Real time observations were undertaken in order to ascertain the type of collision avoidance behaviour displayed by pedestrians in crowded and relatively empty areas. Results indicated that gender was the most significant variable in affecting the type of avoidance behaviour displayed by opposite-direction pedestrians sharing momentary close proximity. The larger percentage of female pedestrians displaying ‘closed pass’ behaviour and the tendency of males to display ‘open pass behaviour’ was significant (Chi Squared=308.396, df=1, p=0.01). The knowledge gained from this experiment confirms current notions of pedestrian avoidance behaviour and encourages the undertaking of further experiments on this topic, involving additional variables including age and direction.
It is recognised that collision avoidance between pedestrians occurs much earlier in relatively uncrowded spaces (Wolff, 1973; Collett & Marsh, 1981; Burgess, 1983). Goffman (1972) noted that pedestrians typically passed each other on the right (‘pedestrian streaming’ or ‘lane formation’). Furthermore, pedestrian avoidance behaviour was noted as being achieved by individuals moving to the right. This was potentially due to the fact that right-handedness is more typical in the population.
1.1 Oncoming pedestrian avoidance
Wolff (1973) noted a difference in oncoming pedestrian collision avoidance behaviour according to crowd density. A detour would be made if potential collisions were noted in time but at the highest densities wherein collision was inevitable, a “step-and-slide” movement was achieved. This involved “a slight angling of the body, a turning of the shoulder and an almost imperceptible side step” and was undertaken to avoid a guaranteed collision had either individual “walked directly ahead”. Collett & Marsh (1981) describe two “step-and-slide” possibilities, either a ‘closed pass’, where the pedestrian turns away from the oncoming person or the ‘open pass’ wherein the body is turned towards the oncomer. Typically, men would favour the ‘open pass’ whilst women tended to use the ‘closed pass’ mechanism.
1.2 Avoiding same direction collisions
Typically, pedestrians were found to position themselves behind the shoulder of another pedestrian. According to the speed of travel, ‘pedestrian clusters’ (pedestrian lanes) develop which grow proportionally with pedestrian numbers (Wolff, 1973). It is suggested that pedestrians believe lane formation aids in reducing collision likelihood. The following study will look at the strategies employed by males and females in order to avoid collisions whilst walking along a crowded street. The study was conducted in order to test the validity of previous experimental findings on this topic.
It is hypothesised that male pedestrians will use a closed pass, and female pedestrians an open pass. A clear difference between adult collision avoidance behaviour by gender is expected. The null hypothesis (H0) was that there will be no difference between the type of pass used by males and females. The proposed alternative hypothesis (H1) was that there would be a difference in the type of collision avoidance behaviour made in adult pedestrians, according to their gender.
2.0 Materials & Methods
The study involved 2279 adults, in a ratio of males: females (n=1161: n=1118 respectively). Further information, such as the age of the subjects was not recorded, as the experimental design was that of a naturalistic observational study. A between subjects design was chosen and three variables were recorded; the ‘closed pass’, the ‘open pass’ and the ‘neutral pass’. The design was considered appropriate as each of the subjects were involved in only one of the above three group parameter variables. The independent (or predictor) variable was defined as the gender of the subjects, whilst the criterion (or dependent variable) was marked as the type of passing behaviour observed in the subjects. The criterion variable was hypothesised to differ according to the gender of the subject. In the design, a ‘closed’- and ‘open pass’ were defined according to Collett & Marsh, (1981).
A data collection grid was used to note pedestrian collision avoidance behaviour (see Fig. 1, Appendix 1). Three types of behaviour variable were measured and two different locations were chosen for the study sites, a ‘high-density’ location and a ‘low-density’ location. At each location, the interaction of two pedestrians was recorded as either a ‘closed pass’, an ‘open pass’, or a ‘neutral pass’. The collision data was collected at one single 15 to 25 minute time period, in the evening of a single weekday, by different observers. Observers were stationed at either the ‘high-density’ or ‘low-density’ location and all the results were subsequently pooled according to the subjects’ gender. When passes were observed, the specific nature of the pass was recorded in the data collection table (see Fig. 1, Appendix 1).
A Chi-square test was performed on the data from ‘open’- and ‘closed pass’ observations. A significant association between, ‘closed pass’, ‘open pass’, and gender was observed (Chi-Squared =308.396, df=1) (see Fig. 2). A total of 40 instances of collision avoidance behaviour, out of ‘closed’- ‘open’- and ‘neutral passes’, across both genders was recorded.
Fig. 2: The frequency of open and closed passes according to gender in the display of pedestrian collision avoidance behaviour
A total of 420 females displayed ‘open pass’ collision avoidance behaviour, whilst 741 female pedestrians displayed ‘closed pass’ behaviour. Conversely, approximately 840 males utilised ‘open pass’ movement to avoid collision, 278 males engaged in ‘closed pass’ movements to avoid collision. As shown in Fig. 2 (above), female pedestrians utilised the ‘closed pass’ most frequently whilst male pedestrians favoured ‘open pass’ motion. More than double the number of male subjects utilised the ‘open pass’, only one third more female subjects used a ‘closed pass’ compared to the ‘open pass’. Indicating gender is a less significant variable in females than males, in dictating the most likely type of pass used for collision avoidance. The results were furthermore significant (Chi-Squared =308.396, p=0.01), showing that there is a statistical significant association between gender and collision avoidance displayed.
Evening data collection was judged the best time to collect data as it coincided with a commuting time. One recording period was undertaken at each location, lasting between 15 and 25 minutes each. This time slot was considered long enough to provide a good representation of collisions, but was limited by the observers’ time limitations.
The locations were decided as being suitable to represent ‘high’- and ‘low-density’ for collision recording, as the observers were familiar with these areas and with evening-typical pedestrian activity. Data was not collected on weekends due to time limitations. Although separate, gender demarcated readings were taken for both the high- and low-density locations. It was believed that pooling the data would be the most effective means of testing the hypothesis. To this end therefore, only the collision data that was recorded for the ‘open pass’ and ‘closed pass’ behaviours was utilised in the analysis. Data collected from subjects employing a ‘neutral pass’ technique was not considered necessary for the scope of this particular investigation.
A difference in the collision avoidance behaviour according to gender was noted. Thus, the null hypothesis, H0, is to be rejected and the alternative hypothesis has been supported by the findings. The results also agree with Collett & Marsh, (1981) and van Basten, et al., (2009), who found that when using a Pairwise comparison to study the avoidance behaviour of males and females in simulations and games, females would be more likely to collaborate for collision avoidance than males.
As Fig. 2 (above), shows, females tended to use ‘open passes’ and males ‘closed passes’ as predominant collision avoidance behaviour. It is believed that women intend psychologically to protect their breasts, by using a ‘closed pass’ (Collett and Marsh, 1981). The finding that females exhibited self-protective conduct is supported by Kobayshi (1994), who noted that on trains, females would be more likely to stand facing the door and position their arms at a lower angle, whilst males would be more likely to face other passengers and the reverse being true.
In a study by Jenni and Jenni (1976), American college students displayed a marked difference in the manner in which they held books; females more frequently carried books closer to the chest and males to the side. This posture-related behaviour may have relevance to support the findings in this study. The psychological reason that it is believed women wish to act with propriety (Collett & Marsh, 1981) again is similar to the suggested reason for their posture on public trains.
4.1 Study strengths & Limitations
As the data collection for this experiment was carried out over the course of only one day, at both locations, future same-location data collection, over three consecutive days, which further encompassed the same allotted time periods throughout the day may improve experimental validity. Three days of data recording at one location (e.g. high density region), at the previously decided time slots, for thirty minutes each time, would be sufficient to effectively map collision avoidance behaviour. The same measurement criteria as used at the first location should be applied to all other study locations, to ensure reliability. An average of the number of days the study was conducted, according to each time interval would then also preferable to undertaken, thereby increasingly the overall reliability of the results. However, this would only give an indication of the type of collision avoidance displayed at those particular locations. Other independent variables may also affect the results obtained, such as weather. However, as the study was a naturalistic study thereby providing a large amount of ecological validity, these concerns may be somewhat corrected for.
4.2 Future Research
To further increase the reliability of the data collected for repeat experiments, different degrees of ‘high density’ and ‘low density’ locations could be visited, to provide an indication of whether collision avoidance behaviour would differ dramatically, (for example), between the most densely populated regions and second most densely populated regions visited. The improved experimental design could include visiting five places across five days (and over three or more time periods throughout the day), according to “five-point scale” (e.g. 1=most densely populated, 5=least densely populated). It may also be useful to include an observation on pedestrian age as collision avoidance behaviour may be age dependent.
Future experiments may also consider same-direction pedestrian collision avoidance behaviour, to test whether the variable of direction affects avoidance behaviour. Instances of actual collisions could also be measured, to see whether, in these cases, avoidance behaviour was not displayed. Experimental subjectivity may have been a consideration in this study as more than one observer collected the tabulated data. This could readily be corrected for by having only one observer conduct all observations.
The results from this experiment lent support to the fact that collision avoidance behaviour definitely has a gender-associated factor (Collett & Marsh, 1981). Furthermore, in larger spaces, with fewer pedestrians the incidence of collision was found to be less likely, as it is believed that collisions were noted and corrected for much earlier. There are a number of topics within pedestrian collision behaviour that could be studied to provide further explanation in the area of gender associated collision avoidance behaviour. Studying same-direction collision behaviour, age-associated behaviour and recording behaviour at different time periods during the day may provide a deeper study of the topic as well as potentially useful ongoing research in the field.
Burgess, J.W. 1983. Interpersonal spacing behavior between surrounding nearest neighbors reflects both familiarity and environmental density. Ethology and sociobiology. 4(1), pp.1117.
Collett, P & Marsh, P. (1981). Patterns of Public Behavior: Collision Avoidance on a Pedestrian Crossing. In T. A. Sebeok & J. Umiker-Sebeok (Eds.), Nonverbal Communication, Interaction and Gestures, pp. 199-217. The Hague: Mouton Press.
Goffman, E. 1972. Relations in Public. Harmondsworth:Pelican.
Jenni, D.A. & Jenni, M.A. 1976. Carrying Behaviour in Humans: Analysis of Sex Differences. Science. 194, pp.859-860.
Kobayashi, T. 1994. Sexual Differences in Posture-related Human Behaviour on Subway Trains, and Their Biological Function. Journal of Ethology. 12(2), pp. 121-130.
Van Basten, B.J.H., Jansen, S.E.M. & Karamouzas, I. 2009. Exploiting Motion Capture to Enhance Avoidance Behaviour in Games. Lecture Notes in Computer Science. 5884, pp. 29-40.
Wolff, M. (1973). Notes on the behavior of pedestrians, in: People in places: the sociology of the familiar. Birenbaum, A.(ed), London, UK: Nelson Press, pp. 3548.