Behavioural Impacts of Rewards for Avoiding Peak-hour Driving: Analysis of the Dutch Spitsmijden Project.
E Ben-Elia, D Ettema, Utrecht University, NL
Behavioural impacts of rewards for avoiding peak-hour driving during the course of the Dutch 'Spitsmijden' project. Longitudinal analysis and mixed model estimation results highlight the potential benefits and problems for congestion management.
The level of congestion on the Dutch road network is already quite high and predicted to increase in the next decade. Road pricing is stated by transport economists as the first-best solution to alleviate congestion externalities. However, public opinion in the Netherlands does not support more taxation. Consequently, rewards for avoiding peak- hour driving have been suggested as an alternative congestion management strategy. This approach is supported by psychological knowledge which suggests that rewarding is more efficient in the long run in sustaining wanted behaviour compared to punishments which only thwart unwanted behaviour but do not promote learning and internalization.
The Dutch 'Spitsmijden' experiment was conducted by a public-private partnership. Its purpose was to collect a large sample of field data (RP) regarding the impact of rewards on daily commuting behavior during the morning rush-hour. During a period of 13 consecutive weeks in Autumn, 2006, 340 recruited volunteers from Zoetermeer, a satelite city of The Hague, participated in a scheme whereby they would receive daily rewards, either of money (N=232) or of credits to earn a 'Yeti' smartphone(N=108). Participants could avoid peak hour travel either by shifting their departure times (earlier or later) or choosing other travel modes (bike or public transport) or by working from home. Yeti users were also provided with real-time traffic information and travel times.
Data was collected in three stages. Upon recruitment, participants filled a web-based survey about their home to work travel routines and socio demographic characteristics. Detection equipment was installed on the outskirts of the town and web-based personal travel log book recorder was applied in the second stage. In the first 2 weeks behavior was tracked but no rewards were given. The reward period lasted 10 weeks. Different reward schemes were assigned in different orders depending on the reward type. Participants receiving money took part in 3 consecutive reward treatments (3?, 7?, 3-7?). They received their rewards at the end of each week. Yeti users participated in 2 consecutive treatments: accredited weeks and unaccredited weeks. At the end of the reward period Yeti users that acquired sufficient credits could keep their phone. Allocation to reward classes was another feature relating to the preliminary frequencies of commuting - the base line for reduction of peak trips. During The last week data was collected without rewards, however information was still available to Yeti users. An evaluation survey was conducted in the third stage of the study.
The data was analyzed using longitudinal methods and mixed panel models. Initial results suggest that rewards can be an effective measure in changing travel behaviour. Specifically rewards reduce the shares of peak-hour driving, shift driving to early and later periods and increase the shares of non-driving alternatives ? public transport and working from home. However, once the rewards were terminated behaviour returned more or less to previous trends. Different behaviour trends were observed for the two different reward types. Money receivers tended to shift to earlier driving times whereas Yeti users were more likely to shift to later times and less likely to use non-driving alternatives.
In addition other factors influence the behavioural impacts of the rewards. These include 5 main categories: information availability; short and long run experience; situational constraints and/or support measures and lastly personal characteristics. Information tends to increase the level of variability in the response: traffic information tends to increase driving shares while public transport information increases non-driving shares. Long run experience (travel habits) and attitudes about alternatives (beliefs) tend to segregate the response to particular travel alternatives or limited the level behaviour change. Practicing before commencing the trial (in the short run) also supported behaviour change. Household or work related constraints restricted behaviour change while supportive measures in the work place had a positive effect. Personal characteristics such as gender and education levels were also found to have an influence on behaviour.
These results provide valuable insights into a future implementation of a reward scheme on a larger setting. Specifically, in an external study it was found that over reaction to the reward could cause significant time-loss to the road network. Thus understanding of what influences individual behaviour is quite important in this setting. Further work is being conducted now in improving the models for predicting the extent of the behavioural changes under the reward scheme.
Association for European Transport