CLEANLINESS AND REGULARITY:

(1)

CLEANLINESS AND REGULARITY:

THE INFLUENCE OF POLLUTION AND ARCHITECTURE ON

CUSTOMER BEHAVIOR

1

CLEANLINESS AND REGULARITY:

HE INFLUENCE OF POLLUTION AND ARCHITECTURE ON

CUSTOMER BEHAVIOR

Jelle de With

Utrecht July 2nd, 2013

CLEANLINESS AND REGULARITY:

(2)

2

CLEANLINESS AND REGULARITY:

THE INFLUENCE OF POLLUTION AND ARCHITECTURE ON

CUSTOMER BEHAVIOR

Master thesis, MscBA, specialization Marketing Management University of Groningen, Faculty of Marketing

Words: 13.465 July 2nd, 2013 Jelle de With student number: 1937138 Cornelis Evertsenstraat 27 3572 JR Utrecht Tel: +31(0)6 1110 5673 E-mail: jelledewith@gmail.com Supervisors:

1st supervisor: Dr. J.A. Voerman 2nd supervisor: Dr. M.C. Leliveld

Federal University of Groningen, Faculty of Marketing Nettelbosje 2

9747 AE Groningen

(3)

3

Management summary

The negative effect of pollution on customer behavior is weakened by non modern

architecture. Customers prefer interesting city-centers. The build environment, known as the servicescape (Bitner, 1992), influences their decisions to approach or to avoid a certain shopping area. Cities and shops rival intensely to attract customers. The municipalities and organizations invest a lot of money in new buildings to improve the customers’ well-being and so influence their behavior. Big new modern buildings, like train stations and shopping malls, are build. But after a few years the buildings start showing signs of decay and disorder is spreading. Customer well-being is decreasing and avoidance lies in wait. What are the effects of architecture and pollution on customer behavior?

The build environment has many elements that influence the customer. From the service management perspective, elements like color, noise, smell, lighting and temperature (Bitner, 1992) influence customers perception of the servicescape. In environmental psychology, openness, complexity and order are elements that determine peoples approach or avoidance (Mehriabian and Rusell, 1972). How do customers react on a polluted environment and what moderating effect has architecture on this relation.

The purpose of this research is to investigate the effects of pollution and architecture on customer behavior. This is tested through comparing the build environment of train stations. First, two clean types of stations are tested: non modern vs. modern. Secondly those two stations are polluted. The effects are explored by using a 2 x 2 experimental design, were the environments are tested in measures of customer behavior. Besides the interaction of pollution and architecture, personal preferences are taken into account for their influence on customers’ perception.

The results indicate that there is a significant relationship between pollution and architecture on customer behavior. Participants who regularly travel by train valued non modern

architecture higher than modern architecture. When the station becomes polluted all participants showed decreases in appreciation. In a modern station this effect was even

(4)

4

Preface

In the summer of 2003, I decided to study Industrial Design in Enschede. I finished my bachelor in 2009 but felt the urge to study further. I was looking for a new challenge and decided to go to the university of Groningen for a masters degree in Marketing. A few hubs and bumps down the road later I have finally made it to my destination. The final test, the master thesis.

Although I started in whole new field of study, design and engineering never lost my interest. In the field of service marketing I found the connection between marketing and industrial design. I always thought, an interesting environment could enlighten people and would positively influence them. When I was ready for my master thesis I decided to research the effects of architecture on customer behavior. It took me quite some effort to get to this point but now I am very proud of the result

(5)

5

1. Introduction

1.1 Background

Architects and contractors always have to find the right balance in how they will satisfy all stakeholders and keep cost to a minimum. They must try to balance building and maintenance cost with building aesthetics. This question has become even more relevant through the economic crisis and the vacancy of shopping and office space. De Volkskrant (1-6-2013) reported over 13 million m2 of vacant commercial space. The website

www.overbewinkeling.nl even argued for re-use of shopping areas instead of building bigger

and newer shopping malls. Shopping areas that were build in the 70’ and 80’ of the previous century are often pointed out to be decayed and reported as unclean (VPRO, 14-5-12) These areas were build with ‘new techniques’ that made it possible to build enormous shopping centers with much lower costs than before. However, those building proved to be not

sustainable and decay started spreading. Through these signs of decay, disorder (e.g. graffiti, destruction, pollution) starts spreading in these environments. This has strong consequences. Disorder in the environment made people behave more opportunistic (Keizer et al. 2008) and signs of disorder triggers more behavior of demolishing and pollution (Wilson and Kelling, 1982; Cialdini, 1990) These signs of disorder in the build environment have their influence on customers and their behavior. (Vilnai-Yavetz and Gilboa,2010; Bitner, 1992). Their

judgments of the environment make them decide whether they want to avoid or approach these places. This is described by Mehrabian and Russell (1974) as the stimulus-response model (SOR). Customers approach or avoidance behavior is triggered through environmental stimuli which can be experienced as arousing or pleasant . Cleanliness, for instance was found to be a strong predictor of customers’ quality perception of the service (Wakefield and

Blodgett, 1996). As such it is of high importance to get to better understand the relation between pollution and architecture and its effect on customer behavior.

(8)

8

Hellenic railways, Greece. In relation to the SOR model of Mehrabian and Russell (1972), an unsatisfying train station can result in avoidance of the train station. The train station than acts as an entry barrier for the actual service of public transport.

1.2 Problem statement

The main objective of this research is to gather insights about how pollution in the build environment influences customer behavior, and how the build environment moderates this effect. As previously mentioned, research in service marketing and environmental psychology showed single effects of pollution and the build environment but so far no connection is made between their relevance on topic of customer behavior. So, hopefully the outcome of this research will give new insights on how customer behavior is influenced by disorder in their surroundings. The following research objective is formulated: ‘How does pollution affect

customers’ behavior in the build environment and how is this effect moderated by elements of the architecture?

In order to define whether pollution and architecture influence customer behavior, the following research questions are formulated.

• What is the effect of pollution in the build environment of a train station on customer

behavior?

• What is the effect of the architecture in the train stations build environment on

customer behavior

• Is the effect of pollution on customer behavior moderated by the train stations build

environment?

1.3 Relevance

(9)

9

1.4 Structure

This research will continue with an exploratory literature research in chapter 2. It will build on the theoretical framework to support the hypotheses formulated at the end of the chapter. The literature research will provide a deep and broad understanding of the subjects. Data will be collected from published academic literature. The theoretical framework will elaborate on the effects of pollution and design aesthetics in the servicescape and how it influences customers. Subsequently, to guide the research the methodology and research design is

discussed in chapter 3. Chapter 4 elaborates on the results from the analysis. Finally, chapter 5 will contain the conclusions drawn from the results. The research question, through the

hypotheses, will be answered and possible additional conclusions are made. This research should be able to identify the effects of pollution and architecture on customer behavior. The chapter will continue with stressing the limitations and finally present the managerial

(10)

10

2. Theoretical framework

Does a polluted servicescape affect customers’ feelings? And what is the moderating effect of the aesthetics of the room/building in which the servicescape sets. Through this literature research the relevant variables are discussed.

2.1 Servicescape

‘All social interaction is affected by the physical container in which it occurs’ (Bennett and Bennett,1979). This physical container is defined by Bitner (1992) as the servicescape, the total configuration of environmental dimensions. It describes how the built environment affects both consumers and employees in service organizations (Bitner, 1992). The servicescape contains the physical evidence. Informational cues that help the customers develop their beliefs, feelings, and behavioral intentions towards the store offerings (Baker et al. 2002) The servicescape model is anchored in the environmental psychology research tradition and also draws together relevant literature in marketing, organizational behavior, human factor/ergonomics, and architecture.

Importance

The servicescape provides a framework which facilitates managers to control the build environment of the service. It can attract or repel customers. Mehrabian and Russell (1974) and Bitner (1992) argued that the servicescape can lead to either approach or avoidance behaviors. Approach behaviors are defined as the desire to explore the environment, lengthen one’s stay there, interact with others and perform activities within the environment.

Avoidance behaviors are defined as the opposite (Russell and Mehrabian, 1974). When environments become distressing, arousing and unpleasant, people try to avoid these places. When entering a servicescape, the customer judges the environment on environmental factors and decides what action to perform. Again customers try to avoid servicescapes that are perceived as distressing and unpleasant. A good servicescape is a pleasant atmosphere in which the service is delivered. The effect of arousal has an inverted U shape. Little or no arousing is perceived as boring and uninteresting. As arousal increases, interest increases. But when arousal exceeds a critical value it can also be negative. People experience the

(11)

11

More practical, the servicescape provides a visual metaphor for an organization’s total offering. The servicescape acts as a package, similar to a product’s package, by conveying a total image. It suggests the potential usage and relative quality of the service (Solomon 1985). Before purchase, consumers commonly look for cues about the firm’s capabilities and quality (Berry and Clark 1986, Shostack 1977). They judge these cues and decide whether or not approach the servicescape. Changing the configuration of the servicescape can also have significant impacts. Spangenberg et al.( 2006) and Bruggen et al. (2011) found increased customer spending, at least in the short term, when servicescape remodeling were perceived positive by the customers.

In addition, the configuration of the servicescape can boost or restrain the organizations marketing goals.. Harris (2008) proved the link between perception of the servicescape and loyalty intentions. The physical setting can hinder or aid the accomplishment of both internal organizational goals and external marketing goals (Bitner 1992) In this way the servicescape can be assumed a facilitating role by either aiding or hindering the ability of customers to carry out their respective activities.

The servicescape as a facilitator can also encourage and nurture particular forms of social interaction among and between employees and customers. The servicescape communicates a certain message, it differentiates in signaling the intended market segment, positioning the organization, and conveying distinctiveness from competitors.

Elements

The servicescape exist of three main dimensions. These are the ambient conditions, spatial layout and functionality, and signs, symbols and artifacts (Bitner 1992). Ambient conditions are the characteristics of the environment pertaining to our five senses. Background

(12)

12

Blodgett 1996) The third main dimensions, signs, symbols and artifacts contain the many items in the physical environment that serve as explicit or implicit signals that communicate about the place to its users (Bitner, 1992). They can be used as labels, for directional purpose, and to communicate rules of behavior. Signage can play an important role in communicating a firms’ image.

So, the servicescape consists of the total configuration of environmental dimensions. The configuration of elements can lead to approach or avoidance. Pollution is one of those elements that impacts customer behavior.

2.2 Pollution

Cleanliness of the service environment is a servicescape element in (Bitner, 1992) and of itself (Vilnai-Yavetz and Gilboa, 2010), and has a strong impact on customers’ responses in multiple service contexts. In the servicescape model cleanliness is part of the ambient conditions but it also states implicit signals about the firms’ image, and rules of behavior (Bitner, 1992). People try to avoid littered and polluted environments. Cleanliness affects customers emotions, attitudes, and behaviors, especially approach behaviors (Vilnai-Yavetz and Gilboa, 2010). In the field of environmental psychology, a littered environment shows cues of chaos and disorder, which are unpleasant to endure (Steg et al. 2012). Conclusion, a polluted servicescape will repel customers. The effect of a clean environment becomes even more important when customers’ stay endures. Wakefield and Blodgett (1996) studied the influence of cleanliness on approach behaviors, and found increased importance when the stay in the environment endures.

What is it

(13)

13

Customer behavior

Cleanliness is a servicescape has a strong impact on customers’ responses in multiple service contexts. Maintaining cleanliness can help in providing a better service and prevent service failures (Vilnai – Yavetz and Gilboa 2010). Cleanliness receives much less attention than other ambient factor like color, noise and odor (Vilnai – Yavetz and Gilboa 2010). But neglecting pollution has its downturn. Uncleanliness was found to be one of the most influential irritants in a shopping environment (d’Astous 2000) and organizations with the highest cleanliness problems had the lowest retention rates (Hofman et al. 2003)

Cleanliness can also act as a motivator, it signals service quality. Customers associate cleanliness with service quality (Wakefield and Blodgett, 1996) It acts as a predictor and sets expectations according the disconfirmation paradigm (Oliver, 1980) in which a cleaner than expected environment will result in more satisfaction and higher service quality and vice versa. Cleanliness predicts if the organization is successful or not (Nguyen and Leblanc, 2002). Cleanliness influences customer behavior. Wakefield and Blodgett (1996) provided a construct to predict customer behavior. Through quality perception and satisfaction they were able to predict behavioral intentions as desire to stay.

Consequences of pollution

People’s behavior concerning littering has been studied in environmental psychology for a long time. When it comes to littering, which can be seen as inappropriate behavior, the applicable norm or rule of that moment plays an important role. Wilson and Kelling (1982) wrote about the so called “broken window” effect. It stated that when a neighborhood is not cleaned regularly and broken or decayed parts like windows are not repaired quickly enough, the neighborhood will enter a negative spiral of pollution, vandalism and criminal behavior. The Broken Window Theory suggests that a setting with disorder triggers disorderly and petty criminal behavior. The ‘broken window’ is the violated norm of, in this case, non committing vandalism. suggests that a setting with disorder triggers disorderly and petty criminal

behavior. When this rule is broken other seem to be triggered to act comparing. Cialdini, Reno and Kallgren (1990) according, executed research on norms in littered

(14)

14

triggered littering behavior. This littered environment constituted a descriptive norm that one other had already violated the norm of non littering. Participants were more likely to follow this behavior. It creates a certain “if a lot of people are doing this, it’s probably a wise thing to do” feeling. Interesting though was the finding that in a slightly littered environment people were less likely to litter than in a totally cleaned space. This was described to the effect of making the norm salient to the people in the environment. When they experienced one scrap of litter it became salient that the injunctive norm of non littering had been violated.

Keizer, Lindenberg and Steg (2008) performed a study about the effect of disorder on opportunistic behavior. According to the Broken Window Theory a littered environment, shows signs of disorder, which result in more littering, theft and other opportunistic behavior. They performed several tests in which the environment differed in being polluted with litter and graffiti and its influence on the behavior of the participants. Participants were more likely to litter in polluted environments, 33% over 69%. They also tested the effect of pollution, on other undesirable behaving, like stealing. Participants stole more if the opportunity was offered, in a polluted environment with litter and graffiti. This effect was less than on littering thought significant. The polluted environment weakened their concern for appropriateness and strengthens the goal to do what makes them feel good (for example, by being lazy and

throwing paper on the street) or the goal to gain resources (say by stealing). This makes it clear that people behave different in a polluted environment and even show social undesirable behavior. Even when people know it is not right, the goal to act appropriately is weakened when people observe that others seemingly did (or does) not pursue the goal to act

appropriately. So when people observe that others violated a certain social norm or legitimate rule, they are more likely to violate other norms or rules, which causes disorder to spread. So in conclusion, it’s not only a hygiene concerns if the environment becomes polluted. People and for this case, customers, will execute more undesirable behavior and the environment will enter a negative spiral of decay and undesirable behavior.

Conclusion

Disordered places, by cause of e.g. litter, have negative influence on peoples behavior, through fear and stress. A littered environment shows cues of chaos and disorder, which are unpleasant to endure (Steg et al. 2012) Visible signs of disorder in the neighborhood increase fear and mistrust amongst its inhabitants (Ross et al. 2001; Skogan, 1990; Taylor &

(15)

15

noisy, dirty, and run-down. Many buildings are in disrepair or abandoned and vandalism and graffiti are common (Ross and Jang, 2000) People who live in such neighborhoods, where they see a lot of disorder, have significantly higher levels of both fear and mistrust than those who live in neighborhoods characterized by social control and order (Ross and Jang, 2000)

Taken together, pollution as part of the service environment creates disorder, and disorder negatively affects people and their behavior. Yet, the role of cleanliness in the perception of service quality seems intuitively important, but as a matter of hygiene rather than a

motivational factor (Herzberg, 1966) Cleanliness of service environments has in general been considered trivial but the environment has to be clean and neat to a certain level otherwise it results to bad performance. We predict that customers will try to avoid places that have high levels of disorder. More specifically, we predict that pollution negatively affects customer behavior.

H1: Pollution in the servicescape creates negative customer behavior.

2.3 Architecture of the environment

Part of the servicescape is the aesthetics or design of buildings architecture (Bitner, 1992). Architecture is part of the spatial layout of the servicescape, mentioned in 2.1. An important element of this visual presentation is the architecture of build environment. The term

architecture will be further used as the collect of elements that define the building aesthetics of the environment. Architecture has always been a point of controversy. What is a nice building to stay in and which one is not? The build environment affects us in our daily life. New buildings and changes to existing buildings affect the quality of street life (Naser, 1994). The quality of buildings can make it attractive to enter and comfortable to stay. But on the contrary it can also make people feel unpleasant and wanting to leave. Can the quality of the building moderate the effect of pollution?

Enclosure, Complexity and Order

(16)

16

The first one is enclosure, which comprises openness, spaciousness, density or mystery. It defines the way about how survivable a place is. In other words, when entering a place, how easy is it to get an overview of the environment. The research on enclosure and related variables suggests that people prefer open space to enclosed spaces (Kaplan & Kaplan, 1989; Ulrich, 1983). It provides a safe feeling, being able to know what is surrounding you. In this light enclosure has evolutionary relations. Darwin (1887/1993, p. 89) suggested that natural selection works on mental capabilities as well as physical attributes. In this perspective, the mental capability of aesthetic pleasure is viewed as positive feedback between attention and things having survival value. If a space is enclosed, then depth of vision will be restricted, it might be easier to hide, and it will be easier for something else to be hiding (Stamps and Smith, 2002). The second variable is complexity. Complexity involves a comparison in which more independent elements, larger difference between them, and less redundancy and pattern produce greater complexity (Nasar, 1994). In relation to the environment, researchers have substituted the term complexity with diversity, or visual richness (Nasar, 1994). Kaplan and Kaplan (1989) refer to visual richness to suggest the removal of negative contents of

environmental complexity, such as litter and other factors that reduce order.

The third variable, order, refers to the degree to which a scene hangs together or makes sense (Kaplan and Kaplan, 1989). Several formal variables, including familiarity, redundancy, and compatibility may affect perceptions of order (Nasar, 1994)

Approach or avoidance through architecture

(17)

17

prefer open spaces. But too much openness also creates discomfort. (Kaplan and Kaplan, 1989; Ulrich, 1983) Those scales, openness, complexity and order, are scales that refer to recurring elements in the literature. It predicts how people react to a certain environment, whether they would approach or avoid the place. People strive to secure their lives in any possible way, they try to reduce uncertainty to a maximum and avoid unpleasant places.

More specifically, we predict that customers will be less vulnerable to pollution, which increases complexity and reduces order, being exposed to moderately complex and high ordered places. The servicescapes’ architecture acts as a moderator on the effect of pollution. So;

H2:the negative effect of pollution on customer behavior will be positively moderated by a

pleasant and comfortable architecture.

2.4 Moderators

Personal preference and attitudes towards the effect of pollution and architecture can differ. Those differences can moderate the effects of the manipulation.

Interest in architecture

Architects have different preferences in building aesthetics than layperson (Akalin et al. 2009). So when a participant is an `architect´ or very interested in architecture he or she can judge the build environment (train station) different than laypersons. This difference in judgments can have its effect on the hypothesis testing and is therefore included as a covariate.

Messiness

(18)

18

Use of train

Users of public transport will have a more positive attitude towards the service than non-users (Pedersen et al. 2011). Therefore the use of train will be measured to distinguish between users and non-users in their attitudes towards the train station.

Personality; extrovert vs. introvert

The complexity of environments can be received different by people. Furham and Allas (1999) issued extroversion as one of the most important qualities of people reacting to complex situations. To prevent personality differences to bias the results, extroversion is included as a covariate.

Age and Gender

The variables age and gender won’t be used as covariates. Age and gender highly correlate with extraversion and in some degree with messiness. Lehmann et al. (2012) found

correlations between age and gender in relation to extraversion. Age is negatively correlated with extraversion and females tend to score higher on extraversion.

2.5 Conceptual model

The argumentation and hypothesis mentioned above can be translated to the proposed conceptual framework. In the figure below, shows a depiction of the hypothesis and their assumed relation. Pollution in the servicescape Customer behavior - Interest in architecture - Messiness - Use of train - Personality

Building aesthetics of the servicescape

H1

(19)

19

3. Method

This chapter will describe the research methods used to provide an answer on the problem statement and the hypotheses. The research design and data collection will be outlined below. Finally, the plan of analysis will be discussed.

3.1 Participants and design

The scope of this experiment is public transport, in specific the build environment of train stations. Dutch citizens are the subject of investigation and respondents were invited through e-mail, face-to-face contact and a facebook ‘event’. To encourage participation, two movie tickets are allotted to one of the participants that has registered its e-mail address afterwards. The research has a 2 x 2 design. The conditions are a ‘modern station’ vs. a ‘classic station’ (in this research further called a ‘non modern station’) and ‘polluted vs. non-polluted’ areas. The conditions of the ‘modern station’ vs. ‘non modern station’ are chosen by comparing 9 major train stations in the Netherlands on architectural features and use. Pollution is

manipulated by bringing specific rubbish into the site. More information about the manipulation is written down in chapter 3.3.The moderating variables, messiness,

architectural interest and personality are not manipulated but are tested by specific items.

sample characteristics

A total of 182 participants have completed the questionnaire. Table 4.1 shows the distribution of participants per test group.

Pollution No pollution

Modern 42 52

Non-modern 46 42

Of the 182 participants 104 are female and 78 are male. The age of the participants ranges from 15 till 72 years. The mean age is just over 33 years and the median is 27.

3.2 Procedure

The participants will be exposed to one of the manipulated conditions. The online survey (appendix A) will equally distribute a picture of one of the four conditions. A picture has been proven to be a successful method to expose participants with a setting, and measure their behavior for this setting. The questionnaire will be distributed through Qualtrics.com. This

(20)

20

online survey tool is licensed by the University of Groningen for their staff and students. The data collected through Qualtrics.com can be downloaded for statistical analysis through SPSS. A hyperlink of the questionnaire was distributed amongst friends, family and

acquaintances by e-mail, by facebook posts and by a facebook ‘event’. People were asked to fill in the questionnaire without further information about the subject to prevent biased answers. People are encouraged to participate through a lottery amongst participants. They have an equal change in winning two movie tickets. Participants are also asked to distribute the questionnaire to their social network. This ‘snowball’ effect must increase the reach of potential participants, in order to get a better sample size and distribution. The distribution by using a facebook ‘event’ has proven to be an effective way of encouraging people to

participate. It makes it possible to personal address all your facebook-friends with a special notification. There is also the possibility to send updates and reminders afterwards.

The questionnaire will be constructed from a list of items that measure the variables. These items are taken from previous studies and have proven to be an effective measurement for the dimensions. This will be discussed more in detail in subchapter 3.4.

3.3 Manipulation setting

To test the hypotheses, two train stations in the Netherlands are selected based on the

(21)

21

3.3 Operationalization

This part will describe how the different variables are constructed and measured. The measuring scales are taken from existing literature, altered to the specific situation or developed on own insights.

3.3.1 Independent variables

Pollution

‘Pollution’ was operationalized as the presence or absence of pollution on the platform of the train station. Pollution was created through deliberately scattering rubbish on the ‘clean’ platform. To check whether the manipulation of the train station with pollution was effective, two questions were taken from the cleanliness scale of Wakefield and Blodgett(1996). They are answered on a 7-point Likert scale ranging from ‘totally disagree’ to ‘totally agree’. The

(22)

22

questions were translated into Dutch and adapted to the specific train station setting. Two other questions of the scale are not applicable through the absence of specific measured items in the setting. Internal reliability is tested through Cronbach’s alpha. Cronbach’s alpha has to be above 0,6 to prove sufficient internal consistency (Malhotra, 2010). The two questions have a very high cronbach’s alpha of 0,97. To prevent participants to become aware of the pollution when answering the questions about the dependent variables, the manipulation check is performed afterwards.

Design aesthetics

Complexity and order of the modern and non modern train stations is measured using a scale based on the complexity scale of Kaplan, Kaplan and Wendt (1972) This scale contains explicit questions about how intricate and complex a setting is, answered with a 5 point likert scale. Based on these questions a dutch translation is complemented to a total of eight items. To improve consistency in questioning through the questionnaire, a 7 point likert scale is used. The scale ranges from 1=strongly disagree to 7=strongly agree. The scale was expanded with 6 questions based on own insight, which brought the scale to 8 items to measure

complexity and order. Items, 1,2,3 and 8 measure complexity, items 4 till 7 measure order. The 8 item scale proved to have insufficient internal reliability (α < .5). Through factor analysis (Appendix B) the 8 items were divided into 4 items which are renamed to order_total which has a Cronbach’s alpha of 0,838. Two other items, which asked explicitly inquire complexity, are recalculated into com_total with a Cronbach’s alpha of 0,909. The remaining two questions which inquire how much there is to see are left out because of little explained variance The manipulation check will be performed in chapter 3.5.

3.3.2 Dependent variables

(23)

23

Desire to stay

The first scale to measure customer behavior is by desire to stay (Wakefield and Blodgett, 1996). This two item scale, answered on a 7-point Likert scale, ranges from 1=strongly disagree’ to ‘strongly agree’. is translated from the original in to dutch. The original scale had a cronbach’s alpha of α = 0.75. The internal reliability for this questionnaire was tested with cronbach’s alpha and showed a high result with α = 0,87.

Quality perception

The second scale to measure customer behavior is also originated from Wakefield and

Blodgett (1996) The original scale to measure quality perceptions of the building (α = 0.76, N = 3) is translated into Dutch and altered to the situation. The original scale exists of a 7 point, bi-polar scale of ‘terrible – great’, ‘much worse than I expected – much better than I expected’ and ‘not all what it should be – just what it should be’ and starts with the statement; ‘The overall quality of the facility is:’ Cronbach’s alpha has improved in this questionnaire from α = 0,76 to α = 0,91.

Satisfaction

The third customer behavior scale measures satisfaction directly. This scale again originated from Wakefield and Blodgett (1996) and contains two items to measure satisfaction.

The original scale holds two questions which are answered by a 7 point, bi-polar scale (α = 0.8). The question starts with the statement ‘The overall feeling I get from this facility:’ Two additional questions have to be answered on a bi-polar scale being; ‘is dissatisfaction – is satisfaction’ and ‘puts me in a bad mood – puts me in a good mood’. The scale is translated into dutch and altered to the specific situation. Cronbach’s alpha is performed and showed α = 0,92 which is again higher than the original scale.

3.3.3 Covariates

To correct for personal preferences, covariates are taken into the test to prevent for biased groups within the sample. Three covariates were taken as to be the most important; ‘interest in architecture’, ‘messiness’, ’personality’ and ‘use of the train’.

Interest in architecture

(24)

24

architecture’, is derived from the ‘the personal involvement inventory’( α = 0.90) by

Zaichkowsky (1994). This scale measures participants’ involvement on a certain subject. The original scale exist of 7 point, bi-polar items. The scale is translated into dutch and altered to ten statements which are ranging from ‘totally disagree - totally agree’. Internal reliability was measured with Cronbach’s alpha and scored α = 0,93 which is more than sufficient (α > 0.6)

Messiness

The extent in which a person is messy and disorganized, here called messiness, can be of influence in how people perceive the independent variable of pollution (Lui et al. 2012). No fitting scale was found in the literature and therefore a new scale is developed. This scale measures messiness through 7 statements. The items are answered one a 7-point likert scale ranging from ‘strongly disagree’ to ‘strongly agree’. The whole scale is set up in Dutch (table 3.6 contains the scale and its items in English). Cronbach’s alpha proved the scale to have sufficient (α > 0,6) internal reliability with α = 0,84.

Personality; extrovert vs. introvert

Complexity of the manipulated environments will be received different by participants.

Furham and Allass (1999) issued extroversion as one of the most important qualities of people reacting to complex situations. Differences in personality are measured through the EPQ-R (Eysenck & Eysenck, 1991) which has its origin in the Eysenck Personality Inventory (1975) scale. A dutch translation by the University of Groningen (1995) is copied for this

questionnaire. The EPQ-R personality test consists of four personality scales, ‘Psychoticism’, ‘Extraversion’, ‘Neuroticism’ and ‘Lie’. For this test the scale Extraversion is extracted to measure the extraversion levels of participants. Questions have to be answered with ‘Yes’ or ‘No’. The amount of questions answered positive rate the participants’ level of extraversion. Cronbach’s alpha on internal reliability proved the scale to be sufficient with α = 0,81.

Use of public transport

(25)

25

3.4 Demographics

To measure differences in age and gender, two items are added. Participants choose between being Male or Female. Age is measured with a sliding bar that has to be moved from 0 to their own age. Gender and age will not be used in the analysis because they highly correlate with personality (Lehmann et al. 2012).

3.5 Measurement overview

The measurement overview shows the variables and their constructs. The first column contains the name of the variable. The second column holds the items that are part of the construct. The third column shows the Cronbach’s alpha for the internal consistency of the constructs. The fourth column holds the scale on which the items are measured. Finally the fifth column contains the source of the constructs.

Variable Questions / items Cronbach's alpha (α)

Scale Source

independent variable Pollution

1. This facility maintains clean walkways and exits

2. Overall, this facility is kept clean

0,97 7 point Likert, Bipolar Totally disagree - totally agree

Wakefield & Blodgett (1996) Design Aesthetics Order Questions: 4, 5, 6 and 7 Complexity Questions: 1 and 2

1. I think this station is complicated

2. I think this station is complex 3. I think there is much to see in this picture**

4. I think everything looks structured and well organized* 5. I think this station looks clear.

6. I think this station is a coherent whole. 7. This station comes chaotically across*

8. I think there is much to see in this station**

Order N=4, 0,838 Complexity N=2, 0,91

7 point Likert, Bipolar 1. Totally disagree 2. Disagree 3. Somewhat disagree 4. Neutral 5. Somewhat agree 6. Agree 7. Totally agree Kaplan, Kaplan and Wendt (1972) Own questions / scale Dependent variables Desire to stay

1. I enjoy spending time at this facility

2. I like to stay at this facility as long as possible

0,87 7 point Likert, Bipolar Totally disagree/totally agree

Wakefield & Blodgett (1996) Quality perception

The overall quality of this facility is:

0,91 7 point Likert, Bipolar 1. terrible – great

2. much worse than I expected – much better than I expected 3. not at all what it should be – just what it should be

Wakefield & Blodgett (1996) Satisfaction

The overall feeling I get from this facility:

0,92 7 point Likert, Bipolar 1. is dissatisfaction – is satisfaction

2. puts me in a bad mood – puts me in a good mood

Wakefield & Blodgett (1996)

(26)

26 Covariates Interest in architecture 1. I think architecture is important

2. I think architecture is boring* 3. Architecture is not relevant to me*

4. I think architecture is exiting 5. Architecture has no importance to me* 6. I think architecture is attractive 7. Architecture fascinates me 8. I attach no value to architecture* 9. I am involved in architecture 10. I think architecture is not needed*

0,93 7 point Likert, bipolar Totally disagree - totally agree

Ziachkowsky (1994) Messiness

1. People think I am messy* 2. Others think my house/room is messy*

3. I think I am a neat person 4. I really dislike mess

5. When I see a mess, I have the tendency to clean it

6. I summon others when they make a mess.

7. Everybody should clean up its own mess

0,84 7 point Likert, bipolar Totally disagree - totally agree Own questions / scale Personality

1. Are you a talkative person? 2. Are you rather lively? 3. Do you enjoy meeting new people?

4. Can you usually let yourself go and enjoy yourself at a lively party?

5. Do you usually take the initiative in

making new friends?

5. Can you easily get some life into a rather dull party? 6. Do you tend to keep in the background on social occasions?*

7. Do you like mixing with people?

8. Do you like to plenty of action and excitement around you?

9. Are you mostly quiet when you are with other people?* 10. Do other people think of you as being very lively?

11. Can you get a party going?

0,81 Bipolar, Yes or No Eysenck & Eysenck, 1975

Use of train How often do you use the train? n/a 1. Less than once a year 2. Less than 12 times a year 3. 1-3 days a month 4. 1-3 days a week 5. 4 or more days a week

Pedersen et al. (2012)

Demographics

Age What’s your age? n/a 0 – 100 years _n/a

Gender What’s your gender? n/a 0. Female

1. Male n/a

(27)

27

3.6 Manipulation check

Manipulated variables will hardly represent the precise concept a researcher has in mind (Perdue and Summers 1986). Therefore the manipulation check is performed. The

manipulation checks are conducted in order to check whether participants perceived the build environment as being different in complexity and order (modern/ non-modern) and whether they have perceived the place as being polluted (polluted / non-polluted).

Complexity and order

The manipulation was tested with the variables order_total and com_total to see if manipulation had the intended effect. Table 4.2 shows the outcome of the independent samples t-test for both variables. The manipulation proved to be effective.

Non-modern Modern Sig.

Order 4,93 4,12 0.000

Complexity 5,57 4,60 0.000

Pollution

The experienced pollution by the participants is tested through a two items scale (Cronbach’s α=0,973, n = 2). An independent samples t-test shows a difference of 5,93 for the polluted environment against 1,99 for the non-polluted setting (sig. 0,000).

Non-polluted Polluted Sig.

Pollution 1,99 5,93 0.000

Tabl e 3.6.1 Manipul at ion check of order and compl exity

(28)

28

3.7 Plan of analysis

In order to test the conceptual model, ANCOVA analyses will be performed. However, before the data is ready for the analysis some preparations need to be done.

First, all items are formulated in the same direction yet. Those items have to be recoded in the opposite direction. The recoded items are marked with a * in table 3.5

Secondly, to test the scale validity, the Cronbach’s alpha’s for all constructs are calculated. To prove sufficient, the score of cronbach’s alpha has to be above 0.6 (Malhotra, 2010). When constructs proved to be non sufficient, additional analysis is performed to increase Cronbach’s alpha by regrouping or deleting items. Results can be found in chapters 3.4 and an overview can be found in chapter 3.5

Subsequently, after the items are recoded and tested on validity, the items can be computed into variables. The new variables are calculated by taking the mean of all remaining items per construct for every participant. After computing the variables the manipulation check is performed through an independent samples t-test. The T-test defines whether there are significant differences in the mean on the variables pollution and complexity and order. The manipulation check is performed to make sure participants have perceived the manipulation correctly. Outcomes of the manipulation are written in chapter 3.5

Furthermore, the results of the experiment will be analyzed by using ANCOVA. A two way ANCOVA is used for analyzing the variance between two independent variables and their effects on a metric dependent variable. The independent variables are both categorical dummy variables that represents the existence of pollution on the one side and the type of architecture on the other side. The dependent variables constitute customer behavior through desire to stay, quality perception and satisfaction. The analysis will test whether there exists some difference in customer behavior for clean or polluted environments and in modern or non modern architecture, while taking into account the covariates interest in architecture,

(29)

29

4. Results

Table 4.2 gives an overview of the differences in the dependent variables of satisfaction, perceived quality and desire to stay. These outcomes are the means calculated from the 4 different settings before analysis.

Settings Satisfaction Perceived Quality Desire to stay

Non-polluted / non-modern M = 4.98 M = 4,71 M = 3,67

Polluted / non-modern M = 3,41 M = 3,21 M = 2,64

Non-polluted / modern M = 4,11 M = 3,95 M = 2,55

Polluted / modern M = 3,17 M = 3,11 M = 2,00

The ANCOVA is performed with the upcoming variables, see table 40.2

Variable Function

Pollution Independent variable

Architecture Independent variable

Train use Independent variable*

Interest in architecture Independent variable*

Messiness Covariate

Extraversion Covariate

Satisfaction Dependent variable

Perceived Quality Dependent variable

Desire to stay Dependent variable

* covariate which became IV after violation homogeneity of slopes Tabl e 4.0.1 Overview means scores dependent variabl es in th e four condit io ns

(30)

30

4.1 Homogeneity of slopes

In the first run, the covariate ‘use of train’ and ‘interest in architecture’ point out to be significant. This violates the assumption of homogeneity of slopes so the conditions are not sufficient (appendix B). To cope with this problem, first the covariate of ‘interest in

architecture’ will be transformed into an independent variable. The covariate was measured on an ordinal 7 point likert scale. Through calculation of the median (5,30), the variable is split into a dummy variable. The lower half of the group will be accounted for ‘0’ which hold the group who have no interest in architecture. The top half will be accounted with ‘1’ which hold the group who have an interest in architecture. By recalculating the covariate into a nominal independent variable some information is lost. Looking to the data, some participants score above 4 (neutral) on ‘interest in architecture’. Because the median cut off the group in two, we assume group 0 ( <5,30) as low interest and group 1 (>5,30) as high interest in architecture. This group with high interest in architecture can be associated with the ‘architects’ in chapter 2 who value building different than ‘non-architects’.

To find out if the assumption of homogeneity of slopes is still violated, the ANCOVA homogeneity of slopes test is run again. Now with ‘interest in architecture’ as third

independent variable (appendix C). Still, ‘use of train’ was significant in its interaction with the independent variables. ‘Use of train’, which is 5 point likert ordinal scale, is also recoded into a dummy variable (median = 2,00). These are the participants with use of train =1 or 2, which are ‘less than once a year’ en ‘less than 12 x a year’. 40 participants used the train less than ones a year and 59 of the participants used the train less than 12 times a month, those two groups use the train about once a year and we will interpret this as ‘non train users’.

The ANCOVA test is than run again with ‘train use’ also as an independent variable (appendix D). Finally, the homogeneity of regression (slopes) assumption is no longer violated. None of the existing covariates do significantly interfere with the independent variables. See for details, appendix 4.1 The list of Independent Variables is complemented with the two new variables, interest in architecture (arch_dummy) and use of train

(31)

31

4.2 Main analysis

With the precondition of the assumption of homogeneity of slopes set. And the

transformation of the both interfering covariates of ‘use of train and ‘interest in architecture’ into independent variables, all circumstances are right to perform the ANCOVA analysis (appendix E). Table 4.3.1 contains the outcomes of the main effect, two-way and three-way ANCOVA for the three dependent variables. Dependent variables that proved to be significant are printed bold.

Overview of between-subjects effects

Desire to stay Quality perception Satisfaction

Source F. Sig. F Sig. F. Sig.

modern 20,344 ,000* 4,950 ,027* 7,611 ,006* vuil 20,115 ,000* 48,013 ,000* 46,427 ,000* train_split ,014 ,906 ,669 ,415 ,035 ,851 arch_dummy 4,263 ,041* ,274 ,602 ,093 ,761 Mess_total ,028 ,867 ,165 ,685 ,175 ,676 Pers_total ,984 ,323 ,403 ,526 ,221 ,639 modern * arch_dummy 2,300 ,131 4,799 ,030* 3,556 ,061** train_split * arch_dummy ,000 ,986 1,407 ,237 2,139 ,145 vuil * arch_dummy 3,164 ,077* ,001 ,981 1,125 ,290 modern * train_split ,182 ,670 1,420 ,235 2,013 ,158 modern * vuil 1,123 ,291 2,637 ,106*** 1,768 ,185 vuil * train_split ,110 ,740 ,050 ,823 ,433 ,512

modern * train_split * arch_dummy ,089 ,766 ,839 ,361 ,215 ,643

modern * vuil * arch_dummy ,129 ,720 ,027 ,870 ,551 ,459

vuil * train_split * arch_dummy ,021 ,885 ,579 ,448 1,596 ,208

modern * vuil * train_split 2,570 ,111 4,015 ,047* 4,704 ,032*

* Sig.< 0,05 ** Sig.< 0,1 *** Sig. < 0.11

The confidence interval is set at 95%, in order to be significant, the condition p < .05 have to be fulfilled. When effects did not meet the condition of p < .05, but are under or close to .1 these effects are also mentioned. This chapter is structured following the order of ANCOVA’s on following desire to stay, satisfaction and quality perception.

(32)

32

4.2.1 Effect of Pollution

Hypothesis H1: Pollution in the servicescape creates negative customer behavior.

tests the effect of pollution on customer behavior with the covariates personality and messiness. Pollution proved to be significant on all three dependent variables, the results are listed below. The tables provides the adjusted means on the dependent variables for each group. Adjusted refers to the fact that the effect of the covariate has been statistically removed.

Desire to stay

The table and graph show the effect for pollution on desire to stay. Desire to stay decrease from 3.149 point to 2,295 on a 7 point likert scale when exposed to pollution.

The effect of pollution*interest in architecture is significant (sig.=0.77) when sig. < 0.10 The graph shows that ‘desire to stay’ decreases stronger for ‘interest in architecture = 1. It

decreases from 3,516 to 2,325, whereas ‘interest in architecture’ = 0, decreases from 2,782 to 2,264.

Pollution on Desire to stay

Pollution Mean Std. Error

95% Confidence Interval lower bound upper bound

0 3,149 ,135 2,883 3,415

1 2,295 ,134 2,030 2,560

Covariates: Mess_total = 4,82, Pers_total = 15,26.

Pollution * interest in architecture on desire to stay

Interest

in arch. Mean Std. Error

95% Confidence Int. Polluti

on _boundlower _boundupper

0 0 2,782 ,192 2,403 3,151

1 2,264 ,183 1,903 2,625

1 0 3,516 ,189 3,143 3,890

1 2,325 ,198 1,935 2,716 Covariates: Mess_total = 4,82, Pers_total = 15,26.

Figure 4.3.1 Pol lut ion

2,2 2,4 2,6 2,8 3 3,2 0 1 2,2 2,4 2,6 2,8 3 3,2 3,4 3,6 0 1 1 0

Tab le 4.3.2 Poll ution

Tab le 4.3.3 Poll ution * int erest in arch itect ure Figure 4.3.2 Pol lut ion * int erest in ar chit ecture

(33)

33

The table and graph show the effect for pollution on quality perception. The score decrease from 3.149 point to 2,295 when exposed to pollution.

Satisfaction

The table and the graph show the effect of pollution on satisfaction. Satisfaction decreases from 4,609 to 3,272 on a 7 point Likert scale when exposed to pollution.

Pollution on Quality perception

0 4,385 ,126 4,136 4,634

1 3,148 ,126 2,900 3,396

Pollution on satisfaction

0 4,609 ,139 4,336 4,883

1 3,272 ,138 2,999 3,545

3 3,2 3,4 3,6 3,8 4 4,2 4,4 0 1

Table 4.3.4 Pollution Figure 4.3.4 Pollution

3 3,5 4 4,5

0 1

(34)

34

4.2.2 Effect of architecture

Hypothesis H2:The negative effect of pollution on customer behavior will be positively moderated by a pleasant and comfortable architecture., tests the moderating effect of the

build environment on customer behavior. First the effect of the architecture is tested on customer behavior, as well as the covariates personality and messiness.

Desire to stay

The table and graph show the effect of the architecture on desire to stay. Desire to stay decreases from 3.147 point to 2,297 on a 7 point Likert scale when exposed to a modern setting.

The table and graph show the effect of the architecture on quality perception. The score decreases from 3,963 to 3,570 on a 7 point Likert scale when participants are exposed to a modern setting.

Architecture on desire to stay

Modern Mean Std. Error

0 3,147 ,133 2,884 3,409

1 2,297 ,134 2,032 2,562

Architecture on quality perception

0 3,963 ,125 3,717 4,209

1 3,570 ,126 3,322 3,819

2,2 2,4 2,6 2,8 3 3,2 0 1

Table 4.3.6 Modern archit ecture Figure 4.3.5 Modern ar chit ectur e

3,5 3,6 3,7 3,8 3,9 4 0 1

(35)

35

Satisfaction

The table and the graph show the effect of modern architecture on the dependent variable satisfaction. Satisfaction decreases from 4.209 to 3,679 on a 7 point Likert scale when participants are exposed to modern architecture.

The influence of modern architecture on customer behavior proved be negative for all three dependent variables. This proves the assumption that non modern architecture is perceived more pleasant and comfortable than modern architecture.

4.2.3 Effect of pollution * architecture

The previous results are all single effects of modern architecture and pollution on customer behavior. Hypothesis H2 specifically tests the effect of the build environment on pollution

through behavior. The interaction of the independent variables are listed below.

Desire to stay

The effect of pollution * modern architecture proved to be non significant with sig. 0,293 and therefore is hypothesis H2 not support for dependent variable Desire to stay

The effect of pollution * modern architecture proved to be non significant with sig. 0,106 and therefore is hypothesis H2 not support for dependent variable Quality perception. However,

the effect of pollution * architecture * use of train proved to be significant (Sig. 0,47). This was expectable because use of train showed significant correlation in the homogeneity of slopes test. See table 4.3.9 and 4.3.10 for the results.

Modern architecture on satisfaction

0 4,209 ,137 3,938 4,479

1 3,673 ,138 3,400 3,946

3,6 3,8 4 4,2

0 1

(36)

36

The table and the graphs show the differences in use of train =0 and use of train=1. For group use of train=0, Quality perception lowers when exposed to a modern setting without pollution. But in a polluted setting quality perception rises when exposed to a modern setting.

For group train_split=1, Quality perception lowers for both polluted and non-polluted when exposed to a modern setting. Therefore we can say, hypothesis H2:The negative effect of pollution on customer behavior will be positively moderated by a pleasant and comfortable architecture is supported for the group train_split=1. The hypothesis is not supported for

train_split=0 because quality perception increased in the modern setting.

Use of train = 1

Pollution * Modern architecture on quality perception

modern Mean Std. Error

0 0 4,748 ,258 4,239 5,256

1 3,538 ,244 3,057 4,020

1 0 4,210 ,277 3,663 4,756

1 2,867 ,265 2,345 3,390 Covariates: Mess_total = 4,82, Pers_total = 15,26. Use of train = 0

Pollution * Modern architecture on quality perception

0 0 4,704 ,258 4,194 5,214

1 2,862 ,243 2,381 3,342

1 0 3,877 ,213 3,456 4,299

Table 4.3.9 quality percepti on _{Table 4.3.10 quality percept ion}

2,5 3 3,5 4 4,5 5 0 1 pollution=0 pollution=1 2,5 3 3,5 4 4,5 5 0 1 pollution=0 pollution=1

(37)

37

Satisfaction

The effect of pollution * modern architecture proved to be non significant with sig. 0,185 and therefore hypothesis H2 is not supported for dependent variable Satisfaction. However, the

effect of pollution * modern architecture * use of train proved to be significant (Sig. 0,32). See table 4.3.11, 4.3.12 and figures 4.3.10 and 4.3.11 for the results.

The graphs and table show again significant differences between the groups based on variable use of train (train_split). Group use of train=0 shows a similar effect seen by quality

perception. The satisfaction score lowers when exposed to modern in a non-polluted setting, but increases when pollution and modern architecture are both present. Group use of train=1 (fig. 4.2.11) shows that both lines of pollution decrease when exposed to modern architecture. This implies that hypothesis H2:The negative effect of pollution on customer behavior will be positively moderated by a pleasant and comfortable architecture is supported for the group

use of train=1. For group use of train=0 the opposite occurs, satisfaction increases.

Pollution * Modern architecture on satisfaction

0 0 5,035 ,284 4,474 5,596

1 3,142 ,268 2,614 3,671

1 0 4,093 ,235 3,629 4,556

Pollution * Modern architecture on satisfaction

0 0 4,980 ,283 4,421 5,539

1 3,677 ,268 3,147 4,207

1 0 4,330 ,305 3,728 4,931

3 3,5 4 4,5 5 5,5 0 1 pollution=0 pollution=1 2,5 3 3,5 4 4,5 5 0 1 pollution=0 pollution=1

Table 4.3.11 use of train=0 Table 4.3.12 use of train=1

(38)

38

4.3 Additional analysis

As found in the homogeneity of slopes test, interest in architecture proved to have significant interaction with the independent variables.

Interest in architecture proved to be significant for the dependent variable Desire to stay. Desire to stay increased from 2,525 to 2,921 when interest in architecture=1. (table 4.3.1 and figure 4.3.1

4.4.1 Interest in architecture and modern architecture

The interaction modern architecture*interest in architecture is significant (Sig<0.05) for quality perception with sig.=0.030. modern architecture*interest in architecture is also significant (Sig<0.1) for the dependent variable Satisfaction (0,061).

The quality perception of participants of group interest in architecture=1 lowers from 4.205 to 3,422 when exposed to a modern setting. The quality perception of interest in architecture=0 showed hardly no difference going from 3.720 to 3,718 on a 7 point Likert scale.

Interest in architecture on desire to stay Interest

in arch. Mean Std. Error

0 2,523 1,33 2,260 2,785

1 2,921 ,137 2,650 3,192

Interest in architecture * Modern architecture on quality perception Interest in arch. Mean Std. Error 95% Confidence Int. Modern _lower bound upper bound 0 0 3,720 ,179 3,366 4,074 1 3,718 ,171 3,381 4,056 1 0 4,405 ,176 3,859 4,552 1 3,422 ,187 3,054 3,791 Covariates: Mess_total = 4,82, Pers_total = 15,26.

2,4 2,6 2,8 3

0 1

Table 4.3.13 interest in ar chitectur e Figure 4.3.11 interest in ar chitecture

3,4 3,6 3,8 4 4,2 4,4 0 1 1 0

Table 4.3.14 quality percept ion Figure 4.3.12 quality perception plot

Table 4.3.15 satisfaction scores Figure 4.3.13 satisfaction plot

(39)

39

Satisfaction scores decrease for both groups interested in architecture when exposed to modern architecture. However, interest in architecture=1 appreciates a non-modern setting more than interest in architecture=0. The reduction in satisfaction is also stronger when exposed to a modern setting.

Interest in architecture * Modern architecture on satisfaction modern Mean Std. Error 95% Confidence Int. Pollutio

n _boundlower _boundupper

0 0 3,994 ,197 3,604 4,383

1 3,828 ,188 3,457 4,199

1 0 4,424 ,193 4,042 4,805

(40)

40

5. Discussion

5.1 Discussion of findings

The aim of this research was to investigate the possible relationship between pollution and customer behavior in a train station. The results show that there is a significant difference in participants’ customer behavior between polluted and non-polluted environments. Pollution increases the level of disorder and that customers’ behavior is negatively influenced by it. This finding contributes to the findings of Steg et al. (2012) that a littered environment shows cues of chaos and disorder, which are unpleasant to endure. Subsequently, an unpleasant and arousing setting causes customer to avoid the environment (Mehrabian and Russell, 1974).

The second aim of this research was to investigate the possible moderation of the buildings’ architecture on the effect of pollution on customer behavior. The results, in the first place, show that there is significant difference in customer behavior between modern and non-modern architecture. Customer behavior is negatively influenced when customers are exposed to modern architecture. Modern train stations are tested low on complexity and moderate on order. This effect is expected because people prefer places with moderate complexity and high order (Nasar, 1994). It also implies that modern stations are perceived less pleasant and comfortable than non modern, classic architecture. Secondly, the interaction of pollution and modern architecture are tested together. The results show a significant difference between polluted environments in modern and non modern architecture. The complexity and order of the architecture moderates the effects of the complexity and (dis)order caused by the

pollution. The interaction between modern architecture and pollution is not significant. However, the interaction becomes significant when the covariate use of train is added as independent nominal variable. For the dependent variables quality perception and satisfaction, the interaction between modern architecture and pollution proves to be significant when participants were regular train users (more than once a month). Desire to stay though, is not significant. The non-modern polluted setting scores higher on the customer behavior

variables, quality perception and satisfaction, than the polluted modern setting. Those

(41)

41

We thus can conclude that for those who are regular train users, the negative effect of pollution is reduced by the architecture of a non-modern design. This research proves the moderating effect of architecture on customer behavior by pollution. It can be clarified by the degrees of complexity and order in the environment. The effect of pollution on customer behavior can also be explained through complexity and order. Through the construct of complexity and order, the effects of architecture and pollution are connected. Non-modern architecture proved to have more visual richness, a higher level of complexity. The

complexity level is influenced by the amount of order from returning patterns and

symmetries. Advanced complexity with high order can isolate disturbing effects of pollution because the proportional small increase can be intercepted by the complexity and order. In other words, in modern architecture with low complexity and high order, the pollution

disturbs more easily. It increases complexity but lowers order. This research links the findings from environmental psychology to the field of marketing, whereas pollution and architecture have proven to be predictors for customer behavior and attitudes towards servicescapes.

The covariates interest in architecture and use of train have proven to be significantly interacting with the independent variables.. The use of train was interacting with both independent variables pollution and architecture. Interest in architecture has less interaction but has its own effect. Interest in architecture showed significant differences on desire to stay. Noteworthy is the effect of participants with high interest in architecture scored .4 higher on desire to stay. Also their decline in desire to stay is stronger. This can be explained by the strengthened engagement. When participants show strong interest in architecture they also appreciated the environment more. Their commitment to a nice environment is higher. A violation of the environment is experienced stronger, and therefore the effect of diminishing customer behavior is stronger.

CLEANLINESS AND REGULARITY:

CLEANLINESS AND REGULARITY:

THE INFLUENCE OF POLLUTION AND ARCHITECTURE ON

CUSTOMER BEHAVIOR

CLEANLINESS AND REGULARITY:

HE INFLUENCE OF POLLUTION AND ARCHITECTURE ON

CUSTOMER BEHAVIOR

CLEANLINESS AND REGULARITY:

CLEANLINESS AND REGULARITY:

THE INFLUENCE OF POLLUTION AND ARCHITECTURE ON

CUSTOMER BEHAVIOR

Management summary

Preface

Table of contents

1. Introduction

1.1 Background

1.2 Problem statement

1.3 Relevance

1.4 Structure

2. Theoretical framework

2.1 Servicescape

2.2 Pollution

2.3 Architecture of the environment

2.4 Moderators

2.5 Conceptual model

3. Method

3.1 Participants and design

3.2 Procedure

3.3 Manipulation setting

3.3 Operationalization

3.3.1 Independent variables

3.3.2 Dependent variables

3.3.3 Covariates

3.4 Demographics

3.5 Measurement overview

3.6 Manipulation check

3.7 Plan of analysis

4. Results

4.1 Homogeneity of slopes

4.2 Main analysis

4.3 Additional analysis

5. Discussion

5.1 Discussion of findings