FEMALE AND MALE DRIVING BEHAVIOUR ON SWEDISH URBAN ROADS AND STREETS
Aronsson, Karin F. M., Ph.D. Candidate
Bang, Karl L., Professor em. Department of Transport and Economics
Royal Institute of Technology (KTH), Sweden
ABSTRACT Knowledge of male and female driver behaviour is essential when addressing gender issues in the planning and design of highways and streets. The objective of the study was to establish if there are significant differences in driver behaviour of male and female drivers for a variety of Swedish urban street designs, environments and traffic conditions. The comparative analysis was based on additional data extraction from an existing dataset. Behaviour data of male and female drivers was analysed through descriptive statistics, multiple regression and comparison of speed profiles. The results showed no significant difference between the mean free-flow speed of male and female drivers. For a number of studied sites there was an indication of slightly higher female free-flow speeds. Men and women had the same probability to be platoon leaders on arterials and urban streets. On suburban streets there was an indication that the likelihood for men to lead a platoon was slightly higher, although the data supporting this was very limited. On average male and female platoon leaders maintained the same speed (i.e. free-flow speed). On average male and female drivers maintained the same time headway to the nearest vehicle in front when driving in a platoon. On one investigated suburban street female drivers kept 0.2 seconds greater headway than male drivers (significant on 90 % level). In the driving simulator study, performed for urban street conditions, men had significantly higher free-flow speed than women (2 km/h). For the event “passing an occupied bus stop” men drove 4 km/h faster than women on average. No difference in behaviour between men and women was recorded for the event of “arriving at a crosswalk with approaching pedestrians ”. The conclusion of the field study was that Swedish male and female drivers differed only marginally in their speed and headway driving behaviour. The simulator study, which enabled “controlled experiments” for different traffic events, indicated some differences for specific traffic situations which should be further explored.
1.1 Background Previous studies indicate a difference in male and female driver behaviour (Ericsson 2000 and Polk 2005). Better knowledge of these matters is essential when addressing gender issues in the planning and design of highways and streets. The objective of the present study was firstly to investigate if there were significant differences in female and male driver behaviour for a variety of street designs, environments and traffic conditions. Secondly, the factors causing the differences should be investigated. Knowledge of male and female behaviour in traffic was obtained through additional extraction from data collected by the authors in a previous project (Aronsson and Bang 2005). The present study was supported by the Swedish Road Administration (SRA). Driving speed is influenced by many factors including street type, environment, traffic flow, and geometric design (Aronsson and Bang 2005). In built-up environment of urban or suburban character, motorized traffic is often mixed with unprotected road users with serious implications for traffic safety. For such facilities the driving speed is also influenced by activities and conflicts with these road users. Aronsson and Bang (2005) studied in-depth using a “bottom-up” approach with main emphasis on improving the knowledge of factors that influence a driver’s choice of speed and speed profile when travelling on urban streets with minor intersections for a variety of street types, street designs, environments and traffic conditions in Sweden.
1.2 Literature Review The attitudes of women and their driver behaviour are increasingly similar to that of men (Forward, Linderholm, and Järmark 1998; Polk 2005; Stutts, Rodgman, Reinfurt, and Staplin 2001). A qualitative study of driver irritation and aggressive behaviour (Björklund 2005) proved that driving very close to the rear of the subject’s car was causing most irritation. Women experience more often stress by this event than men study. Young drivers are more likely than older to be involved in rear-end crashes, and young males more so than young females (Singh 2003). A study of speed patterns (Ericsson 2000) showed higher average speed for male than for female subjects. An American study of spot speed and a regression model of speed proved that male mean speed was lower than female (Poe, Tarris, and Mason 1996). Furthermore it was shown that road design factors best explained the speed variation. Driver variables, such as age and gender, merely explained a few percent of the speed variation. In conclusion, three groups of variables were found in the literature to be of interest to examine further when analyzing gender issues and driving speed: Average free-flow speed choice Speed profiles Driver behaviour in restrained traffic flow
2.1 Introduction A previous project (Aronsson and Bang 2005) studied factors influencing speed on urban streets for three main types of facilities; arterials, suburban and urban streets links. Streets segments of each type were selected in six cities: Stockholm (population: 1.8 million); Uppsala (0.2 million); Linkoping (0.15 million); Vasteras (0.1 million) and Nykoping (0.05 million). The average daily traffic flow on the selected streets ranged from 5 000 to 20 000 vehicles. The data collected was of great use in the study of male and female driver behaviour described in this paper, since individual driver behaviour was investigated in great detail. Several data collection techniques were employed as below.
i) Mobile data collection ii) Short-base and long-base data collection iii) Area wide video data collection iv) Driving simulator studies v) Microscopic simulation
In the mobile data collection, a specially equipped passenger car travelled on nine urban roads logging speed, time, travelled distance, distance to vehicle in front of them, as well as recording on video camera any side-friction events. The short-base stations were equipped with double pneumatic tubes, video camcorders and manual observers. The latter were placed out of sight of passing cars. Consequently, the data gathered at eighteen short base stations consisted of traffic flow, spot speed, time headway, vehicles identity and passage time. Matching of vehicles identity and passage time data from entry and exit short-base stations provided results in form of average travel times and ratio of through traffic. The technique of area wide video data collection consisted of a video tower equipped with two remote controlled digital video cameras. The tower could cover a street segment of around 300 meter. Semi-automatic video analysis was used to track the movements of all vehicles and unprotected road users with the purpose to obtain driver behaviour data for speed profiles and speed impact of side friction events. The field studies described above were supplemented with driving simulator studies, investigating impact of side-friction events on driver behaviour of 21 male and 18 female subjects. Speed profiles and average travel times were analyzed and isolated impacts of different side friction events were quantified. Lastly, the gained speed knowledge was used to calibrate driver behaviour parameters in a microscopic simulation model. The model was applied to develop speed profiles and speed-flow relationships for one road type. Final speed prediction models are recommended on a synthesis of field data, driving simulator and simulation results (Aronsson and Bang 2006).
The field data collection was executed under the time period of June 2002 to May 2003. All measurements were made under fair weather conditions and normal traffic performance of the studied street. The driving simulator experiments were conducted in October 2004. For the purpose of the present study, additional data regarding driver gender was extracted that enabled the analysis of the male and female driver behaviour.
2.2 Investigated Traffic Measures Based on the reviewed literature several traffic measures were found to be of interest when addressing gender issues in traffic planning. The measures included average speed choice, speed profiles and driver behaviour in restrained traffic flow. To be more specific, the following parameters were measured and evaluated for a male and female driver population respectively. Short-base studies
1. Free-flow speed (male/female) 2. Speed of platoon leaders (male/female) 3. Percentage male/female drivers leading a platoon 4. Percentage male/female drivers being restrained by a vehicle in front
of them 5. Headway of restrained drivers (male/female) 6. Time distance to the vehicle behind, restrained traffic (male/female) 7. Speed of restrained drivers (male/female)
Long-base studies 8. Average travel speed on long-base routes (male/female)
Driving simulator studies
9. Average free-flow (male/female) 10. Speed profiles (male/female) 11. Multiple regression analysis of impact of various factors and side-
friction events on male and female average travel speed.
2.3 Short Base Study Each vehicle passing the short-base stations was registered by three actions in the previous project (Aronsson and Bang 2005). Firstly, when the vehicle crossed the pneumatic tubes information of passage time, vehicle type, axle spacing, speed and headway were registered. Secondly, a human observer registered manually the vehicles licence plate numbers and time interval. Third, the vehicle identity was backed up by video camcorder registering vehicle front view and passage time. For the purpose of the present study, information of driver gender was extracted from reviewing the video films (Figure 1) and entered into the traffic data log file for the specific site. A total of nine short base sites out of eighteen field sites were chosen for the gender data collection.
Figure 1 Example of video recording from a short-base station; Male driver at site Repslagare Street.
2.4 Long Base Study Vehicle identity and passage time were collected at two consecutive short base stations constituting a long base route. Travel times of the through vehicles were thereafter computed. Average travel speed of individual vehicles driving on the long base route was calculated from route distance and the vehicle’s travel time. The gender of the driver, in the case of through traffic, was extracted from the data from one long base route. Average male and female travel speeds were calculated and compared.
2.5 Driving Simulator Study Speed profiles and travel data for forty drivers were also collected using a driving simulator where the drivers encountered numerous sets of events when driving a street modelled after one of the actual streets surveyed in the mobile study. Free-flow traffic conditions in the own direction of travel, was studied on a street with on-coming traffic. Thirty-nine licensed drivers (21 men, 18 women, age ranging from 22 to 65 years) with varying driving experience performed the driving task in the simulator prepared with seven scenarios. Subjects were mainly recruited from university faculty and staff not otherwise engaged in the project. Participation was rewarded with a cinema voucher. Each subject encountered the same fixed sequence of scenarios
including routes with a varying frequency of on coming traffic, traffic on cross streets, pedestrians and buses at bus stops. The experiments were conducted in a laboratory room with a mock-up car and wall projection of the simulated drive. The mock-up car was equipped with a steering wheel, turn signal, adjustable car seat and pedals for breaking and accelerating. The simulator vehicle had automatic transmission. A relatively high quality sound system reproduced engine noise and vibrations, tire screeching and crash sounds. The street environment in the driving simulator was produced from drawings and videotapes collected during previous steps of the data collection. The traffic flow and the frequency of events at the time the field site data was collected were important inputs when the synthetic traffic environment was designed. The purpose of the driving simulator experiment was to obtain extensive data on driving patterns. This was achieved a) through the inclusion of many different types of drivers in the experiment, b) through the study of variables in the street environment that were not included in the field site data collection, c) by studying traffic situations that did not occur during the field site data collection and d) by evaluating the effects of interacting factors on driving patterns.
2.6 Analysis Methodology Behaviour data, collected for the present study, was stratified for male and female driver populations during various traffic conditions. Of importance when analyzing the speed data of individuals was to note if the driver was restrained by a vehicle in front of them or not. The speed choice of unrestrained drivers was considered to best represent speed choice made with regards to street design, environment and individual driving style preferences. Descriptive statistical analysis was applied on the data collected from the short and long base stations. The data was tested for differences in driver behaviour between men and women. The speed profiles of male and female subjects produced in the simulator study were examined and multiple regression analysis was performed to single out variables affecting male and female average travel speed.
3. RESULTS
3.1 Short base Study
Calculated traffic measures The data of individual vehicles passing the selected short base stations was analysed for each site and for time periods of interest. Comparisons between
traffic data for male and female drivers were performed for the driver behaviour parameters listed below. Average free-flow speed, male and female driver. The measure represents travel unrestrained from vehicles ahead in the driven direction. Vehicles with headway of 5 seconds or more were selected for the analysis. Percentage of male/female drivers leading a platoon. The measures indicated if there was a male or female predominance to drive as platoon leaders. The share of male drivers leading a platoon was calculated and compared to the share of female drivers leading a platoon. Drivers who were free from a vehicle ahead, but were closely followed by a vehicle behind, were studied. Headway greater than 3 seconds and time distance to the vehicle behind of 2 seconds or less defined a platoon leader in this study. Average speed of platoon leaders (male/female). The parameter represents the speed choice of a free driver who is closely followed by one or more vehicles. Headway greater than 3 seconds and a time distance to the vehicle behind of 2 seconds or less defined a platoon leader in this study. Percentage of male/female drivers being restrained by a vehicle ahead. The measure explains if there was a male or female predominance to be hindered by vehicles in front of them and to drive in a queue. Headways less than 3 seconds and a time distance to the vehicle behind of 2 seconds or less defined restrained travel. Average headway of restrained drivers (male/female). This measure represents drivers driving in restrained flow. A headway less than 2.5 seconds defined restrained traffic in this study. Average time distance to the vehicle behind, restrained traffic (male/female). This measure represents drivers driving in restrained flow. A headway under 2.5 seconds defined restrained traffic in this study. Average speed of restrained drivers (male/female). This measure represents drivers driving in restrained flow. A headway less than 2 seconds defined restrained traffic in this study.
Vehicle Type Selection Initial speed analysis from the sites led to the decision to exclude heavy vehicles from the study. On several sites, the heavy vehicles were in many cases driven by men at speeds and headways that notably differed from average cars and vans. A comparison of men and women driving cars, or vans, was judged to best represent gender driver behaviour.
The results of the calculated traffic measures were presented in tables of which examples are given below. The measures average speed, headway and time distance were tested to find out if there were any significant differences between male and female driving behaviour. Arithmetic means of the measures were computed and presented as the average values of the tables. A weighted average figure was computed per road type with the purpose to summarize the findings. The analysis showed that men and women drove with similar average speed on the studied streets during free-flow traffic conditions (Table 1). The average speed was on the studied street types; Arterials 48 km/h; Suburban streets 45 km/h and Urban streets 37 km/h. On arterials and urban streets, women drove in average over 0.5 km/h faster than men. It was significant on the 95 % level for one of the arterial sites. At the studied suburban sites, men drove on average 0.5 km/h faster than women did. The difference in share of men driving as a platoon leader, compared to women was small (Table 2). On the suburban sites, male representation was slightly higher. At other sites, the percentage of male drivers was equal to the percentage of female drivers. Male platoon leaders drove in average 0.5 km/h faster than female platoon leaders on arterials and during off-peak periods (Table 3). The difference was significant for one of the studied arterial sites. On suburban and urban streets, the difference in average platoon leader speed was insignificant. The hypothesis that women are overrepresented among restrained drivers was slightly supported by the calculation of off-peak travel on the arterial road sites (Table 4). A control of average headway for restrained drivers was performed for vehicles driving in a queue or simply restrained by a vehicle ahead. The difference in average headway for men and women was null for arterials and urban streets. On the suburban sites, women drove on average with a 0.2 seconds greater headway than men did on average, which was significant for one of the suburban sites at the 90% level. Average time distance to the vehicle behind was checked for men and women driving in restrained traffic. Men and women drove with similar time distance to the vehicle following them on most road sites, except on one arterial road on which for women the distance to the car following them was somewhat greater (significant on the 90% level).
3.2 Long base Study Average travel speeds of the through vehicles on ten studied streets were investigated in the previous project by Aronsson and Bang (2005). Average travel speed was calculated from the long base distance divided by the average travel time of a through vehicle. Information of the gender of the drivers of the through vehicles was extracted for one of the routes for the objectives of the study and entered into the database. The selected route was a suburban street and the time period was an afternoon hour. The design of the street was two lane, two way undivided traffic, and narrow sidewalks for pedestrians. Posted speed was 50 km/h. The traffic flow was approximately 900 vehicles per hour for both directions on the day of field study. The average through vehicle drove the route with a travel speed of 39.7 ± 1.4 km/h. The average travel speed of the fifty-tree vehicles driven by men was 39.4 ± 1.7 km/h. Fourteen vehicles were driven by women with an average speed of 40.0 ± 3.9 km/h. Seven drivers of through vehicles could not be identified for gender and their average travel speed on the routes was 42.1 ± 4.4 km/h. The speed distribution of female drivers was compared to the whole population. Through vehicles driven by women follow a speed distribution similar to the whole through vehicle population. The average speed of men and women was tested for equality of means. The result proved difference in average travel speed of men and women, driving a through vehicle along the specific route, was not significant on the 95 % level.
3.3 Driving Simulator Study
Speed profiles at side friction events Speed profiles of forty drivers were collected in the driving simulator study. The drivers encountered many types of side friction events when driving in the simulator. Two of the tested side friction events were proved to reduce driver speed:
a) Pedestrians crossing the street b) Buses stopped at a bus stop.
The effect of pedestrians crossing the street on driver speed was also studied in previous field study (Aronsson and Bang 2005), which made it possible to make a comparison of speed profiles collected with the two techniques. Both studies were of unrestrained travel, i.e. free-flow travel. The collected speed profiles showed firstly good resemblance between simulated and observed behaviour for the event of a pedestrian approach a crosswalk, and secondly a great similarity between male and female speed selection. Next, the driver behaviour at the event “buses at bus stop” was investigated using the driving simulator data (Figure 2). The lane width of the street in this scenario was 3.5 meters. The driven speed past the bus stop area was notably higher for male drivers than female.
Regression Analysis of Average Travel Speed Average travel speed data from the simulator study was used as a basis for multiple regression analysis of the impact of different types of events producing the following significant results (t < 0.05). The R square value of the analysis was 0.48. The simulator study demonstrated a considerable impact on the free-flow speed of crossing pedestrian (-19.0 km/h), and buses stopped at the side of the street (-14.0 km/h). The factor of pedestrians walking on the sidewalk, right hand side, decreases the free-flow speed by less than 2 km/h. On-coming traffic in the simulator increased the free-flow speed by 2.4 km/h. Gender was an influential factor on the speed driven in the simulator and the analysis proved in general that men drove 2.3 km/h faster on this street type. Regression analysis for separated male and female driver populations verified gender differences of specific side friction events. The average free-flow speed of male drivers was reduced by 12.3 km/h when passing an occupied bus stop. The equivalent value of female drivers was 16.0 km/h. This implies women reduced their vehicle speed to a greater extent than men did, when passing a bus at bus stop. The difference in drop of average speed was 4.3 km/h, which the speed profiles also demonstrated. The lane width of the street was 3.5 meters in some scenarios and 4.0 meters in others. The buses at the bus stop were placed on a distance of 3.9 and 4.1 meters from the centre line in both cases of street width.
4. CONCLUSIONS Field survey results Free-flow speed The results showed detected no significant difference between the mean free-flow speed of male and female drivers. For a number of studied sites there was an indication of slightly higher female free-flow speeds. Platoon leaders Men and women had the same probability to be platoon leaders on arterials and urban streets. On suburban streets there was an indication that the likelihood for men to lead a platoon was slightly higher, although the data supporting this was very limited. On average male and female platoon leaders maintained the same speed (i.e. free-flow speed). Headway during restrained driving On average male and female drivers maintained the same time headway to the vehicle in front when driving in a platoon. On one investigated suburban street female drivers kept 0.2 seconds greater headway than male drivers (significant on 90 % level).
Simulator study results In the driving simulator study, which was performed for urban street conditions men had significantly higher free-flow speed than women (2 km/h). For the event “passing an occupied bus stop” men drove 4,3 km/h faster than women on average. No difference in behaviour between men and women was recorded for the event of “arriving at a crosswalk with approaching pedestrians ”. Synthesis The conclusion of the field study was that Swedish male and female drivers differed only marginally in their speed and headway driving behaviour. The simulator study, which enabled “controlled experiments” for different traffic events indicated some differences for specific traffic situations which should be further explored. REFERENCES Aronsson, K and K L Bang (2005) "Factors Influencing Speed Profiles on Urban Streets" in TRB 3rd International Symposium on Highway Geometric Design. Chicago, USA. Aronsson, K and K L Bang (2006) “Influence on vehicle speed profiles of interactions with other road users” in European Transport Conference 2006 conference, Strasbourg, France. Björklund, G (2005) Driver Interaction; Informal Rules, Irritation and Aggressive Behaviour. Uppsala University, Sweden. Ericsson, E (2000) Urban driving patterns - characterisation, variability and environmental implications. Bulletin 186 Lunds University, Lunds Institute of Technology, Sweden. Forward, S, I Linderholm, and S Järmark (1998) "Women and Traffic Accidents, Causes, Consequences and Considerations" in the 24th International Congress of Applied Psychology. San Francisco, California, USA. Poe, C M, J P Tarris, and J M Mason, Jr. (1996) "Relationship of Operating Speeds to Roadway Geometric Design Speeds" Vol. FWHA-RD-96-024: Federal Highway Administration, USA. Polk, M (2005) "Women's and men's valuations of road system infrastructure in Sweden". Borlange: Swedish Road Administration, Sweden.
Singh, S (2003) "Driver Attributes and Rear-end Crash Involvement Propensity". Washington DC, USA: National Center for Statistics and Analysis, National Highway Traffic Safety Administration. Stutts, J, E Rodgman, D Reinfurt, and L Staplin (2001) "The Role of Driver Distraction in Traffic Crashes". Washington D.C.: AAA Foundation for Traffic Safety, USA.
