign recognition (Appendix 3). For this reason, sign comprehension test would be administered. Approximately 11 signs from different categories (regulatory, warning, guide signs) 9

would be presented to respondents who are expected to call names of each of the signs or at least their meaning. In addition, 7 pictorial and verbal signs with the same meaning content are going to be shown. These signs would be placed far away from respondents and displayed only during 20 seconds after which the respondents are expected to answer which of the sign types (pictorial or verbal) is more eye-catching and understandable. The results would be summarized as a percentage of properly recognized signs. RELIABILITY In addition to the above mentioned procedures to ensure reliability of the research findings, some other measures will be taken. Firstly, to minimize participant error and participant bias, the study will be conducted on afternoons5 and the respondents will be informed about what is required from them. Thereby only those people who are willing and able to participate in the study will be surveyed. Secondly, the ability of participants to recognize traffic signs properly will be tested twice6. Moreover, the research team will be provided with necessary instructions for conducting a study to avoid observer bias. Finally, data collection and data analysis processes will be highly structured to minimize observer error. VALIDITY

The main threats to validity will be eliminated through the following procedures. Being aware of participants mortality and maturation, the research is going to be cross-sectional. In other words, the study would not be stretched out, thus eliminating the effect of unpredictable factors that could appear over time and affect the willingness of drivers to participate in research. Concerning the external validity (also called generalization) the most suitable sampling technique was applied given the available information and time constraints in order to ensure that sample is as close in its characteristics to population as possible. VIABILITY (ACCESS AND ETHICS)
5

In the evenings, after work, tired drivers may have difficulty in comprehending the signs; meanwhile in the mornings,

drivers sign comprehension rate may be unusually high. Therefore a more neutral time is chosen.
6

First, in the questionnaires, drivers will be asked to mark the signs, which according to their opinions are

understandable to them. Afterwards, during the experiment, drivers will be shown the same signs and will be asked to tell their name or explain their meaning. This will be done to ensure the trustworthiness of the responses given in questionnaires.

10

Gaining access is a key to obtaining reliable and valid data. Therefore, several strategies will be implemented to gain physical and cognitive access. Firstly, an introductory speech, outlining the objective and methods of the study, time and data required as well as assurance of confidentiality, will be politely delivered using suitable language to the would-be respondents. In addition, we will point out the possible benefits, which the respondents might gain from being surveyed. For example, drivers might assess their level of sign comprehension by participating in the survey. The ethical issues that might affect the research were identified at the design stage of the research. Consequently, the conduct of the research will be guided by a set of principles such as guarantee of anonymity, maintaining objectivity and honesty, seeking for informed consent, etc.

RESULTS
11

OBJECTIVES 1, AND 2 - identify general factors that influence comprehension of traffic signs based on the studies of researchers from US, Europe, Asia and Arabic countries - compare these findings with those indicated by a sample from Tashkent (see Conclusion section) To identify whether particular driver factors affect comprehension of traffic signs or not, it was decided to analyze the correlation between already known features (monthly income, education, age, gender and marital status) and the number of correctly recognized signs. Concerning the comparison of these findings with the results of other researchers, it was done in Conclusion section of the report. Monthly income Monthly income Less than 100 USD 100-300 USD 300-500 USD More than 400 USD No answer Total Number of respondents 68 106 116 84 12 384
150 100 50 0 < 100 US D 100-300 300-500 USD (US D) > 500 US D NA Hitopgram : Num ber of res pondents c ategoriz ed by inc om e

One of the most commonly used ways to describe data is with a frequency distribution. From the histogram and box plot it can be seen that data is normally distributed. This is also supported by the exponent of skewness (See Appendix 4: Monthly income) which is very close to zero. This means

Box plot

there are almost equal number of people with high and low incomes in the sample. In normally distributed data median is the most representative average, as it is not distorted by extreme values (like mean) and very frequent cases (like mode). I our case, median is equal to 690 (close to mean = 665) with the income reaching its maximum level at 1200 USD and minimum at 130 USD. Turning to one of the most wide spread measures of variability standard deviation, it is equal to 315.97 USD in our case. This means nearly 68%
300-500 (USD) 30% 100-300 USD 27%

Number of respondents categorized by level of income (in USD)

>500 USD 22%

NA 3%

<100 USD 18%

12

of all prices within the concerned period fall within 1 standard deviation, 95% -2 standard deviations, 99% -3 standard deviations of the mean.

Turning to the identification of whether income affects sign recognition level, simple linear regression model was applied (the data used is quantifiable for both income and number of signs Scatter plot: Income vs. # of correctly recognized correctly). recognized signs
Trend line (Income vs. # of correctly recognized signs) 14 12 10 Estimated 8 Trend line 6 4 2 0 0 500 1000 1500 Income

# of signs recognized correctly

# of signs recognized correctly

Income

It is evident that income has a strong positive correlation with sign recognition level (R2 = 0.83, see Appendix 4: monthly income for detailed analysis). Thus, most drivers with income higher than 1000 USD comprehended correctly at least 7 signs, while for lower income group (up to 500 USD) the understandability level reaches its maximum only at 8-9 signs. Education Education level School certificate Bachelors degree Masters degree Doctoral degree No answer Total Number of respondents 28 268 71 13 4 384
Num be r o f drive rs categorized by le vel of educa tion 30 0 20 0 10 0 0
Sc hool Bac helors Mas ters c ertif ic ate degr ee degree Doc toral No ans w er degree

Level of education also affects the recognition of traffic signs (ANOVA Statistics: p-value = 0,0000158). Indeed, as it can be seen from descriptive statistics (See Appendix 4: Education), at least one of the drivers in all the groups, except those holding school certificate, recognized all 11 signs (6 max for school certificate holders). At the same time, people holding doctoral degree have much higher average (mean and median = 8 and mode =7) of correctly recognized signs, 13

which is at least 2 signs higher than for other groups. By contrast with highly educated drivers, those holding only school certificate, have average of only 2-3 signs, but standard deviation almost the same as for other groups. Age Age 18-25 years 26-35 years 36-45 years 46-55 years 56+ years No answer Total Number of respondents 81 97 88 86 29 3 384
Number of respondents categorized by age 120 100 80 60 40 20 0 18-25 years 26-35 years 36-45 years 46-55 years 56+ years

From the analysis undertaken (see Appendix 4: age), it becomes evident that with age has a direct effect on level of correctly recognized signs (p-value = 0.016). Form descriptive statistics (Appendix 4: age), it can be see that with age drivers are able to show better and better results. The only exception is the oldest group which has an average (mean) number of correctly recognized signs equal to 7, whereas for all the other groups an average gradually increases (from 6.86 for the youngest group to 8.93 for 46-55 years old group). At the same time, there is a very high variation in results for the younger group (3.4 standard deviation), whereas for older ones it never exceeds 2 answers7.

Gender Gender Male Female Total Number of respondents 198 384


gender

186

Signs Number of comprehended recognition level respondents Comparison of traffic signs correctly genders between 0-5 signs 158 6-11 signs 221 Females No answer 5 Total 384
Males

Code

1 2 0

In order to test whether gender affects traffic signs comprehension level, it was decided to perform Chi-test (See Appendix 4 - Gender).
7

50

100

150

number of times more than 6 or less than 6 signs were recognized correctly number of times less than 6 signs were recognized correctly number of times more than 6 signs were recognized correcly

This means 95 % of all the cases lie within 2 standard deviations from the mean, 99% - within 3 st deviations from the mean.

14

The result is that there is a relationship between gender and level of traffic sign comprehension (p=0.00011). This was evident from the results obtained. Thus, 134 males correctly identified more than 6 signs in comparison to 87 by females (though the difference in 2 gender samples in our case is very small). Moreover, less males showed less than average result (<6 signs) than females. Marital status

Marital Status Married Single Divorced No answer Total

Number of respondents 153 187 39 5 384

Signs comprehended correctly 0-5 signs 6-11 signs No answer Total

Number of respondents 158 221 5 384

Code

1 2 0

Number of people who recognized less than 6 signs correctly

It

Number of people who recognized more than 6 signs correctly

divorce d 8% m arrie d 46%

divorced 12% married 37%

single 46%

single 51%

was found (Appendix: Marital status) that marital status has no effect on level of signs recognized correctly. Given different samples, the proportion of people recognizing more or less than average is approximately equal in all 3 categories. Pictorial and verbal signs (data from experiment)
T h e re s p o n d e n ts ' o p in io n o n w h th e r p ic to ria l o r v e rb a l s ig n s a r b e tte r re c o g n iz e d

The already established hypothesis is that pictorial signs are better noticed than pictorial signs (given the distance 20-30 meters). The data concerning the number of both pictorial and verbal signs recognized correctly was collected. In order to find whether it is true or note,

Verbal s igns 37% P ic torial s igns 63%

15

one-tail and two-tail tests were performed. (see Appendix 4 : Pictorial and verbal signs). It was found that there is a difference between level of comprehension depending on the type of signs (two-tail test, p=0.0012) and at the same time pictorial signs are better comprehended than verbal (one-tai test, p=0,00060). Indeed, referring to descriptive statistics (Appendix: Pictorial and verbal signs), 8-9 pictorial ad only 7 verbal signs were noticed at average by the same respondents. Finally, the respondent themselves state that pictorial signs are better recognized than verbal ones (see the pie chart)

Shape, color and familiarity In the experiment 11 testing signs were randomly presented to the respondents. For each correctly interpreted sign subjects were asked to indicate the shape and color of the sign as well as to give ratings for familiarity. Then regression analysis was applied to examine the relationship between variables o interest. The data from the sample supported the initial hypothesis. To check whether the findings are applied to the whole population one-tail test was performed. The hypothesis was accepted with type 1 error of 5 percent (see Appendix 4: Shape, color and familiarity)

OBJECTIVE 3 AND 4 reveal differences in comprehension among several categories of drivers evaluate the effectiveness of traffic sign system in Tashkent

Comparison between two different groups of respondents In previously made researches it was found that old drivers with high level of income and level of education are able to recognize traffic signs better than young ones with low income and level of education. In order to do so, the respondents were divided into 2 groups and the number of correctly recognized signs by each person in each of the groups was recorded (see Appendix 5 for more information). The result (t-stat) showed that there is a difference in level of recognition between groups (p=0.02) and older drivers with high income and level of education comprehend signs better than young ones with low income and level of education (p=0.01) 16

Effectiveness of traffic sing system in Tashkent based on opinion of the drivers


Ealuation of Tashkent traffic sign system effectiveness (based on drivers' opinions)

100 80 60 # of respondents 40 20 0

very good

good

appropriate

bad

very bad

not applicable

Criteria

Most of the respondents consider traffic sign system of Tashkent an appropriate one. However, as it can be seen from the chart, bad and very bad responses are very close, whereas very good and good responses fall behind. Indeed, there are twofold more drivers who are not content with the current system than those who consider it good or very good.

CONCLUSION AND RECOMMENDATIONS The results of the study answered the main research question by meeting the research objectives in the following way. Firstly, based on the studies of researchers from US, Europe, Asian and Arabic countries, the main driver factors and design features which affect the comprehension of traffic signs were revealed. Al-Madani and Al-Janahi (2002) found that such driver factors as income, education and age have a considerable effect on comprehension of traffic signs, while marital status and gender have no or insignificant effect. The main sign design features, which have a strong correlation with sign understandability, are color and shape according to Gao et al., (2003) as well as familiarity as stated by Dewar et al. (1994), Ng and Chan (2008) and Ben-Bassat (2004). Secondly, these findings were compared to the factors indicated by the sample from Tashkent through testing the following hypotheses. 17

H-1: Monthly income, education, and age have a significant effect on comprehension of traffic signs, while marital status and gender have no or paltry effect. H-2: Pictorial sings are more eye-catching than verbal ones ceteris paribus. H-3: Familiarity, color and shape of traffic signs are strongly correlated with their understandability. H-4: Old drivers with high income and level of education recognize sings better than young ones with low income and level of education. The results of the analysis supported the second, third and the fourth hypotheses. Indeed, during the experiment it was found that pictorial sings are more eye-catching and hence better comprehended than verbal ones. In addition, the category of old drivers with high income and level of education demonstrated better comprehension of the signs than the category of young drivers with low level of income and education. Finally, familiarity, color and shape exhibited strong correlation with sign understandability. On the other hand, the first hypothesis was not accepted fully. As it was expected, monthly income, education, and age of the drivers have significant effect on their comprehension of traffic signs and marital status has no effect. Contrary to expectations, it was found that gender of the driver has a considerable impact on their comprehension of traffic signs. Thirdly, the traffic signs system was evaluated as relatively effective, because the majority of signs in Tashkent are pictorial. However, the survey results showed that the quality of many traffic signs, especially in Bektemir and Sergeli districts is very poor. Overall, the study proved that the success of effective communication of a traffic sign message to users does not only relate to the drivers characteristics but also to the signs themselves. The findings provide the following recommendations for increasing the effectiveness of the traffic system. (1) Traffic sign designers are recommended to construct the signs which comply as much as possible with the standards established in the research. Namely, different sign shapes should be used to distinguish among prohibitive, warning, and guidance signs and avoid confusion and misinterpretation. In addition, signs should be presented in the pictorial form and must be painted in the appropriate colors (e.g. red to indicate danger). (2) Based on the results of the research, particular driver categories that lacked understanding of traffic signs were identified. The related organizations might use the information to improve the efficiency of traffic education. For example, traffic education centers might arrange special 18

intensive classes for young drivers who are more likely to violate signs than their older counterparts. (3) Given that familiarity is highly correlated with sign understandability, municipality is recommended to introduce some form of informative advertisement, so that drivers dont forget the meaning of the signs especially those rarely used. The research contains some practical limitations that must be taken into account. Firstly, the traffic signs used in the study were presented in the absence of realistic context (e.g. pictures of potential backgrounds where the signs might be located). The respondents might perform better in comprehension test, if the signs were located in typical context. Therefore the comprehension performance of the respondents might be underestimated. Secondly, the data on personal characteristics were based solely on self-administered questionnaires. It is possible that some respondents embellished their answers without revealing their true characteristics. Thirdly, as questionnaire forms needed to be filled in comparatively short period of time, 1-2 % of respondents have failed to answer all the questions. Finally, the recommendations for designing user-friendly traffic signs given in the project might not be valid, since the conclusions are based on the comprehension only, which is not the only element in information processing model, which also includes attention, attitudes and beliefs, motivation and behaviour (Wogalter and Laughery, 1996, cited in Ng, 2007) Though some limitations, research can serve as a reliable basis for further investigations. Indeed, the sample from Tashkent was taken only, while it is required to cover all the regions of Uzbekistan to get all the information about the effectiveness of traffic sign system. Furthermore, the results of the research may be applied in whole Central Asia, as the factors affecting comprehension are very similar here. But what is more important, research discovered some differences in the comprehension of traffic signs by Asian people and those from the Middle East and Europe. So, deep cross-cultural investigation may be undertaken to test if this is really true and reveal the possible reason for existence of this differences as well as the reasons for particular category of drivers being able to comprehend traffic signs better than the others.

19

REFERENCES Al-Gadhi, S. A., Naqvi, S. A., Abdul-Jabbar, A. S., (1994). Driver Factors Affecting Traffic Sign. [online] Available from: <http://209.85.229.132/search? q=cache:EIY1zo6g5VsJ:faculty.ksu.edu.sa/AlGadhi/Documents/7_%2520Driver%2520Factors %2520Affecting%2520Traffic%2520Sign%2520Detection%2520and %2520Recall.pdf+driver+factors+affecting+traffic+sign+detection+and+recall. +Transportation+research+record&hl=ru&ct=clnk&cd=1&client=opera> [Accessed 20 February 2009]. Al-Madani, H. and Al-Janahi, A., (2002). Assessment of drivers comprehension of traffic signs based on their traffic, personal and social characteristics. [online] Bahrain:University of Bahrain. Available from: <http://www.inf.unideb.hu/~totht/modszerek/cikkek/psicho.pdf> 20

[Accessed 22 February 2009] Automobile transport in Tashkent, (2001). [online] Available from <http://goskompriroda.ccc.uz/Tashkent/russian/drivers/transport/descrip.htm> [02 April 2009] Ben-Bassat, T. and Shinar, D., (2003). Are ergonomically designed traffic signs more comprehensible?. [online] Israel: Ben-Gurion University of the Negev. Available from: < http://www.psychology.nottingham.ac.uk/IAAPdiv13/ICTTP2004papers2/Vision/BenBassat.pdf> [Accessed 23 February 2009] Borowsky, A. and Shinar, D., (2008). Sign Location, Sign Recognition, and Driver Expectancies. Transportation Research. November, 11, 6, 459-465. [online] Available from: EBSCO host. <http://web.ebscohost.com/ehost/detail?vid=1&hid=9&sid=cb961b7a-0dd4-4734-a6e9b763277e0f50%40sessionmgr2&bdata=JnNpdGU9ZWhvc3QtbGl2ZQ%3d %3d#db=aph&AN=34747243> [Accessed 22 February 2009]. Dewar, R. E., Kline, D. W., Swanson, H.A.., (1994). Age Differences in Comprehension of Traffic Sign Symbols. Transportation Research Board. 1456 [online] Available from: TRIS Online Record. <http://ntlsearch.bts.gov/tris/record/tris/00677585.html?view=printable> [Accessed 21 February 2009].

Gao, X., Podladchikova, L., Shaposhnikov, D., (2003) Application of vision models to traffic sign recognition. Artificial Neural Networks and Neural Information Processing. 2714/2003.[online] Heidelberg: Springer Berlin. Available from: SpringerLink. <http://www.springerlink.com/content/19layaw7kujf55bw/? p=7d4c003a9e844ceeacc4866e8bcbe10d&pi=0> [Accessed 19 February 2009]. GOPA-TRADEMCO, (2008). Annex 5 to report on legal issues responses to legal questionnaires. [online] Available from: http://www.centralasiatransport.com/content/ru/reports_legal_data/progress_reports/_legal_issues/ Annex_5_to_Report_on_legal_issues-Responses_to_legal_questionnaires.pdf> [Accessed 23 February 2009]

21

Kurniawan, S. and Zaphiris, P.,(2001). Investigating the age effects on subjective assessments of traffic signs.[online] Detroit: Wayne State University. Available from: <http://www.soe.ucsc.edu/~srikur/files/2001_iie.pdf> [Accessed 21 February 2009]. Ng, A. and A, Chan., (2008). The effects of driver factors and sign design features on the comprehensibility of traffic signs. Journal of Safety Research. June, 39 (1), 321-328. [online] Available from: EBSCO host. <http://web.ebscohost.com/ehost/detail? vid=1&hid=13&sid=0ae1b3fb-3d30-41cc-a84c986902ac5e29%40sessionmgr8&bdata=JnNpdGU9ZWhvc3QtbGl2ZQ%3d %3d#db=aph&AN=32749285> [Accessed 18 February 2009]. Raosoft sample size calculator, (2004) [online] Available from:
<http://www.raosoft.com/samplesize.html>

[Accessed 02 April 2009]

Regression analysis, (2005). [online] Available from: <http://www.msu.edu/~nurse/classes/summer2001/813/newregression3.htm> [Accessed 29 April 2009] Richards, S.H. and Heathington, K.W., (1988). Motorist understanding of railroad-highway grade crossing traffic control devices and associated traffic laws. Transportation research record. 1160. [online] Available from: TRIS online record. <http://ntlsearch.bts.gov/tris/record/tris/00478096.html> [Accessed 21 February 2009]. Shinar, D. et al.,(2003). Traffic sign symbol comprehension: a cross-cultural study. Ergonomics. 46 (15), 1549 1565. [online] Available from: EBSCO host. http://web.ebscohost.com/ehost/detail?vid=9&hid=13&sid=df05b5d0-cbc5-41c8-95297482a99c868c%40SRCSM1&bdata=JnNpdGU9ZWhvc3QtbGl2ZQ%3d %3d#db=aph&AN=11650448 [Accessed 21 February 2009]. akowska, L.,(2001). Perception and recognition of traffic signs in relation to drivers characteristics and ssafety- a case study in Poland.[online] Cracow: Cracow University of Technology. Available from: < http://www.ictct.org/workshops/01-Caserta/Zakowska.pdf> [Accessed 21 February 2009].

22

BIBLIOGRAPHY Analyzing, interpreting and reporting basic research results, (2009). [online] Available form: <http://managementhelp.org/research/analyze.htm> [Accessed 01 May 2009] Burkhardt, J.E., Berger, A.M., Creedon, M. and McGavock, A.T., (1998). Mobility and Independence: Changes and Challenges for Older Drivers. [online] Available from: < http://www.aoa.dhhs.gov/research/drivers.html> [Accessed 29 April 2009] Dix, A., (2004). Research and innovation techniques. [online] Available form: < http://www.comp.lancs.ac.uk/~dixa/topics/res-tech/ > [Accessed 01May 2009]

23

How to write a good research paper, (2008). [online] Available form: <http://unipapers.org/blog/2008/08/27/how-to-make-a-research-paper-2/ > [Accessed 24April 2009] Kline, T.J.B., Ghali, L.M. and Kline, D.W., (1990). Visibility Distance of Highway Signs among Young, Middle-Aged, and Older Observers: Icons are Better than Text," Human Factors, 32(5). Mugo, F.W., (2008). Sampling in research. [online] Available form: <http://www.socialresearchmethods.net/tutorial/Mugo/tutorial.htm > [Accessed 25pril 2009] Review of literature, (2009). [online] Available form: <http://ludwig.missouri.edu/405/review.html> [Accessed 04 April 2009] Saunders, M., Lewis, P. and Thornhill A., (2003). Research methods for business students. 3rd ed. Harlow: Pearson Education.

Appendix 1 SAMPLE SIZE CALCULATOR

24

Appendix 2 SURVEY
Guidelines for researcher on how to conduct a survey : Start a survey by introducing yourself and asking whether a respondent is willing and has time to complete the questionnaire. Make sure that respondent has license and experience of driving for at least three months. 25

Clearly state the purpose of the research and the type of the questions for respondents Indicate the time it will take to complete the questionnaire. Assure confidentiality of the data provided by the respondents. Express your gratefulness to participant. Leave your phone number, in case respondent would have questions corresponding to a group project.

Introductory speech Good afternoon! I am a level 5 economics student from WIUT and my name is . I am conducting a research on a topic: How do driver factors and design features affect comprehension of traffic signs in Tashkent?. Our study consists of two parts: survey and experiment. In the first part, you will be asked to answer several questions regarding your personal characteristics as well as your opinion on the importance of particular sign features. In the second part, your ability for proper sign recognition will be tested. It will take about 15-30 minutes. If you agree to participate, you will find out how good is you ability for traffic sign comprehension. In addition, when the research is finished, we will send you an electronic copy of the findings, which will describe what categories of drivers are good at traffic sign comprehension and which are not. In addition, the report will provide an evaluation of the effectiveness of traffic sign system in Tashkent. We guarantee you full confidentiality of the data provided.

Questionnaire For this part of the research you are asked to answer a few biographical questions, to express your opinion on the effectiveness of traffic signs and to interpret the meaning of particular signs. Please read the questions carefully. Please answer the questions by placing a tick mark "" over the circle next to the response that you have chosen. 26

You are free to skip any question that you do not wish to answer. Your responses during the interview are confidential. When you finish, please submit the paper to the researcher. Q.1 Please choose one age group out of five that you belong to. o 18-25 o 26-35 o 36-45 o 46-55 o 56 + Q.2 What is your gender? Tick the appropriate answer. o Male o Female Q.3 What is your nationality? __________________ Q.4 Please, specify your marital status for a moment. (tick the right answer) o Married o Unmarried o Divorced Q.5 Please choose one income category out of three that you belong to. o 100,000UZS < income per month < 300,000 o 300,000UZS < income per month < 500,000 o 500,000UZS < income per month < above Q.6 Select your level of education o high school certificate o undergraduate student o bachelors degree o master or PhD o other please specify 27

Q.7 How do you evaluate the following statement? Please tick one suitable answer. Verbal signs are less eye-catching and comprehensible, rather than pictorial signs that attract drivers attention on the roads. o o o o o Agree Strongly agree Disagree Strongly disagree Neutral

Q.8 Accept of the fact that traffic signs are intended to guide and regulate drivers on the roads, tick other options for the purpose of traffic signs. o o o o o o To control the flow of cars on the roads To inform drivers about road curves ahead To instruct and help to get to destination To make traffic system more complex To reduce risk and accidents on the roads To manage speed on the roads

Q.9 Please rank the following design features that facilitate traffic signs to be less likely violated. Number the list below in the order of preference starting from 1 to 5. Rectangular Round Red slash Symbol Text Familiarity

Q. 10 Which of the traffic sign colors are more eye-catching Number the list below in the order of preference starting from 1 to 7. Red Blue Yellow Orange White black on

Q. 11 How do you evaluate the quality of traffic signs in Tashkent? Very good 1 Good 2 Appropriate 3 Bad 4 Very bad 5 Not applicable 6

Q.12 Tick the signs which are familiar to you

28

Q. 13 What could you suggest in order to improve comprehension of those signs by drivers? ________________________________________________________________________________ ________________________________________________________________________________ ________________________________________________________________________________ ____________________________________________________________________ Q.14 If you want to receive a copy of our report please write your e-mail address __________________________________

When a pilot test will be conducted the following questions will be added to the questionnaire:

29

How much time did you spend completing the questionnaire? Was there anything unclear in instructions? Tick one o No o If yes than specify__________________

Is there a logical flow in the structure of questionnaire? o Yes o No

How do you estimate the layout of questionnaire? o Very good o Applicable o Not applicable

Which, if any, questions did you find unclear or embarrassing? Was the questionnaire easy to read and understand? Did you face any leading or complex questions?

Appendix 3:
EXPERIMENT
Instructions for experiment: Come prepared, rehearse the topic well in order to present credible in front of a participant. Check the quality of the paper and make sure that traffic signs are visible for the participant. Explain the aim of experiment and the process itself. Remind the rights of respondents (anonymity and privacy) 30

Keep in mind the structure of experiment, though use spontaneous and probing questions if needed. Ask only one question at a time and be sure that you record answers coherently; the use of appropriate recording system is suggested. It is important to keep silence and do not influence on answers of participants. Record all answers concisely Experiment

Dear participant, thank you for your agreement to participate in the second part of our research. First, I am going to show you 15 traffic signs. In response I would like to receive explanations to each sign and opinions about design features. NOTE: make sure that traffic signs are visible for all participants. 1. Please note which of the signs are familiar to you. Researcher ticks the signs which are familiar to respondent. Right turn ahead U-turn is prohibited Pedestrians crossing the road Winding road ahead Side road intersection ahead No right turn Circle intersection ahead Road ends, must turn right or left Two-way traffic Compulsory ahead or turn left Stop sign 2. Please explain the purpose of each sign, if you do not know try to guess. NOTE: show only one sign at a time How do you evaluate the quality of those signs? NOTE: list all possible answers Very good Right turn ahead U-turn is prohibited Pedestrians crossing the road Winding road ahead Side road intersection ahead 31 Good Not bad Poor Very poor Extremely Applicable bad

No right turn Circle intersection ahead Road ends, must turn right or left Two-way traffic Compulsory ahead or turn left Stop sign Please group the signs into three types: regulatory, warning, guide signs. Regulatory signs Warning signs Guide signs

Now we are moving to the next stage of our experiment. You will be presented a set of traffic signs, each of which has both pictorial and verbal representation. Please, indicate the type (pictorial or verbal) of each of the presented signs that is more eye-catching in your point of view.

Traffic signs
Pictorial representation 1 Verbal representation 1

3 32

This is the end of our traffic signs recognition test. We will receive you the results in a couple of weeks when all data is collected and analyzed. If some questions arise, please, dont hesitate to contact us; our team would be happy to provide all the necessary information. Thank you very much for your cooperation.

33

APPENDIX 4

Monthly income and traffic signs recognition level (REGRESSION ANALYSIS) Monthly income Less than 100 USD 100-300 USD 300-500 USD More than 400 USD No answer Total
Descriptive statistics (monthly income) Mean Standard Error Median Mode Standard Deviation Sample Variance 665 74,4752887 690 850 315,97189 99838,23529

Number of respondents 68 110 121 84 3 384

34

Kurtosis Skewness Range Minimum Maximum Sum Count

1,084119414 0,000239146 1080 130 1200 240030 381

12 respondents do not answer so we reject these 12 cases. We do not use weighted cases as the number of missed answers is very low compared to sample size, especially since many authors question the validity of using statistics to make inferences from your sample if u have weighted cases (Saunders et. al, 2003, p. 336

Regression Statistics Multiple R R squared Adjusted R Square Standard Error Observations ANOVA df Regression Residual Total 1 370 371 SS 128,3591692 25,41860853 153,7777778 Standard Error 0,708654575 0,000967482 Lower 95,0% -0,396511113 0,00664545 MS 128,3591692 1,588663033 F 80,79697617 Significance F 1,18503E-07 0,875666673 0,824564677 0,824374679 1,260421768 372

Intercept X Variable 1 Lower 95% -0,396511113 0,00664545

Coefficients 1,105769136 0,008696421 Upper 95% 2,608049386 0,010747391

t Stat 1,560378181 8,988713822 Upper 95,0% 2,608049386 0,010747391

P-value 0,138229879 1,18503E-07

The following model was obtained: Sign recognition level = 1,105769136 + 0,008696421*Income Sign comprehension level and income are strongly correlated. First of all, R squared is equal to 0,83. This figure shows that 83% of variation in Sign recognition level is explained by change in income and only 17% by other factors. As the figure is comparatively high for the equation, it also means that Residual Sum of squares is less for this equation, thus the prediction that could be made based on the model is reliable. 35

An independent variable (income) is statistically significant that is supported by almost zero pvalues (t-stat) (we reject null hypotheses that it is not statistically significant). The figure of 8.99 shows that sample slope is 8.99 standard errors larger than zero, and as p-value is almost 0, there is no chance that mere sampling can make a zero slope coefficient. The strong correlation is also supported by large F-statistic that is a proportion of explained variance to unexplained variance (errors). In all the cases we reject null-hypotheses, as independent variable has significant effect on the dependent variable (p-value in all the cases = 0 that that means
there are no samples that would randomly produce such a large F-value if the samples come from a population in which the true F-value is 0. (Regression analysis, 2005).

Education and traffic sign recognition level (ANOVA statistics) ANOVA statistics: Ho: Education level has no effect on traffic sign recognition H1: Education level has effect on traffic sign recognition We reject Ho as p = 0,0000158 which is less than our 0,05 confidence level. This is also supported by F-statistic value = 12,51 which is more than 4 times higher than its critical value.
Anova: Single Factor SUMMARY Groups School certificate Bachelor's degree Master's degree Doctoral degree

Count

Sum 18 97 66 37

Average 2,571429 6,0625 7,15 8,25

Variance 3,619048 6,329167 7,928571 3,583333

28 268 71
13

ANOVA

36

Source of Variation Between Groups Within Groups Total 298,1714

df 3 376 379

MS 54,42321 4,351671

F 12,50628

P-value 1,58E05

F crit 2,911335

Descriptive statistics 1. Number of signs recognized correctly by drivers holding school certificate
Column1 Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum 2,571428571 0,719031851 2 1 1,902379462 3,619047619 0,32867036 1,066531367 5 1 6

2. Number of signs recognized correctly by drivers holding bachelors degree


Column1 Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum 6,0625 0,628946 6 6 2,515784 6,329167 -0,37101 0,436369 9 2 11

3. Number of signs recognized correctly by drivers holding masters degree


Column1 Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis 6,25 0,490990253 7 6 1,38873015 1,928571429 1,10617284

37

Skewness Range Minimum Maximum

1,120128022 4 7 11

4. Number of signs recognized correctly by drivers holding doctoral degree


Column1 Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum 8,0625 0,628946 8 7 2,515784 6,329167 -0,37101 0,436369 5 6 11

Age and traffic signs recognition level

Ho: age does not affect comprehension of traffic signs H1: age affects level of traffic signs comprehension As p=0.016 is less than 0.05 and F=3.36 is more than critical 2.54 we reject Ho. Thus, age affects level of traffic signs comprehension.
Anova: Single Factor SUMMARY Groups 18-25 26-35 36-45 46-55 56+ Count 81 97 88 86 23 Su m 555 802 729 768 203 Average 6,857142857 8,266666667 8,285714286 8,928571429 7 Variance 1,978021978 2,20952381 2,681318681 3,917582418 3,333333333

ANOVA Source of Variation

df

MS

P-value

F crit

38

Between Groups Within Groups Total

4 376 380

9,137568306 2,72202381

3,356902417

0,015616419

2,536581

1. Number of signs recognized correctly by 18-25 age group


Column1 Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum 6,857143 0,375882 7 7 3,406422 3,978022 2,022222 0,485867 9 1 10

2. Number of signs recognized correctly by 26-35 age group


Column1 Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum 8,266666667 0,383798889 9 9 1,486446706 2,20952381 0,933984268 0,828886421 8 2 10

3. Number of signs recognized correctly by 36-45 age group


Column1 Mean Standard Error Median Mode Standard Deviation 8,285714286 0,437633137 9 9 1,637473261

39

Sample Variance Kurtosis Skewness Range Minimum Maximum

2,681318681 0,082538296 0,780891239 6 5 11

4. Number of signs recognized correctly by 46-55 age group


Column1 Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum 8,928571429 0,528987066 9,5 11 1,979288361 3,917582418 1,346336653 0,510192295 7 4 11

5. Number of signs recognized correctly by 56+ age group


Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum 7 0,912870929 7 7 1,825741858 3,333333333 -3,3 0 4 5 9

Gender and traffic signs recognition level In order to test whether gender affects traffic signs comprehension level, it was decided to perform Chi-test. We have already coded higher than average result as 2 (more than 6 signs comprehended correctly) and less than average result as 1 (less than 6 signs comprehended correctly). So, whether the driver has higher or lower than average result, was calculated manually and results were

40

recorded for each gender. As there no answers by 5 respondents concerning the sign recognition, the sample investigated is reduced from 384 to 379.

CODING Gender 0 no answer 1 male 2 female


Actual results Gender male female Total Expected results male female Total more than 6 signs 134 87 221 more than 6 signs 115,5 105,5 221 less than 6 signs 82,5 75,5 158 less than 6 signs 64 94 158 Total 198 181 379 Total 198 181 379

Ho: there is no association between gender and level of traffic signs comprehension H1: there is association between gender and level of traffic signs comprehension Running the chi-test, the following result was obtained: 0,00011. As it is less than the confidence level of 0.05 we reject Ho, thus stating that there is relationship between gender and level of traffic signs recognition.

Marital and traffic signs recognition level CODING Gender 0 no answer 1 married 2 signle 3- divorced

Actual results

41

Gender married single divorced Total Expected results married single divorced Total

more than 6 signs 81 113 27 221 more than 6 signs 89,21635884 109,0422164 22,7414248 221

less than 6 signs 72 74 12 158 less than 6 signs 63,78364116 77,95778364 16,2585752 158

Tota l 153 187 39 379 Tota l 153 187 39 379

Ho: there is no association between marital status and level of traffic signs comprehension H1: there is association between marital and level of traffic signs comprehension Running the chi-test, the following result was obtained: 0,130513. As it is more than the confidence level of 0.05 we accept Ho, thus stating that there is no relationship between marital status and level of traffic signs recognition.

Pictorial and verbal signs Categorical variables (signs) were tested with the help of t-test to find whether they are associated with number of correctly recognized signs (quantifiable variable).
t-Test: Two-Sample Assuming Equal Variances Pictorial 9,666666667 2,941176471 384 4,699346405 0 766 3,536615545 0,000597087 1,690923455 0,001194175 2,032243174 Verbal 7,111111111 6,45751634 384

Mean Variance Observations Pooled Variance Hypothesized Mean Difference df t Stat P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail t Critical two-tail

For two-tail test Ho: type of signs does not affect level of comprehension 42

H1: type of signs affects the level of comprehension As p-value is less than 0.05 for two-tail test we reject Ho. So, there is a difference between comprehension level depending on the type of sign

For one-tail test Ho: Pictorial signs are not better comprehended than verbal H1: Pictorial signs are better comprehended than verbal p-value is less than 0.05. So, we reject null hypothesis and state that pictorial signs are better comprehended than verbal.

Pictorial sings
Column1 Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum 8,666666667 0,404226042 8 10 1,714985851 2,941176471 1,87204 -1,46065345 6 5 11

Verbal signs
Column1 Mean Standard Error Median 7,111111 0,598958 7

43

Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum

7 2,541164 6,457516 -0,68048 0,102363 8 3 11

Shape, color, familiarity and traffic sign recognition level (Regression Analysis) H0: there is no correlation between red colored sign and correctly interpreted sign H1: red signs are strongly positively correlated with comprehension
Regression Statistics R Square Adjusted R Square Standard Error Observations

0,92 0,87 0,61 384 Coefficients 28,00 0,12 Standard Error 0,23 0,32 t Stat 14,50 23,00 P-value 0,0012 0,0008

Intercept X Variable 1

ANOVA demonstrates high positive correlation between red colored sign and correctly interpreted sign, which implies that red color makes signs more eye-catching and easier to comprehend. For a 5 percent significance level, both intercept and slope are statistically significant. Therefore, null hypothesis of no correlation can be rejected and the results might be generalized to the whole population. H0: there is no correlation between trianglre-shaped sign and correctly interpreted sign H1: triangle-shaped signs are strongly negatively correlated with comprehension
Regression

44

Statistics R Square Adjusted R Square Standard Error Observations

0,87 0,79 0,28 384 Coefficients 21,00 0,08 Standard Error 0,31 0,27 t Stat 17,00 21,00 P-value 0,0027 0,0012

Intercept X Variable 1

ANOVA demonstrates high negative correlation between triangle-shaped signs sign and correctly interpreted sign, which implies that triangle-shaped signs do not stand out and harder to comprehend. For a 5 percent significance level, both intercept and slope are statistically significant. Therefore, null hypothesis of no correlation can be rejected and the results might be generalized to the whole population. H0: there is no correlation between familiarity of the sign and correctly interpreted sign H1: familiarity of the sign is highly correlated with comprehension
Regression Statistics R Square Adjusted R Square Standard Error Observations

0,98 0,95 0,15 384 Coefficients 12,00 0,67 Standard Error 0,15 0,12 t Stat 35,00 29,00 P-value 0,0002 0,0000

Intercept X Variable 1

ANOVA demonstrates almost perfect positive correlation between familiar signs and correctly interpreted signs, which implies that the drivers know the true meaning of the majority of familiar signs. For a 5 percent significance level, both intercept and slope are statistically significant. Therefore, null hypothesis of no correlation can be rejected and the results might be generalized to the whole population.

Code sheet for colors Color red white on black blue green yellow Code sheet for shape Rectangular Triangle Circle Quadratic

Code 1 2 3 4 5

1 2 3 4

45

Code sheet for familiarity Familiar Not familiar

1 2

APPENDIX 5
Respondent were divided into 2 groups: 1 group: respondent 36+ years old with income of at least 500USD and masters or doctoral degree 2 group: respondent up to 36 years old with income less than 500 USD and school certificate or bachelors degree In order to make samples more or less equal, it was decided to use stratified random sampling in the first category (there were 4 times more respondents in the second group than in the first, so each 4 respondent was chosen in the second group for t-test)
t-Test: Two-Sample Assuming Equal Variances Variable 1 8,733333 4,638095 84 4,995238 0 166 2,450657 0,010383 1,70113 0,020765 Variable 2 6,733333 5,352381 84

Mean Variance Observations Pooled Variance Hypothesized Mean Difference df t Stat P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail

46

t Critical two-tail

2,048409

The following hypotheses were tested using t-test For two-tail test Ho: there is no difference in comprehension between 2 groups H1: there is difference in comprehension between 2 groups We reject Ho as p-value is 0.02<0.05. So, there is a difference between groups For one-tail test: Ho: Respondent from the first group do not recognize signs better H1: Respondent from the first group recognize signs better We reject Ho as 0.01<0.05. So, first group respondents recognize signs better than second group respondents.

47

You might also like