The effect of different kinds of messages on the number of incorrectly placed bicycles : a field experiment

(1)

The effect of different kinds of messages on the number of incorrectly placed bicycles: A field experiment

Theo Dekker 10004598 Universiteit van Amsterdam

1

UNIVERSITEIT VAN AMSTERDAM

The effect of different kinds of messages on the

number of incorrectly placed bicycles: A field

experiment

Master’s Thesis

Student: Theo Dekker

Student Number: 10004598

Date: 21-06-2016

Education: MSc Business Economics – Managerial Economics & Strategy

Number of ECTS: 15 credits

Institution: University of Amsterdam, Faculty of Economics and Business

Supervisor: Prof. Hessel Oosterbeek

(2)

Theo Dekker 10004598

Universiteit van Amsterdam 2 Content 1 Introduction 5

2 Literature review and hypotheses 9

2.1 Cycling in the Netherlands 9

2.2 Bicycle parking in the Netherlands 10

2.3 Bicycle parking in Japan 11

2.4 Messages and rule compliance 16

3 Methodology 20 3.1 Field experiment 20 3.2 Background Houten 21 3.3 Locations 23 3.4 Treatments 24 4 Findings 26 4.1 Main findings 28 4.2 Other findings 37

4.3 Limitations of the findings 38

5 Discussion and conclusion 43

Appendix 1: Locations 45

Appendix 2: Examples of incorrect placed bicycles per location 50

Appendix 3: The messages 53

Appendix 4: Variable definitions 54

(3)

3 Statement of Originality

This document is written by Theo Dekker, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(4)

4 Abstract

This study held a natural field experiment to see which kind of message leads to the highest rule compliance. Incorrectly bicycle parking is a problem in many cities, this study tried to decrease the number of incorrectly placed bicycles by informing people. The experiment varied the text on the messages attached to the handlebars of bicycles that were misplaced. There were three different kinds of messages, one basic message, and two messages that contained additional information on top of the basic message. One message also highlighted the consequences towards others, and another message added additional information about a penalty. The results of this study showed that adding additional information did not affect bicycle users’ behavior differently from the basic message. But the number of misplaced bicycles was lower in the post-experimental period. The experiment showed that applying messages on bicycles appears to have beneficial effects, against minimal costs.

(5)

5 1 Introduction

In the Netherlands many people use the bicycle as a mode of travel, most people use it for travelling midrange distances that are too short to use the car and too long for a walk. Add reference. However, some bicycle riders park their bike at places that are not permitted for parking. To illustrate this problem people may park their bicycle at streets close to automobile traffic, on sidewalks, in pedestrian thoroughfares, and so forth.

People can park their bicycles at different areas in the city. Some cyclists may park their bicycle on street in an urban area for example near a shop, to go shopping. This is a problem, but the problem of inappropriate parking is often larger near railway stations. It is common in Japan and in the Netherlands to use the bicycle in combination with the railway for commuting trips (Martens, 2006). The number of users of a train station is large, and therefore some of the users may park their bicycle incorrectly.

For example, the municipality of Houten wrote the following in their weekly news update (Houten Actueel, 2015): “Around train station Castellum there are everyday bikes that are not placed right. Especially local residents suffer from it. Some sidewalks are even blocked by parked bicycles, allowing wheelchair users not to get through. This can lead to dangerous and troublesome situations.”

Moreover, misplaced bicycles lead to some other socially undesirable effects, for example it may prevent the flow of other bicycle users and constituting an eyesore (Fujii, 2005). This clearly indicates that inappropriate bicycle parking is a serious problem for municipalities.

When a bicycle user sees other illegally placed bicycles, the cyclist is more likely to also park his/her bike on a street illegally. This implies that on a location were a lot of bicycles are incorrectly placed, a large number of people will become accustomed to parking illegally. According to Dawes (1980), a situation like this is called a social dilemma. An individual tends to follow the actions of the majority. But in this case it may also bring a situation of inferior equilibrium. This is an undesirable state of society in general, and is robust against slight perturbations, including some policy interventions (Moffit, 2001). In such a situation it is very important which policy intervention should be implemented in order to break the aggregate and collective behavior that leads to this situation. Which policy

intervention can change the behavior of people with respect to inappropriate bicycle parking? Many municipalities try to encourage citizens to act in line with the rules. This is not

(6)

6

only focused on the right parking of bicycles and cars. But also about clean streets, intervene at attempts to steal and helping an old woman with shopping. In all these cases the

municipalities have a role, but also the citizens themselves.

In the last years there is a trend that most municipalities have to cut their expenditures. Some municipalities try to have smaller role in creating a pleasant and safe environment, and ask for a greater contribution off their citizens. The idea is that all citizens in a municipality have a key role in creating a pleasant, safe environment. In an ideal pleasant and safe

environment there is no illegal bicycle parking. To keep the city easily accessible for cyclists and livable for users and local residents, all bicycles should be placed correctly.

A municipality can try to change the environment surrounding bicycle parking. First of all, a municipality can expand the parking capacity. When there are more (or quicker) places to park the bicycles, the problem may be smaller. Moreover, it can monitor more frequently and punish people who park their bicycle inappropriate.

In some cities, for example Amsterdam, the municipality hires “bicycles coaches”. These coaches ensure that bicycles be parked in the designated areas. In this way local

residents do not suffer from illegally parked bicycles, however this is a costly option. Another option is to try to influence bicycle riders’ voluntary motivation with respect to observance of the laws. This can be done by public-education campaigns (Messick and Brewer, 1983; Dawes, 1980).

This study focuses on the trying to influence the motivation of the cyclists by informing them. Can informing the cyclists who park their bicycle incorrectly reduce the bicycle-parking problem? Many municipalities already inform cyclists about misplaced bicycles. They hang a message on the handlebar or put a sticker on the bicycles if they are misplaced. But does the message that is communicated matter? It is likely that people respond different to various messages. It might be the case that if people do know what the effect is of their action, they might be willing to change their behavior. But it is also possible that a threat and removing the bikes is more effective. The message can emphasize the consequences for others such as wheelchair users or emphasize the penalty for incorrectly bicycle parking. The main question is: does it matter which message is used, and if so which message is the most effective one to place the bicycles properly? To examine the effect of different kinds of messages on inappropriate bicycle placement, the following research question is formulated:

(7)

7 Research question:

What is the effect of different kind of messages on the number of incorrectly placed bicycles?

There is already some research on rules compliance and motivating individuals. However, the current literature cannot answer the research question. A field experiment addresses the research question above. The field experiment was held in Houten, a city with approximately 50,000 inhabitants. During the experiment there were three different periods.

In the pre-experimental period, this study counted the number of incorrectly placed bicycles per location. In total there were six locations in the city where the incorrectly placed bicycles were counted. There were observations from different days, and each day had three different observations: one observation in the morning, one in the afternoon and one in the late afternoon.

In the experimental period, a message was put on the misplaced bicycles. Hanging a message on handlebar belongs to the class of interventions known as nudges (Camerer et al., 2003). Nudge is a concept in political theory, behavioral science and economics, which argue that positive reinforcement and indirect suggestions to try to achieve non-forced compliance can influence individual’s motives and decision-making.

In the post-experimental period, the misplaced bicycles were again counted for multiple days. The expectation was that there would be on average significantly fewer bikes placed incorrectly. Moreover, this study also tested if different kind of messages had a different effect on the number of misplaced bicycles. In total there were three different types of messages. All the three messages informed the person that their bike is placed wrong. The first message only contained this information. The other two messages gave also some additional information. The second message highlights the consequences for other people if people place their bicycle incorrectly. The third message highlighted a penalty. If people incorrectly park their bicycles there is a chance that their bicycle will be removed. People can get their bicycle back if they pay a fine.

In summary, the findings of the experiment showed that the number of incorrectly placed bicycles was lower in the post-experimental group. This finding suggests that sending a message towards people can lead to higher rule compliance, and in this particular case can lead to fewer misplaced bicycles. Moreover, the experiment in Houten showed that adding additional information did not affect bicycle users’ behavior differently from the basic message.

(8)

8

This was not in line with expectations. The experiment showed that applying messages on bicycles appears to have beneficial effects, against minimal costs.

The content of this paper is divided as follows. The second section will provide an overview of the related literature and described the hypotheses. The third section will explain the methodology of the experiment. The fourth section will give the results of the experiment and the fifth part will conclude after discussing these results. The conclusion includes some limitations of this study and recommendations for further research.

(9)

9 2 Literature review and hypotheses

This section describes the related literature and hypotheses. First, some facts about cycling in the Netherlands will be described. The second part of this section will discuss literature related to bicycle parking in the Netherlands. Next, three studies that focused on bicycle parking in Japan will be described in more detail. At the end of this section the hypotheses are formulated with some additional literature about messages and rule-compliance.

2.1 Cycling in the Netherlands

The bicycle is a popular way to transport (Fietsberaad, 2009). After the car, the bicycle is the most popular way of transportation in the Netherlands. The bicycle is used in more than a quarter of all journeys (26%). Based on numbers of the 2007, 34% of all trips up to 7,5 km were made by bicycle. Bicycle use depends very much on the distance covered. Most journeys are still shorter than 7,5 km (70%). Moreover, it is interesting to note that that in 15% of journeys in the category 7,5-15km the bicycle is chosen.

There is a high level of diversity in the bicycle users in the Netherlands (Fietsberaad, 2009). The use of bicycles is not restricted to school-going children. The bicycle can be used for different kind of motives such as touring, education, shopping and business trips. When the journey has education as motive, the likelihood that some takes the bike is high (50%) in the Netherlands. When a person can use a car or a bicycle for shorter distances, many people do not make absolute choices. Most people sometimes use their car, and sometimes their bicycle. This depends per situation.

Cycling is popular in the Netherlands, but the urban bicycle share differs per municipality (Fietsberaad, 2009). In 2003 Groningen had the highest cycle share 38%, followed by Zwolle (37%) and Leiden (34%). The lowest cycle share was found in Heerlen (10%) and Rotterdam (16%). The municipality of Houten has an average cycle share of 22%. In the Netherlands there are more bicycles than people, this is unique in Europe. On average a Dutch person owns 1.11 bicycles. Moreover, the number of bicycles sold in the Netherlands is also high, 1.2 million in 2005 (Fietsberaad, 2009). In some countries cycling represents low social status and has a bad image. This is not the case in the Netherlands, the bicycle is used by almost every population group. The lowest and highest educational level are the ones that cycle the most, but the group with a higher incomes tend to use the bicycle less often compared to the group with a lower income. Bicycle theft is a serious problem in

(10)

10

the Netherlands, every year around 750.000 bicycles are stolen.

It is hard to tell why there is so much cycling in the Netherlands. It clearly lies in a combination of factors. Spatial and morphological factors, cycling is easier on a flat polder than in an area with mountains (Fietsberaad, 2009). Moreover, many trips in the Netherlands have a short distance and therefore can easily be covered by bicycle. Furthermore, some historical and cultural factors play a role, cycling is embedded in the Netherlands. Every child learns how to use it, and most children get a bicycle for their birthday (Fietsberaad, 2009).

2.2 Bicycle parking in the Netherlands

In the Netherlands there is an expertise center for cycling policy called the “Fietsberaad”. This center has done several studies about cycling and bicycle placement in the Netherlands. Based on research of Borgman (2010) member of the “Fietsberaad” there are too few bicycle racks in 1/3 of the locations in the Netherlands. This was based on data of 42 municipalities. On times with a large number of bicycle users the racks are full and there are still large quantities of bicycles outside the racks. Only at 38% of the locations there were enough bicycle racks.

The storage behavior of cyclists is strongly determined by the parking pressure (Fietsberaad, 2008). The number of incorrectly parked bicycles increases as the number of free places decreases during the morning. In the morning most cyclists attempt to still park their bicycle somewhat orderly. However, if there are fewer places left, the number of incorrectly placed bicycles increases, and those bicycles causes severe annoyance. Moreover, Fietsberaad (2008) held a survey around four train stations in the Netherlands about bicycle parking around train stations. The surveys were conducted in Eindhoven, Haarlem, Leiden and Nijmegen. Those cities have on average 150.000 inhabitants, and around 40.000 train travelers per day per station.

The main findings of the survey are that travel motive was often work (47%) or school related (31%). Another finding was related to the question if people always park on the same location around the station area. In almost all cases (91%) this was the same, this means that the choice where to place a bicycle is very strongly influenced by habitual behavior. Habitual behavior is hard to break.

The participants of the survey gave two main motives for selecting the storage

location. The arguments the ‘storage is on my route’ and ‘distance to platform/destination’ are the most important motives. Moreover, if a person can easily place their bicycle it may be

(11)

11

another reason to park their bicycle on that location.

In the four cities there were guarded parking locations around the train station, but a large group of users do not use those locations. The main arguments to do not use the guarded parking are the price and placing the bicycle takes more time.

The final finding of the survey is that it is also in the interest of the cyclists to decrease the number of illegally placed bicycles. However, there is no support amongst the cyclists for removing all illegally parked bicycles.

The studies done by the “Fietsberaad” were very informative about bicycle parking in the Netherlands. They were well organized and structured. However, most studies did only focused on bicycle parking around train stations. But inappropriate bicycle parking is a problem in the whole city. Moreover, most studies were from five to ten years ago, and therefore a new experiment related to bicycle parking could give some additional insights, or confirm the findings of previous research.

2.3 Bicycle parking in Japan

There is not very much literature about bicycle parking, and the problems related with bicycle parking. But there are three related studies that focused on inappropriate bicycle parking in Japan. One study describes reasons why people may park their bicycle incorrectly. The second study held a field experiment, and the other study focused on a survey. Bicycle parking is a problem in Japan (Fujii, 2005). To illustrate the problem local governments in major cities in Japan have enacted municipal bylaws, and tried to expand the capacity of bicycle parking spaces. Moreover, local governments regularly monitor areas in the city and remove inappropriately parked bicycles. This is the same in the Netherlands. This study will further discuss the three studies because they are related and relevant for the current study. However, it is beforehand unclear if those findings are generalizable to the Netherlands. For this reason, an experiment in the Netherlands could confirm some findings, or set some findings in a different perspective.

Hossain et al. (2003) did a case study about the bicycle-parking problem in Saga. This is a Japanese city with about 200,000 inhabitants. Both the legal and illegal parking behavior have been investigated by means of questionnaires and field surveys. Based on this paper there are two main factors that determine the manner of parking, legal or illegal. The first factor is the walking distance between the destination and the parking place. When the walking distance for a person increases the chance that the person will illegally park his/her

(12)

12

bicycle increases. The second factor relates to the type of the trip, for example a person can go shopping, or is the person moving to his/her employment, or does the person park his bike at a place and travels further with public transport.

The study focused on four locations around the train station of Saga. Many bicycles are left near the station in a disorderly manner because the cost of parking and of access distance.

There was a field survey that counted the number of wrongly placed bicycles on one day from 06:00-19:00. The total number of wrongly placed bicycles was counted but also the duration. Moreover, the people that held the survey made a guess about the age and gender of the cyclists who park their bike incorrectly. Most people who parked their bike incorrectly were young people (10-20 years old). This was a proportion of around 47%. A large majority of them were university and high-school students. The proportion females were high in the group of young people (67%).

People may park their bicycle wrong for a short period, but it may also be the case that people park their bike for a longer period. One the day of the survey 21% of the people

parked their bicycles for more than three hours, but there was also a group of people (71%) who parked their bicycles shorter than one hour. Based on this finding the authors conclude that most people who parked their bike incorrect were visiting one of the shopping centers. Based on an additional survey on parking lots, the authors suggest that shopping people are more sensitive to the distance to the parking lot compared to railway feeder passengers. And for this reason the group shoppers tend to park their bicycles more often misplaced compared to railway feeder passengers. Shoppers tend to park their bicycles near the shop in a non-parking zone because of short time parking. The chance that your bicycle will be removed is smaller if you do not park your bicycle very long. The local government of Saga periodically identifies and removes all illegally parked bicycles. Therefore it is not likely that people park their bicycles incorrectly for longer than a day.

A strong point of this paper was that it combines the findings of two surveys, which both gave insights about the bicycle-parking problem. Moreover, the paper shows the main factors why people park incorrectly, and made a suggestion that shopping people are more likely to park incorrectly. However, the field survey was based on data of one day, does these results hold for multiple days? Moreover, the experimenters made a guess about the age of the people who incorrectly parked their bicycle. This may had an influence on the results.

(13)

13

Furthermore, in Saga it was not possible to park your bicycle incorrectly for longer than a day, which is often possible in other cities.

The second study that will be discussed is the study of Fujii (2005). Fujii held a field experiment in Kyoto (Japan), a city with approximately 1.5 million residents. Fujii argued based on previous research (Klandermans, 1992; Hovland, Janis and Kelly, 1953; Petty and Cacioppo, 1986) that persuasive communication aimed at inducing cooperative behavior in social situation (such as inappropriate bicycle parking) might be a potentially effective strategy.

The author made a set of hypotheses that state how inappropriate bicycle parking may be reduced. It may be reduced by taking into account an awareness of consequences,

behavioral and implementation intentions, moral obligations and the influence of persuasive communication. The author hypothesized that persuasive communication could have an effect on the number of inappropriate placed bicycles. He hypothesized that persuasive

communication aimed at promoting behavior intention would not actually achieve its objective, but persuasive communication aimed at promoting an implementation intention would achieve its objective, reducing the frequency of inappropriate bicycle parking.

There were 99 participants in the field experiment. All the participants were students, and there was a large group of males. There were two different phases in the experiment. In both phases the participants had to answer a questionnaire about inappropriate bicycle parking. In both phases the participants were asked to answer self-reported frequencies of inappropriate bicycle parking. They were asked in which of the 23 situations they would park their bicycle incorrectly. All of the situations differed with respect to reasons for bicycle parking (leisure, shopping, using the train). Furthermore, all participants had to answer questions about awareness of consequences, moral obligation and the behavioral intention of desisting from in appropriate bicycle parking. This was done on a 7-point scale ranging. In the second phase the participants were divided in four different groups, one control group and three experimental groups. The three experimental groups received persuasive communication. There was one planning group, one with advice groups and one without advice group. The author tested if the number of inappropriate parked bicycles were lower after the persuasive communication.

It should be kept in mind that repeating a questionnaire can be seen as a treatment, which had a possible effect on all groups. The control group was only affected by this effect, the three other experimental groups were also affected by persuasive communication.

(14)

14

All the three experimental groups had a four-page leaflet and were requested to read it for five minutes. The leaflet illustrated four situations with consequences of inappropriate bicycle parking. Such as the effect on other people: a bicycle parked on a pedestrian thoroughfare prevents pedestrians from walking unimpeded; the Kyoto city spent

approximately 180 million yen in one year for monitoring and removing bicycles. The main message was “Let’s try to desist from inappropriate bicycle parking”.

The with advice group and planning group were given an additional four-page leaflet after the first leaflet and were also requested to read it for five minutes. The main message of the leaflet was “Let’s try to decrease inappropriate bicycle parking”. There was a statement indicating that it might be difficult to completely eliminate the inappropriate bicycle

placement problem but it might be possible to decrease it to some extent if riders were made aware of the effects of their behavior. Some suggestions in this leaflet were try to use bicycle-parking areas, even if they are not easily accessible or even if they are not free of charge.

The planning group was asked to read the two leaflets and after that they were asked to complete a questionnaire on how to desist from inappropriate bicycle parking for different situations.

After the persuasive communication the self-reported frequency of inappropriate bicycle parking did not significantly decrease in the control group. In the without-advice group it was reduced by 13%; in the with advice group the decrease was about 31%; and in the planning group the number of inappropriate parked bicycles was reduced by 26%. It is worth mentioning that it were mainly students who participated in this experiment and it was reported behavior. The actual behavior of the participants may be different from the self-reported behavior. Therefore, the question is to answer is: how does the actual behavior differ from the self-reported data? This could be tested in a field experiment.

The results of the experiment clearly showed that persuasive communication might be effective in reducing the number of inappropriate parked bicycles. More general, persuasive communication may be effective in inducing voluntary desistence from inappropriate

behavior. However, the hypothesis related to the planning group is not supported. The results did not show a higher reduction in inappropriate bicycle placement for the planning group. Furthermore, the study shows that an individual’s choice of bicycle parking location strongly is influenced by the decisions made by other bicycle users in his/her reference group.

The third study that will be discussed is the one of Fukuda and Morichi (2007). This study used a discrete choice framework as an application to analyze illegal bicycle parking

(15)

15

behavior. The authors argue that many types of travel behavior involve a positive social interaction (a conformity effect) and in some cases it leads to undesirable results. The study proposed a methodological framework by incorporating collective average choices in an individual’s discrete choice model and it conducts an empirical analysis to analyze inappropriate bicycle parking behavior.

The data was collected using a mail-back survey conducted in Tokyo in 2001. All the respondents could freely answer the questions, because their responses would be anonymous. A questionnaire was send to all of the 1616 randomly selected bicycles users. The sample consisted of 42% males. Moreover, 62% was the ratio of married people, and the average duration of parking was four and a half hour. However, the main disadvantage of a mail-back survey is that a certain type of persons responds to it. Is it likely that a person illegally parks his/her bicycle and also responded towards the mail-back survey? Moreover, based on the study of Hossain et al. (2003) mainly young people parked their bicycle incorrectly. But 62% of the participants were married in this study.

In total 163 participants send the survey back (10,1% response rate). The three

stations were Sugamo, Tabata and Ayase. These stations were selected because the proportion of illegal parking differs across these three reference groups. The percentage of illegal parking was 23,1 at Sugamo, 69% at Tabata and 84,6% at Ayase.

The questionnaire consisted of three modules. The first module focused on questions about individual and household characteristics. The second module was about the main purpose of the trips to the station, frequency and duration of bicycle parking, choice of bicycle parking location and walking time to the station. The choice of bicycle parking was a binary one: on street or off-street parking. Whereby on street parking can be seen as

inappropriate bicycle parking.

In the third module the authors collected information about the two main variables of the model of respondents’. The first variable is their attitudes towards risk. It was assumed that the more a person cares about risk, the more likely they will park their bicycles at the right places. The second variable is the level of public morality. If people demonstrate a higher level of public morality, people are more likely to park their bicycle correct.

The findings indicate that persons who frequently visit stations are more likely to park their bicycles off-street (right place). Frequent visitors of stations have a higher possibility of having their bicycles confiscated by the police, and might therefore prefer to park correctly. In line with this finding, the duration of parking increases the likelihood of correctly parking the

(16)

16

bicycle.

In line with the expectations the variable attitudes towards risks showed that risk aver se people are more likely to park their bicycles incorrectly. Moreover, individuals with a high level of public morality are more likely to avoid incorrect bicycle placement.

The most important finding of this email-back survey is related to the average choice level of other bicycles users. The choice of other bicycles users has a large significant influence of the parking choice. The result implies that people have a strong tendency to conform to the behavior of other people. When the majority parks their bicycle on the street, people will follow.

2.4 Messages and rule compliance

This paper contributes towards a better understanding of the mechanisms that may promote rule compliance. A cyclist who parks his/her bicycle incorrectly is violating a rule, and generates a negative externality on the population. While economics often think about monetary incentives, there is some literature that shows that monetary incentives may backfire (Gneezy and Rustichini, 2000; Benabou and Tirole, 2006), or that the monetary incentives are not feasible due to legal or political restrictions. For this reason, it is crucial to understand whether there are other possibilities to promote rule compliance. The goal of the paper is to analyze the effect of various types of message on inappropriate bicycle placement. Our interest in potential effects of informing cyclists who park their bicycles incorrect is that it has relatively low cost and is easy and flexible to implement. It is surprising that there is little known about the effectiveness of message intervention.

By studying compliance this study adds to a small but growing literature on messages and rule-compliance in other settings. Apesteguia et al. (2013) studied how to promote

compliance with rules that carry low penalties in public libraries in Barcelona. Different types of messages were sent by email towards users of the public libraries. The researchers found that showing a simple reminder was effective for promoting rule compliance. Surprisingly, adding other contents (social information or a reminder of a penalty) to the email did not increase the compliance significantly. Fellner et al. (2013) studied citizens’ subscriptions to TV licenses in Austria. Those researchers found no effect of social information or moral appeal mailings. However, they found that a legal threat email significantly increased the compliance rate. Furthermore, Pruckner and Sausgruber (2013) found that a moral message increased payments for newspapers sold based on an honor system. Cadena and Schoar

(17)

17

(2011) did research about loan-repayments in Uganda. They showed that a showing a text message increased repayment rates. Finally, Pomeranz (2010) did a study about firms’ tax compliance in Chili. She found that messages indicating a higher audit probability generate an increase in VAT payments. A unique setting of this studies experiment is that a message is only send to the people who are violating a rule. In the other literature, a message was send to a larger group of people.

In this study’s experiment, the first message represents a general reminder to the cyclist to park their bicycle at the correct place. There is evidence that people fail to pay attention to the relevant contingencies they are involved in (Szilagyi and Adams, 2012; Karlan et al. 2011). For this reason, mechanisms calling the attention of individuals to a specific message may improve their behavior related to it. Moreover, the above-described literature suggests that informing people leads to higher rule compliance. The first message mainly tests whether a message that simply recalling the fact that the bicycle is parked incorrectly improves behavior towards it.

Hypothesis 1: People are less likely to place their bicycle incorrect after receiving a message.

The framing of a message can have an effect on the amount of persuasion it elicits (Smith and Petty, 2014). Various processes can explain the differential persuasiveness of positively (or gain-) framed messages versus negatively (or loss-) framed messages. For example, the negativity bias hypothesis, cue effects and fear arousal hypothesis.

Risk and uncertainty in decision making was studied by Meyerowitz and Chaiken (1987), the findings of this research were that messages framing behavioral outcomes as negative consequences, or losses were more effective than message framed as a positive consequence of gains. For this reason this study used a negative message framing in the second and third message. In this way, both messages have the same kind of framing, which has a positive influence on the internal validity.

The second message adds additional information to the first message. It adds

information about the consequences towards other people, such as people living nearby and wheelchair users. There is a large literature showing that individuals often care about the well being of others, even when this comes at a personal cost (Charness and Rabin, 2002;

Andreoni and Miller, 2002).

(18)

18

obligation to perform a socially desirable behavior (such as incorrect bicycle placement) is related to the awareness of consequences of that behavior. A person may or may not be aware of the consequences of his/her action. To illustrate this, previous literature showed (Stern, Dietz and Kalof, 1993; Garling et al., 2003) that several types of awareness of consequences (e.g. to others, to the environment) with regard to undesirable behavior influenced the moral obligation to desist from such behavior.

In line with this, when a bicycle rider is informed about the socially undesirable consequences of bicycle parking on incorrect places (such as on a pedestrian thoroughfare), his/her moral obligation to desist from illegally parking might be activated. In line with Fujii (2005) the author of this study believes that in absence of such awareness, moral obligation will not be activated. For this reason this study expects that there is an effect of informing bicycle users about the consequences of inappropriate bicycle placements.

However, it may be the case some persons have formed an intention to desist from inappropriate bicycle parking, however still continue with such behavior. The reason for this is that people fail to implement their intention to desist from the behavior. The reasons include a weakness of the intention itself (Fujii and Garling, 2003a; Fishbein and Ajzen, 1975), an unrealistic intention or plan to desist (Garling et al., 1997) and a lack of actual behavioral control (Ajzen, 1985, 1991). The purpose of the second message is to test whether an appeal to the influence of one’s behavior on others improve rule compliance on top of the first message. This study expects that the effect of the additional information about the consequences on others will lead to a stronger decrease inappropriate bicycle placement compared to the message that only the standard information.

Hypothesis 2: The effect of the reduced number incorrectly placed bicycles is stronger if the

message contains additional information about the consequences on others compared to the message that gives only basic information.

The third message builds on the first message and adds information about a penalty, showing that incorrectly placed bicycles will be removed. If the bicycle is removed, people could get the bicycle back after paying a fine. Sanctions may play a role in the maintenance of

conformity and social order. According to Zimring and Hawkins (1968) sanctions are not relevant for the whole population. Sanctions are only relevant to those who have not

(19)

19

this setting the people who park their bicycle incorrectly. For this reason, only those people received a message.

This study aims to direct users’ attention to their own cost when breaking the rules. This is based on the large literature on the deterrence effect of punishment (Becker, 1974). The deterrence hypothesis predicts that the introduction of a penalty that leaves everything else unchanged will reduce the occurrence of the behavior subject to the fine (Gneezy and Rustichini, 2000). However, Gneezy and Rustichini showed that introducing fines can sometimes produce counterintuitive effects.

It is claimed that people are on average risk averse (Holt and Laury, 2002; Baker et al., 2007;), they engage in various forms of risk taking behavior such as gambling and smoking. But also in inappropriate bicycle placement can be seen as risking taking behavior if people received the third message. When informed, people know that there is a possibility that their bicycle will be removed. Based on the research of McAdams (2011), the probability of detection and punishment has a much greater deterrent effect if they belong to the active knowledge.

Furthermore, as prospect theory posits (Tversky and Kahneman, 1981), losses loom larger than gains when evaluated in comparison to a neutral reference point. Based on this theory informing people about the cost when breaking the rules should decrease the number of misplaced bicycles. A possible loss when parking incorrectly looms larger than a possible gain from incorrectly parking your bicycle. The purpose of the third message is to test

whether focusing the attention of users towards a penalty improves rule compliance on top of the basic message. This study expects that the effect of a penalty will lead to a stronger decrease inappropriate bicycle placement compared to the message that only the standard information.

Hypothesis 3: The effect of the reduced number incorrectly placed bicycles is stronger if the

message contains additional information about a penalty compared to the message that gives only basic information.

A priori it is hard to expect which of the two messages that contained additional information is more effective against inappropriate bicycle placement.

(20)

20 3 Methodology

This section describes the methodology used in this study. First, the aspects of the field experiment will be discussed. Second, there will be some background information about the city of Houten, where the field experiment took place. Thirdly, the locations will be

discussed. At the end of the section the different treatments will be shown that were used to influence the behavior of individuals.

3.1 Field Experiment

A field experiment was conducted to test if informing the bicycle users who park their bicycle incorrectly can decrease the number of incorrectly placed bicycles. The experiment took place in a natural setting in which antisocial behavior (misplaced bicycles) is relatively anonymous. This is a problem based on an everyday scenario where there is a simple and well-defined rule, which is known by almost all users. That people should park their bicycle on the right places.

There were three different periods in the experiment: the pre-experimental period, the experimental period and the post-experimental period. The first period consisted of two weeks of counting the number of incorrectly placed bicycles. The second period was a week of informing the bicycle users who incorrectly park their bicycle. In the third period the number incorrectly placed bicycles were again counted for two weeks. The misplaced bicycles were counted on six different locations. In the weeks during the first and third period the number of incorrect placed bicycles were counted on every Tuesday, Wednesday and Thursday. During these days there are three measurement moments: one in the morning (09:00), one in the afternoon (13:00), and one end of the afternoon (16:00). The first period is from 11 April to 24 April, the second from 9 May to 15 May and the third period is from 16 May to 29 May. An advantage of this setup is that the number of incorrectly placed bicycles was counted on fixed days.

There is a fixed routine in counting the incorrectly placed bicycles. First of all, there is a fixed route along the locations, whereby first location 1 was counted, second location 2 etc. At the start of each location the start time and end time were denoted. And in each location there is also a fixed route. This is important because it may happen that a bicycle will be incorrectly placed during the counting period, or the owner will remove an incorrectly placed bicycle while the misplaced bicycles were counted.

(21)

21

In the experimental period the cyclists who parked their bicycle incorrectly, received a message on their bike. There were three different kinds of messages. The messages were attached to the handlebars. On Tuesday 10 May, Wednesday 11 May and Thursday 12 May messages were attached to the bicycles. The same route was cycled as in the pre- and post-experimental period, however when a bicycle was placed incorrectly it received a message. On those days there was one bicycle trip in the morning and one in the afternoon. It took some time to provide each bike with a message. In total there were 741 messages put on the bikes in the experimental period. The first type of message was put 237 times on bikes, the second type 143 times and the third type of message was put on 361 bicycles. However, the number of messages that were putted on the bikes was not counted per day and location. In the post-experimental period the number of incorrect placed bicycles were again counted. Based on the hypotheses the expectation is that the number of incorrect placed bicycles was lower in this period. However, other factors may explain the variation in the number of incorrect placed bicycles. An example is the weather. On a sunny day, on average more people will use their bikes. When more people are using their bicycles, it is likely that the number of incorrect placed bicycles is higher on this day. To control for this, this study will measure the effect for multiple days. And on every day that the bicycles will be counted the temperature will be noted. Furthermore, the bicycle grade of Houten according to

Weeronline will also be noted. With this additional information we can compare the days before the messages and afterwards.

There was a period of two weeks wherein no bicycles were counted because of some free days in the Netherlands, and this was a holiday period for primary- and high schools. For example on 27 April almost all people are free because of Kingsday, a national holiday. This will have an effect on the number of incorrectly placed bicycles on the 27th

of April, but most likely also on the number of incorrectly placed bicycles on the 28th

of April.

3.2 Background of Houten

The experiment was held in Houten. In this section there is some background information about Houten, cycling in Houten and the two large bicycle parking’s in the city. Houten has approximately 50,000 inhabitants, and most shops are in the center of the municipality. The inhabitants of Houten do almost all their food shopping and half of their non-food shopping in their own municipality (Fietsberaad, 2009). Houten is a town in the vicinity of Utrecht. The city is oriented towards cyclists, and Houten won the title of “Fietsstad” in 2008, top bike city

(22)

22

of the Netherlands (Houten, 2016). Groningen and Zwolle are two other examples of bicycle friendly towns, and a bicycle friendly town results in a better quality of urban life according to a publication of the Fietsberaad (2009).

The urban design of Houten remains a special example for other cities. Because of the structure of the city, people are encouraged to travel by train and bike. The shopping center and railway station form the core of the city. The residential areas are around the center, but have a lower construction density. From the center there are several bicycle and pedestrian networks forming direct routes to the residential areas. Every residential area can still be reached by car, but when people use their car they always have to use the ring road to get to the center or to another residential neighborhood. For this reason, in Houten there is relatively more cycling and walking than in comparable centers. Moreover, this is one of the reasons why Houten is one of the safest cities in the Netherlands (Fietsberaad, 2009). Cars and cyclist are able to avoid each other, and there are child-friendly bicycle paths.

Houten has a good bicycle network and there is a bicycle route that passes under the railway. Moreover, it is worth mentioning that Houten has two unique bicycle parking locations near the train stations. The bicycle parking locations called “Transferiums” provide an easy and quick transition from bike to train. The municipality Houten is the boss on the parking locations. In most of the cases the NS (Dutch Railways) owns the parking location of the train station. Both locations are managed by NS bicycle and U-stal. The two bicycle-parkings have free security, with enough racks to place 5000 bicycles.

Van Beek (2012) held a survey amongst inhabitants of the municipality Houten about the two bicycle “Transferiums”. The main findings of the survey are that 57% of the

inhabitants did use the parking place at the center of the city, and 26% of the inhabitants used the parking place at Houten Castellum in 2011. The expectation is that nowadays that the usage rate of the second parking place at Houten Castellum will be higher. The users of the parking’s places rated both places with an eight. The main advantages according to the users are that the locations are easy to find, good located, and are safe. The main disadvantage is that is hard to place your bicycle in the upper racks. Almost all users make use of the parking place because they travel by train (93%), but there are also some users that place their bicycle to go shopping (25%).

Two times a year the municipality removes abandoned bicycles and bicycle wrecks on three locations in the city, “Het Rond”, “Oude Dorp” and “Castellum”. The center areas are not intended to leave bicycles or bicycle racks to store long. These three center locations are

(23)

23

the locations that are used for the experiment. But the municipality does not do that much about incorrectly placed bicycles. There are some people who work for the cities surveillance and enforcement. Those people rarely replace a bicycle that is parked incorrectly. The author of this is study believes that the municipality could have done more to change the

environment surrounding bicycle parking, and decrease the bicycle-parking problem. It could expand the parking capacity, by more or quicker places to park the bicycles. Furthermore, it could monitor more frequently and punish cyclists how park their bicycle incorrectly. Finally, the municipality could try to influence bicycle riders’ voluntary motivation with respect to observance of the laws. It already tries to motivate bicycle riders’ voluntary motivation by showing some message about bicycle parking. For example, a message that shows the text “bicycle wrong = bicycle gone”, or a symbol that shows forbidden to park bicycles and mopeds.

The experiment of this study also focuses on the motivation of the cyclists to park correctly.

3.3 The locations

The number of incorrect placed bicycles was counted on six different locations in Houten. The six different locations will now be described. Figure 1 of appendix 1 shows a map of Houten with the six locations, and figure 2 shows a detailed map of the center of Houten. In this area there are four locations. Furthermore, a detailed map is shown for each location (Figure 3-8). In each location there are one or more shops, this can be seen in the maps. Some pictures of incorrectly placed bicycles are shown to illustrate the problem in Appendix 2. For each location, there is one picture that shows at least one bicycle that is not parked correctly. Next, the six locations will be described. Whereby it is important to understand that it differs per location when a bicycle is labeled as misplaced. In some areas it is forbidden to park bicycles, and in some areas bicycles should be placed inside the racks. A bicycle that is parked in an area where it is forbidden to park (Picture 1 of Appendix 2) has more

consequences for the municipality than a bicycle that is parked near a bicycle rack, but not in the rack (Picture 2 of Appendix 2). The locations:

1: Houten Castellum around the “Transferium”

In this area it is forbidden to park your bicycle, every bicycle should be placed inside the parking. There are some apartments in this area, a fitness center and some shops.

(24)

24

2: Houten Oude Dorp

At this location there are several bicycle racks near the different restaurants, bars and shops. The bicycles should be placed in the bicycle racks. Moreover, there are some iron bars where people can place their bike. A bicycle is placed wrong is the place there bicycle not in the racks or not near the iron bar.

3: Houten around the Transferium

In this area it is forbidden to park your bicycle, every bicycle should be placed inside the parking. In this area there is library and some shops.

4: Houten around the Jumbo/Vershof

At this location there are several bicycle racks for the different shops. The bicycles should be placed in the bicycle racks.

5: Houten het Rond

At this location there are several bicycle racks near the different restaurants, bars and shops. The bicycles should be placed in the bicycle racks. Every Thursday there is a market around this location. The market is from 08:00 to 14:00.

6: Houten de Raat

At this location there are several bicycle racks for the different shops. The bicycles should be placed in the bicycle racks. This is the location with the fewest incorrectly placed bicycles.

3.4 Treatments

This study used three different messages to influence behavior of individuals. All the three messages informed the person that their bike is placed wrong. The first message only contained this information. The other two messages gave also some additional information. The second message highlights the negative effect on other people if you place your bicycle incorrectly. The third message will tell the person that there is a chance that their bicycle will be removed and after paying a fine they will get their bicycle back.

In Appendix 3 the different kinds of messages are shown. However, those were written in Dutch. The translated messages are:

(25)

25

1) Your bicycle is incorrectly parked. Place your bicycle in the designated areas!

2) Your bicycle is incorrectly parked, wheelchair users and local residents experience being inconvenienced if you do not park your bicycle properly.

Place your bicycle in the designated areas!

3) Your bicycle is incorrectly parked, the municipality will remove misplaced bicycles. If your bicycle is taken away, you will have to pay a fee if you want your bicycle back. Place your bicycle in the designated areas!

Moreover, in Appendix 3 there is an example of a message attached to a handlebar of bicycle of the experiment. In the right picture you see a message that was used in Zwolle. This is used as an example.

Location 1 and 4 received the first message, location 2 and 5 received the second message and location 3 and 6 received the third message. This means that the message is the same for two locations.

(26)

26 4 Findings

In this section the findings of the experiment will be discussed. First, the number of

incorrectly placed bicycles will be shown for the different periods. Second, a comparison will be made for the two different periods. Additional regressions will be shown to show the effect of the different kind of messages, and if the messages had an effect. Furthermore, some other related findings will be described about the experiment. Finally, some limitations for the findings will be described.

Table 1

Number of incorrect placed bicycles in the pre-experimental period

Date 12-Apr 13-Apr 14-Apr 19-Apr 20-Apr 21-Apr Total Mean Std. Dev

Temprature 17 17 17 14 14 17 16.00 1.549 Bicycle grade 8 10 8 8 10 9 8.83 0.983 Location 1 Morning 31 44 41 43 36 42 237 39.50 5.010 Afternoon 55 50 53 49 54 52 313 52.17 2.317 Late afternoon 49 48 53 43 44 48 285 47.50 3.619 Location 2 Morning 9 12 12 6 11 10 60 10.00 2.280 Afternoon 15 31 18 20 24 22 130 21.67 5.538 Late afternoon 25 34 34 26 38 29 186 31.00 5.138 Location 3 Morning 44 51 47 45 48 45 280 46.67 2.582 Afternoon 81 71 85 86 75 82 480 80.00 5.865 Late afternoon 68 67 67 54 56 62 374 62.33 6.088 Location 4 Morning 15 21 19 15 15 19 104 17.33 2.658 Afternoon 26 31 21 20 29 27 154 25.67 4.367 Late afternoon 29 43 27 19 20 25 163 27.17 8.681 Location 5 Morning 10 8 14 7 9 11 59 9.83 2.483 Afternoon 14 16 31 22 20 26 129 21.50 6.317 Late afternoon 24 26 17 28 20 24 139 23.17 4.021 Location 6 Morning 3 9 7 5 5 6 35 5.83 2.041 Afternoon 8 8 10 8 6 9 49 8.17 1.329 Late afternoon 13 9 12 8 10 11 63 10.50 1.871

Note: This table shows the number of incorrect placed bicycles in the pre-experimental period. This period was from 10 April to 23 April, and had six measurement days. The source used to collect information about the temperature and bicycle grade was Weeronline

(27)

27 Table 2

Number of incorrect placed bicycles in the post-experimental period

Date 17-May 18-May 19-May 24-May 25-May 26-May Total Average Std. Dev

Temprature 17 18 18 14 16 20 17.17 2.041 Bicycle grade 9 7 9 7 8 9 8.17 0.983 Location 1 Morning 40 42 37 40 39 37 235 39.17 1.941 Afternoon 51 46 46 47 52 242 48.40 2.881 Late afternoon 45 46 48 39 42 48 268 44.67 3.559 Location 2 Morning 8 9 12 9 10 7 55 9.17 1.722 Afternoon 19 25 11 13 18 86 17.20 5.495 Late afternoon 17 25 32 19 27 29 149 24.83 5.811 Location 3 Morning 44 45 41 31 36 39 236 39.33 5.241 Afternoon 72 64 73 60 69 338 67.60 5.505 Late afternoon 56 61 59 54 56 60 346 57.67 2.733 Location 4 Morning 17 5 7 11 8 14 62 10.33 4.546 Afternoon 16 19 23 25 22 105 21.00 3.536 Late afternoon 32 14 25 20 28 26 145 24.17 6.338 Location 5 Morning 6 10 13 5 7 12 53 8.83 3.312 Afternoon 19 21 11 15 22 88 17.60 4.722 Late afternoon 16 21 20 18 19 23 117 19.50 2.429 Location 6 Morning 2 6 3 1 2 4 18 3.00 1.789 Afternoon 7 5 4 4 5 25 5.00 1.225 Late afternoon 5 6 6 7 5 8 37 6.17 1.169

Note: This table shows the number of incorrect placed bicycles in the post-experimental period. This period was from 16 May to 29 May, and had six measurement days. On May 17 there was no measurement in the afternoon, for this reason the average and total number of incorrectly placed bicycles is based on 5 measurement moments. The source used to collect information about the temperature and bicycle grade was Weeronline. Additional information about the locations can be found in Appendix 1.

(28)

28 4.1 Main findings

Table 1 and 2 report the number of incorrectly placed bicycles in the first period and third period of the experiment. It shows the number of incorrectly placed bicycles per

measurement day for the different locations. In each period there were six measurement days. In the first period those were from 12 April to 14 April and from 19 April to 21 April. In the third period those were from 17 May to 20 May and from 24 May to 27 May. On these days the number of incorrectly placed bicycles were counted three times per day for six locations. However, in the third period the number of misplaced bicycles was not counted on 17 May in the afternoon. It was not possible to count on this moment because of study duties at the university. For this reason, the measurement moments in the afternoon are based on five observations in this period.

In the first period the average temperature was 16 Celsius, and the average bicycle grade in this period was 8.33 according to Weeronline. In the third period the temperature was slightly higher with an average of 17.17 Celsius. The bicycle grade in the third period was 8.17.

For each location and measurement moment the total, the mean and standard deviation for the number of misplaced bicycles are shown in table 1 and 2. The number of misplaced bicycles varies per location. Location 3 had the highest number of misplaced bicycles and location 6 the lowest number.

In both periods the number of misplaced bicycles was the lowest in the morning. It is worth noting that on the locations with a train station and bicycle parking (location 1 and 3) the highest average number of misplaced bicycles is in the afternoon. For the other four locations the highest average number of misplaced bicycles is in the late afternoon.

Table 3 reports a comparison of the average number of incorrectly placed bicycles for the different periods. It shows the mean values based on in the information provided in table 1 and 2. Furthermore, it shows the absolute- and percentage difference between the two periods. It compares the pre- and post-treatment periods. The last column in table 3 reports the p-values for the F-test of equality of variable means across the groups. This was calculated with a one-way Anova test.

Based on the information in table 3, the temperature and bicycle grade were not significantly different for the two periods. The table shows a p-value of 0.291 and 0.267, this means that the periods are comparable in temperature and bicycle grade. In both periods it has hardly rained.

(29)

29 Table 3

Comparison of the average number of incorrectly placed bicycles

Period I III Difference Difference in % p-value

Temprature 16.00 17.17 -1.17 -7.29% 0.291 Bicycle grade 8.83 8.17 0.67 7.55% 0.267 Location 1 Morning 39.50 39.17 0.33 0.84% 0.882 Afternoon 52.17 48.40 3.77 7.22% 0.039 ** Late afternoon 47.50 44.67 2.83 5.96% 0.201 Location 4 Morning 17.33 10.33 7.00 40.38% 0.009 *** Afternoon 25.67 21.00 4.67 18.18% 0.087 * Late afternoon 27.17 24.17 3.00 11.04% 0.510 Location 2 Morning 10.00 9.17 0.83 8.33% 0.491 Afternoon 21.67 17.20 4.47 20.62% 0.214 Late afternoon 31.00 24.83 6.17 19.89% 0.080 * Location 5 Morning 9.83 8.83 1.00 10.20% 0.567 Afternoon 21.50 17.60 3.90 18.14% 0.262 Late afternoon 23.17 19.50 3.67 15.83% 0.085 * Location 3 Morning 46.67 39.33 7.34 15.72% 0.012 ** Afternoon 80.00 67.60 12.40 15.50% 0.006 *** Late afternoon 62.33 57.67 4.67 7.49% 0.118 Location 6 Morning 5.83 3.00 2.83 48.57% 0.029 ** Afternoon 8.17 5.00 3.17 38.78% 0.003 *** Late afternoon 10.50 6.17 4.33 41.27% 0.001 ***

Note: This table shows a comparison of the number of incorrectly placedbicycles for the different periods. It additionally shows the absolute difference and percentage for the periods. The first period was from 10 April to 23 April, and the third was from 16 May to 29 May. The messages were different per location. The first and fourth location received the first message, the second and fifth location received the second message. The third message was received by the third and sixst location. Additional information about the locations can be found in Appendix 1. Anova tests were done to test if the difference was significant. ***, ** and * indicate significance at the 1%, 5% and 10% significance levels respectively.

The average number of misplaced bicycles was in the post-experimental period always lower compared to the pre-experimental period. This is in line with the first hypothesis. A priori was expected that the number of misplaced bicycles was lower in the period after the experimental period. In ten of the eighteen measurement moments the number of misplaced

(30)

30

bicycles was significantly different. However, the absolute and percentage changes differ a lot per location and measurement moment. The largest percentage (48.57%) change was in the morning at the sixth location, however the absolute decrease for this moment was only 2.83 bicycles. The smallest percentage change was 0.84% at the first location in the morning. In the afternoon at location 3 the number of bicycles decreased by 12.4 bicycles. This was the largest decline.

The table shows significant differences between the different types of messages. The first and fourth location received the first message with information that their the bicycle is misplaced. The second and fifth location received the second message with some additional information on top of the basic message about the consequences towards other people such as wheelchair users and local residents. The third and sixth location received the third message with additional information on top of the basic message about a penalty if people do not correctly park their bicycles. Next the results per message will be described.

The first location received the basic message. But at location 1 the messages only had a significant effect on the average misplaced bicycles in the afternoon. The absolute

difference was a decrease of 3.77 bicycles in the afternoon, this was a decrease of 7.22 % compared to the first period.

The fourth location also received the first message. At this location the number of incorrectly placed bicycles decreased significantly in the morning and afternoon. The number of incorrectly placed bicycles decreased by 7 in the morning, this was decrease of 40.38%. This difference was significant at a 1% level.

The second location received the second type of message with additional information about the consequences towards others. The messages had only a significant effect in the late afternoon. In the late afternoon the number misplaced bicycles decreased by 6.17, this was a decrease of 19.89%. The Anova test showed that the means of the different periods were not equal, and therefore the decrease had a significant effect (10%).

The message that showed additional information about the consequences towards other people was also used at the fifth location. Table 3 reports a significantly effect of the messages in the late afternoon at the fifth location. It shows a decrease of 15.83% which was significant at a 10% level. The number of incorrectly placed bicycles decreased from 23.17 to 19.50.

(31)

31

The third type of message was used at the third and sixth location. The third type of message was the message that showed the most significant effects. This type of message highlighted a penalty for people who do not park their bike correctly. At the third location the number of incorrectly placed bicycles decreased with 15.72% in the morning, 15.50% in the afternoon and 7.49% in the late afternoon after the experimental period.

At the sixth location the effect of attaching messages on the handlebars of bicycles had the strongest effect. This was the location with the lowest number of incorrectly placed

bicycles, but the number misplaced bicycles decreased about 40% at all the three measurement moments.

It should be kept in mind that the findings are based on a field experiment with only six locations and six measurement days per period. The locations differ in the number of incorrectly placed bicycles. Therefore the findings are harder to analyze and compare the results for the different locations. For example the decrease in the first period in the afternoon was 7.22% and significant at a 5% level. A decrease at the sixth location in the morning was also significant at a 5% level, but this decrease was a larger namely 48.57%.

The first hypothesis stated that people are less likely to place their bicycle incorrect after receiving a message. To answer this question it was more convincing if this study used one or two control locations in Houten. However, this study did not use a control location. To answer the first hypothesis this study used a regression, in which the data is pooled from the different locations and in which there is controlled for the location, measurement moment, temperature and bicycle grade. The following research model is used in this study:

Misplaced Bicycles = β0 + β1 After+ β2 Location2+ β3Location3 + β4 Location4 + β5

Location5 + β6 Location6+ β7 Moment2+ β8 Moment3 + β9 Temperature +β10 BicycleGrade + ∈i

The dependent variable is the number of misplaced bicycles. The variable of interest in this regression is the variable After. This is a dummy variable that equals 1 for the

post-experimental period, and 0 for the pre-post-experimental period. There is controlled for the different locations, measurement moments, bicycle grade and temperature. The omitted treatment to which the locations are compared is the first location. Moreover, the omitted treatment to which the measurement moments are compared is the one in the morning. Table 4 reports the results of the OLS regression of the number of misplaced bicycles. Column (I) represents an OLS regression with control variables and the t-statistics are corrected for