Master Thesis
A research on the influencing factors on eco-‐driving
behaviour:
The case of PostNL
CONTENT TABLE
ABSTRACT ... 3
PRE-‐FACE ... 4
1. INTRODUCTION ... 5
1.1 Workplace Pro-‐Environmental Behaviour
... 5
2. PROBLEM ANALYSIS ... 7
2.1 Sustainability and Eco-‐driving
... 7
2.2 Problem Definition
... 8
3. THEORETICAL BACKGROUND ... 10
3.1 Workplace Pro-‐Environmental Behaviour
... 10
3.1.1 Leading edge models
... 10
3.2
Additional fields of interest
... 12
3.3 Gamification
... 13
3.3
Overview of the model
... 15
4. METHODOLOGY ... 16
4.1 Sample and demographics
... 16
4.2 Measurement Instruments
... 16
4.3 Procedure
... 19
5. PRELIMINARY ANALYSIS ... 20
5.1 Why comparing PCA and CATPCA?
... 20
5.2 Why validating OCB and CLIO?
... 21
5.3 Eco-‐driving: PCA vs. CATPCA
... 21
5.3.1 Detailed analysis of CATPCA
... 21
5.4 OCB: PCA vs. CATPCA
... 22
5.4.1 Detailed analysis of CATPCA
... 22
5.5 CLIO: PCA vs. CATPCA
... 22
5.5.1 Detailed analysis of CATPCA
... 23
5.6 Reliability remaining questionnaires
... 23
5.7 Concluding remark
... 23
6. ANALYSIS ... 24
6.1 Research question 1
... 24
6.2 Analysis – Research question 2
... 26
7. DISCUSSION ... 27
7.1 Discussion on validation
... 27
7.2 Discussion research question 1
... 28
7.3 Discussion research question 2
... 29
8. MANAGERIAL IMPLICATIONS ... 30
9. LIMITATIONS AND FURTHER RESEARCH ... 30
REFERENCES ... 32
Appendix Part A -‐ Eco-‐driving ... 36
Appendix Part B -‐ OCB ... 40
Appendix Part C -‐ CLIO ... 43
Appendix Part D -‐ Research Question 2 ... 46
ABSTRACT
This research builds on a series of recent studies that have reported factors that influence workplace pro-‐ environmental behaviour (WPEB). The aim of this study is to understand which of these factors can explain the variance in eco-‐driving behaviour of chauffeurs at PostNL. These factors include self-‐concordance, descriptive norms, daily affect, organizational citizenship behaviour and transformational leadership style. A preliminary analysis was performed to validate the questionnaires used to measure eco-‐driving, transformational leadership and organizational citizenship behaviour. Second, a multiple regression analysis was used to measure the relationship between the predicting variables (self-‐concordance, descriptive norms, daily affect, organizational citizenship behaviour) and outcome variable (eco-‐driving). Results suggested that the output was substantially biased. Causes for this biased output might be found in the sample, e.g. chauffeurs misunderstood the questions, or were reluctant to fill in a questionnaire on eco-‐driving, which can both result in chauffeurs randomly assigning answers. It is suggested to find and compare alternative methods -‐other than validated questionnaires-‐ that can provide a more reliable output on predicting variables of eco-‐driving behaviour.
PRE-‐FACE
I started working on this research four months ago. I have met a lot of inspiring people, have been to many places (where there was no public transport anywhere near) and spend a lot of hours traveling by train. I had absolutely a great time working on this project and have learned a ton. I could not have finished this project without the help of others. Therefore, I would like to thank all the people who have advised and supported me during these final four months of my career at the University of Groningen.
To start, I want to thank prof.dr. J. Wijngaard in specific for his time, support and advice. Whenever necessary, I could call for advice. In addition I would like to thank N. Ziengs and prof.dr. Van Wezel for statistical advice. Moreover I want to thank prof.dr. J. Bokhorst for rejecting my first proposal, otherwise I would not have done an internship at PostNL Parcels.
From March 2014 I worked at PostNL parcels to write my thesis. Without the help, support and taxi-‐services of Kees Willem Rademakers I could not have done this. He has inspired me to research eco-‐driving behaviour of chauffeurs. With his unconditional love for the environment, he made sure that this project could be continued. I also want to thank Lucia Smit-‐Kleiman for having faith in this project. Furthermore I would like to thank the depot managers who have put a lot of effort into helping me with the questionnaires and I like to thank the chauffeurs for filling in the questionnaires. Special thanks to Gert Jan who was kind enough to pick me up in the middle of nowhere when I was lost at 6:00AM, near the depot in Waddinxveen.
Finally, I would like to thank Klaas Jan Kooiker for his unconditional support and advice. It helped me to focus and keep faith during writing my thesis.
1. INTRODUCTION
1.1 Workplace Pro-‐Environmental Behaviour
Jack is well aware of the discussion on climate change and decided to contribute to a sustainable environment, i.e. engage in pro-‐environmental behaviour (PEB). He separates waste, buys paper bags and even takes the bicycle to work. And yet -‐if Jack is stressed, late or has too much work to do-‐ he takes the car to work, even though he knows CO2 emissions are harmful to the environment. This example of Jack sounds probably familiar to most of us. We become increasingly aware of climate change but still, many of us understand the reasons why Jack took his car to work. This raises the question of how humans can be motivated to continuously engage in PEB (at work). PostNL, the Dutch Post and Package Delivery Company, is currently facing the same question, i.e. how to motivate their employees to continuously engage in PEB at work (eco-‐driving). As PostNL is aware of its environmental footprint, they want to reduce CO2 emissions. In 2012 they introduced a pilot ‘eco-‐driving’ at a depot in Waddinxveen, facilitated by an external company called VGL. Results showed that CO2 emissions and costs decreased by 12% in one year. However, savings fluctuated highly during the year (see figure 1). The question is how to motivate chauffeurs to continuously engage in eco-‐driving so that these fluctuations in savings diminish.
In literature, the need for human behavioural modification towards more PEB is well recognized by researchers. In addition they point to the need for empirical research that investigates how promoting workplace PEB (further referred to as WPEB) can be achieved (Robertson & Barling, 2013). WPEB can be defined as ‘any action taken by employees that she or he thought would improve the environmental performance of the company’ (Ramus & Steger 2000, p 606). To date, scholarship about business and the natural environment has been dominated by studies that address firm-‐ and industry level phenomena (legislation and stakeholder pressure: Gonzalez-‐Benito & Gonzalez-‐Benito 2006, media attention: Bansal 2005, organizations culture or structure: Harris and Crane, 2002; Tudor et al 2008, corporate codes of conduct and corporate guidelines: Somers 2001) with little attention paid to individual behaviours (Andersson et al. 2013). According to Unsworth et al. (2013) ‘the need to increase PEB in employees (WPEB) is readily apparent and rapidly increasing, but different attempts to describe models and interventions that improve WPEB are not working as well as theorising suggests.
on traditional motivation theory. According to Shamir (1991) and Mishel (1973), traditional motivation theories cannot be applied in the case of WPEB. Shamir (1991) and Mischel (1973) argue that WPEB is often not formally required for an employee’s job, which means that WPEB is conducted in so-‐called ‘weak situations’. Shamir (1991) and Mischel (1973) state that in such weak situation, traditional motivation theory cannot be applied. Based on this, it is not clear which theory PostNL can apply to motivate their chauffeurs to engage in eco-‐driving; those who are based on traditional or on non-‐traditional theories?
The goal of this study is to better understand which factors influence eco-‐driving behaviour of chauffeurs at PostNL. Eco-‐driving behaviour of chauffeurs can be seen as WPEB, since it as ‘an action taken by employees that she or he thought would improve the environmental performance of the company’. The effect of electronic monitoring devices will be excluded, due to a lack in resources to test this option. In addition, the effect of gamification will be tested, which is currently used as a tool to motivate chauffeurs to engage in eco-‐ driving.
In the next section I will start with a detailed problem analysis. Subsequently, the theoretical background and methodology are provided and an outline is given on how the rest of this paper is structured.
2. PROBLEM ANALYSIS
In this section I will describe eco-‐driving, which is part of PostNL’s sustainability policy, and the problem they face with motivating chauffeurs to engage in eco-‐driving behaviour.
2.1 Sustainability and Eco-‐driving
A key focus of PostNL Parcels is sustainability, as they recognise their influence on the environment. The aim of the organization is to become the most eco-‐friendly logistical company of Europe in 2017. In order to reach this goal, PostNL is working on various projects such as electric cars, delivery-‐by-‐bicycle, green-‐gas fuel and ship delivery. Another project is the one explained in the introduction, i.e. eco-‐driving. Eco-‐driving is a strategy composed of rules, such as ‘not leave the car idling’ or ‘use the breaks only when necessary’. Based on previous calculations, eco-‐driving results in CO2 emission reductions and cost savings of 12%. However, the difficult part is actually motivating chauffeurs to engage in eco-‐driving the entire year. Once chauffeurs step into the PostNL owned van, until they return, chauffeurs are as happy as the king. They do not face direct control over their driving behaviour; they do not own the vans themselves and do not pay the fuel bill. As a result, chauffeurs are not triggered to engage in eco-‐driving. The solution to make chauffeurs co-‐owners or complete owners of the PostNL vans is (for now) not a solution, nor is it to let them pay their own fuel bill. In 2013, PostNL started an 11-‐month pilot in Waddinxveen. VGL Eco-‐ drive, a local company specialized in eco-‐driving, facilitated the pilot. Interesting to note is that the owner of VGL Eco-‐drive is a former PostNL chauffeur. The aim was to motivate chauffeurs to engage in eco-‐driving in order to reduce CO2 emissions and costs. Prior to the pilot, data was collected on fuel use and CO2 emissions in order to create a benchmark per chauffeur. In addition, chauffeurs participated in an eco-‐driving training. Consequently, the pilot started in February 2013 and data was collected and compared to the benchmark. Data collection was done via the help of an application. Chauffeurs needed to register their fuel use after refuelling. Employees from the planning department controlled this registration process by checking chauffeurs each time they finished their shifts. Outcomes of the pilot are interesting (see figure 1). CO2 emission and cost reductions started slowly, then increased in the months June and July and decreased again after summer. At the end of the pilot, savings were down to zero again. Based on these outcomes, the question is raised why there are significant changes in eco-‐driving behaviour.
Figure 1: Outcome eco-‐driving pilot Waddinxveen
§ Average fuel use
2.2 Problem Definition
PostNL Parcels employs around 480 full-‐time chauffeurs, which are divided among 18 depots in the Netherlands. These chauffeurs drive six million kilometres per year and as a result use 800.000 litres of fuel, which costs 900.000 euro’s yearly. Based on these figures, CO2 emissions are 2.0 million kg per year. In 2013, PostNL implemented an eco-‐driving pilot. The outcomes showed a reduction in fuel and CO2 emissions of 12% in one year. However, outcomes were interesting. Cost savings started slowly, then increased during June and July after which it decreased again to 0. This leads to the following research question:
RQ1: ‘ Which factors can explain the variance in eco-‐driving behaviour of PostNL Parcel chauffeurs?’
In an attempt to motivate chauffeurs to engage in eco-‐driving behaviour, PostNL introduced a gamification tool: the Drivers Challenge and Drive Me Challenge. The Drive Me Challenge is only held with PostNL chauffeurs from both the Parcel and Postal department. The Drivers Challenge is held with other companies such as ANWB, ENECO, Ministry of Milieu and Infrastructure, BAM and Siemens. Every company can let two of their best-‐performing chauffeurs participate in the Drivers Challenge. For the Drive Me Challenge, ten best-‐ performing chauffeurs of PostNL are invited. In case both challenges are held (which is not always the case), the two best performing PostNL chauffeurs at the Drive Me Challenge can participate in the Drivers Challenge. Different games aimed at reducing CO2 emissions, car damages and costs are held at the Challenges and the winner takes home a flat screen television. The reason for organizing the Drivers/Drive Me Challenge is the assumption that games and the possibility to participate in these games will motivate chauffeurs to drive more efficiently.
RQ2: ‘To what extent is the Drive Me Challenge an effective tool to motivate PostNL Parcel chauffeurs to engage in eco-‐driving behaviour during the entire year?’
3. THEORETICAL BACKGROUND
3.1 Workplace Pro-‐Environmental Behaviour
Workplace pro-‐environmental behaviour (WPEB) can be defined as ‘any action taken by employees that she or he thought would improve the environmental performance of the company’ (Ramus & Steger 2000, p 606). Eco-‐driving can be additionally defined as a strategy aimed to change a person’s driving behaviour by providing static advice (Barth & Boriboonsomsin, 2009). When chauffeurs engage in eco-‐driving, it is an action taken by an employee to improve the environmental performance of PostNL. Therefore, I argue that eco-‐driving is a form of WPEB. As mentioned before, the difficulty of an eco-‐driving strategy at PostNL are not the rules to which chauffeurs must comply. Rather, it is the motivational aspect, i.e. how can chauffeurs be motivated to actually engage in eco-‐driving. At home, many people seem to be willing to engage in PEB as long as it does not require too much time, money and effort (Abrahamse, Steg, Vlek, & Rothengatter, 2007), but motivating employees is a different story. At home, people face a personal incentive for energy conservation: individual reductions in gas and electricity use will eventually be reflected in a lower energy bill. Within the work environment, such motivational forces are lacking. Employees jointly contribute to overall organizational energy consumption and typically have no way of seeing how their individual conservation behaviors (e.g. eco-‐ driving) can contribute to the organization’s overall goal of energy conservation (Bolderdijk, Steg & Postmes, 2013). To understand how employees are motivated to engage in workplace PEB, new insights are needed.
3.1.1 Leading edge models
As it is relatively new to describe motivational/influencing factors related to workplace pro-‐environmental behaviour, there is only a limited set of articles that provide valuable insights. Andersson’s et al (2013) identified six articles in their paper that illustrate an array of novel theoretical and empirical approaches that comprise the leading edge of research of WPEB. I will follow his lead by describing 3 out of the six articles. The other three articles, written by Delmas and Pekovic (2013), Walls and Hoffman (2013) and Bolderdijk et al (2013) will be excluded based on their inapplicability in the case of PostNL. Delmas and Pekovic (2013) describe the relationship between adopting environmental management standards and labour productivity, which is less relevant for the case of PostNL whereas Walls and Hoffman (2013) focus on the relationship between aspects of the institutional environment and organizational environmental actions. The paper written on electronic monitoring devices by Bolderdijk et al (2013) is excluded since PostNL is currently not monitoring their chauffeurs with electronic monitoring devices, which makes it impossible to test for (due to time restrictions)
Descriptive Norms: Robertson & Barling (2013) develop and test a model that links environmental-‐specific
control and results in motivation to engage in the activity that is the target of the passion (Robertson & Barling, 2013). Descriptive norms refer to how most people behave in a situation. Descriptive norms motivate both private and public actions by informing individuals of what is likely to be effective or adaptive behaviour in that situation (Cialdini et al, 2008). When individuals follow the lead of others, they speed up the decision-‐making process such that time and cognitive effort are saved, while appropriate behavior likely results (Goldstein, Griskevicius, & Cialdini, 2007). This theory is based on the social comparison theory developed by Festinger (1954). In the environmental context, environmental descriptive norms provide information that pro-‐ environmental behaviors are effective and adaptable in the given context, and they have been shown to have powerful effects on pro-‐environmental behavior (Robertson & Barling, 2013). An example includes the study of Hotel Patrons by Goldstein et al (2007) in which messages containing a descriptive norm (e.g. ‘join your fellow guests in helping to save the environment’) had a significantly greater effect on towel reuse than did messages containing pleas to protect the environment. In addition, Goldstein et al (2008) demonstrated that the effect of descriptive norms is even greater in case people reference the behavior of similar others. Messages describing similar situations (e.g. ‘others who also stayed in the same hotel room reused their towels) resulted in higher rates of towel reuse among hotel patrons that did messages that were describing less similar situations. In this study I am interested in whether descriptive norms of chauffeurs will increase their eco-‐driving behaviour, i.e. do chauffeurs have similar others who engage in eco-‐driving behaviour and does this lead to engagement in eco-‐driving behaviour of the chauffeurs?
Transformational Leadership: In addition, Robertson & Barling (2013) highlight the positive effect of
transformational leadership style. They argue that transformational leadership includes four behaviours: idealized influence, inspirational motivation, intellectual stimulation, and individualized consideration, each of which can be applied to influencing environmental sustainability within organizations. Applied to the case of PostNL, I am interested in whether transformational leadership style is recognized as such by chauffeurs and whether it positively influences eco-‐driving behaviour.
Daily Affect: Another example is from Bissing-‐Olson et al. (2013). They found that daily positive affect and a
pro-‐environmental attitude are positively related to daily WPEB. Daily positive affect can be subdivided into unactivated and activated positive affect. Unactivated positive affect includes feelings of contentment, being at rest, and feeling relaxed, whereas daily-‐activated positive affect involves feeling excited, euphoric and enthusiastic (Bissing-‐Olson 2013). Results showed that the more employees felt calm, relaxed, and content, the more they carried out their required work tasks in environmentally friendly ways. In the case of PostNL, this might be an interesting factor. A high workload per chauffeur can result in the opposite of being calm and feeling relaxed, which might influence a chauffeur’s eco-‐driving behaviour.
Self-‐Concordance: Unsworth et al. (2013) present a model of psychological conditions underlying WPEB and
thus not expressing any higher order value, then it will not be connected to the higher order goals in the employee’s goal hierarchy. The goal hierarchy states that goals operate within a system with higher order, abstract, long-‐term goals (i.e. values) at the top of the hierarchy and day-‐to-‐day goals at the bottom. Klein et al., (2008) state that goal activation depends upon expected utility as a key determinant, which is a combination of the goals efficacy and attractiveness. Unsworth et al, (2013) argue that the success of an intervention or method will be affected by the employee’s perception of his or her ability to achieve the pro-‐ environmental goal (efficacy) and the degree to which he or she values that goal (attractiveness). The perception of efficacy will be affected by the characteristics of the intervention. The perception of attractiveness will result from an interaction between the intervention and the employee’s initial perception of the self-‐concordance of the pro-‐environmental behaviour. In the case of PostNL, this means that chauffeurs would only engage in eco-‐driving if eco-‐driving activates a higher order goal.
3.2 Other fields of interest
In addition to the leading edge models related to WPEB, there are additional fields of knowledge that are considered important for eco-‐driving behaviour by PostNL Parcels, which is Organizational Citizenship Behaviour.
Organizational Citizenship Behavior (OCB) was first termed as such in 1983 by Organ et al. Drawing on Chester Barnard’s concept (Barnard, 1938) of the “willingness to cooperate,” and Daniel Katz’s (Katz, 1964; Katz & Kahn, 1966, 1978) distinction between dependable role performance and “innovative and spontaneous behaviors,” Organ (1988: 4) defined organizational citizenship behaviors as “individual behavior that is discretionary, not directly or explicitly recognized by the formal reward system, and that in the aggregate promotes the effective functioning of the organization. By discretionary, we mean that the behavior is not an enforceable requirement of the role or the job description, that is, the clearly specifiable terms of the person’s employment contract with the organization; the behavior is rather a matter of personal choice, such that its omission is not generally understood as punishable.” (Podsakoff et al, 2000). At initiation, OCB was not as popular as it is today. During past years, the interest in OCB and its related concepts, such as extra-‐role behavior, pro-‐social organizational behavior, organizational spontaneity, and contextual performance has increased dramatically (Podsakofff et al, 2000). OCB can be measured by 5 original dimensions, which are altruism (ALTR), conscientiousness (CONSC), sportsmanship (SPORT), courtesy (COURT), and civic virtue (CIVIC) (Organ, 1988). Definitions of each of these dimensions can be found in table 1.
Table 1: Description OCB dimensions
Altruism Voluntary actions that help another person with a work problem—instructing a new hire on how to
Courtesy Subsumes all of those foresightful gestures that help someone else prevent a problem—touching base
with people before committing to actions that will affect them, providing advance notice to someone who needs to know to schedule work
Sportsmanship
A citizen-‐like posture of tolerating the inevitable inconveniences and impositions of work without whining and grievances.
Conscientiousness
Is a pattern of going well beyond minimally required levels of attendance, punctuality, housekeeping, conserving resources, and related matters of internal maintenance.
Civic Virtue Is responsible, constructive involvement in the political process of the organization, including not just
expressing opinions but reading one’s mail, attending meetings, and keeping abreast of larger issues involving the organization
Source: Podsakoff et al, 2000 p 518)
Podsakoff et al. (2000) argues that OCB may serve as an effective means of coordinating activities between team members and across work groups. Exhibiting civic virtue by voluntarily attending and actively participating in work unit meetings would help the coordination of effort among team members, thus potentially increasing the group’s effectiveness and efficiency. In addition, exhibiting courtesy by “touching base” with other team members, or members of other functional groups in the organization, reduces the likelihood of the occurrence of problems that would otherwise take time and effort to resolve. Secondly, the authors state that OCB may enhance an organization’s ability to adapt to environmental changes. Employees who are in close contact with the marketplace volunteer information about changes in the environment and make suggestions about how to respond to them, which helps an organization to adapt. Second, employees who attend and actively participate in meetings may aid the dissemination of information in an organization, thus enhancing its responsiveness. Finally, employees who exhibit sportsmanship, by demonstrating a willingness to take on new responsibilities or learn new skills, enhance the organization’s ability to adapt to changes in its environment. Applied to the case of PostNL, this means that a high level of Organizational Citizenship Behavior would result in chauffeurs engaging in eco-‐driving behavior as they (or some) volunteer information about changes in the environment to which chauffeurs can adapt (via eco-‐driving). In addition, time is not wasted to solve coordination problems and thus can be spend on issues such as eco-‐driving. Moreover, team meetings and activities are more effectively coordinated which might result in providing more information about eco-‐driving issues, which results in eco-‐driving behavior.
3.3 Gamification
figure 2). This approach can help companies to introduce gamification in a systematic way. The model includes the possibility that an alternative strategy (one that does not include the use of games) is a better solution.
It falls outside the scope of this study to research whether gamification is the best option to increase awareness on eco-‐driving behaviour among chauffeurs at PostNL or whether there are alternative solutions that result in better outcomes. However, it will be researched whether the Drive Me/Drivers Challenge in its current form creates awareness and motivates chauffeurs to engage in eco-‐driving behaviour throughout the year.
3.3 Overview of the model
In sum, based on the theoretical framework, I hypothesize that eco-‐driving behaviour can be greatly explained by a chauffeur’s descriptive norms, daily effect, self-‐concordance, transformational leadership and organizational leadership behaviour. Gamification is not included in the model. The reason for not including gamification is the fact that it is already implemented by PostNL as a solution to motivate chauffeurs to engage in eco-‐driving behaviour. The aim of this study is to understand which factors can explain the variation in eco-‐ driving behaviour of chauffeurs before any solution is implemented. Therefore, gamification will be excluded from the model. As the Drive Me/Drivers challenge is not widely known among the chauffeurs, and in addition, is not communicated extensively, it will not have a (large) influence on the outcomes. Therefore, the conceptual model can be presented as follows:
4. METHODOLOGY
This section covers the methodology in which I discuss the sample and demographics. In addition, I explain the measurement instruments that are used to test the conceptual model. Finally, I will elaborate on the procedure.
4.1 Sample and demographics
The research questions are tested with the help of five validated questionnaires, which were combined into one questionnaire. In total there are 480 PostNL Parcels chauffeurs, divided among 18 depots. In total 71 chauffeurs filled in the questionnaire. Eight questionnaires were deleted due to incompleteness. This leads to a total of 63 respondents. Among the respondents were two female chauffeurs. See table 2 for an overview of the descriptives.
Table 2: Descriptives
Depot Frequencies Age Frequencies Experience Frequencies VGL Frequencies
Breda 22 >60 4 >20 42 1 Day 4 Halfweg 9 18-‐30 3 0-‐5 7 Yes 19 Kolham 21 31-‐40 5 11-‐15 9 No 40 Waddinxveen 3 41-‐50 18 16-‐20 5 Son Total 8 63 51-‐60 Total 33 63 Total 63 Total 63
4.2 Measurement Instruments
Demographic questions: This is part one of the questionnaire. The demographic questions include questions
on age, sex, at which depot the chauffeurs work, how many year they work as a chauffeur for PostNL (experience) and whether they followed a VGL training on eco-‐driving.
Eco-‐driving Behaviour: Because of a lack of existing measures, I created scales specifically for this study to
(always). Items specifically build for this research are provided in the appendix, part D.
Self-‐concordance: To measure the level of self-‐concordance, i.e. whether eco-‐driving behaviour of chauffeurs
activates a higher order goal of pro-‐environmental behaviour, I will use the New Ecological Paradigm (NEP) Scale (Dunlap & Van Liere, 1978; Dunlap et al., 2000), which is the most widely used measure of pro-‐ environmental attitude since it was first published in 1978 (Hawcroft & Milfont, 2010). The NEP Scale measures the extent to which a person possesses an ecological worldview (e.g., a concern for the natural environment and recognition that humans are altering natural processes in the environment) and includes 15 items (7 reverse scored) (Bissing-‐Olson et al, 2013). The NEP Scale is a highly reliable and valid measure (α = .86) of pro-‐ environmental behaviour (Dunlap et al., 2000). Cronbach α is a way to determine the internal reliability of a scale, i.e. if multiple items belong to one scale. This is tested based on the underlying correlation between different items. In this case, the 15 items together measure PEB and has an α = .86, which is high. Any value above .7 is considered reliable. Additionally, the 15 items were translated to Dutch and back translated with the help of an English Professional. Consistent with recommendations by Dunlap et al. (2000), I computed a single pro-‐environmental attitude factor (after re-‐coding reverse stated questions) by averaging the 15 items. Example items are “We are approaching the limit of the number of people the earth can support” and “When humans interfere with nature it often produces disastrous consequences.” Participants indicated their level of agreement with each item on a 5-‐point scale ranging from 1 (strongly disagree) to 5 (strongly agree).
Daily Positive Affect: I will follow Bissing-‐Olson et al. (2013) in measuring daily affect by using 16 items from
the self-‐report circumplex scale (Lasden & Diener, 1992). This affect scale assesses four sets of affective experiences, which are organized around two dimensions: hedonic valence (i.e., positive or negative) and activation level (i.e., activated/high arousal or unactivated/low arousal) (Bissing-‐Olson et al, 2013). The in total 16 items were translated to Dutch and back translated with the help of an English professional. The four items used to measure daily activated positive affect were enthusiastic, elated, excited, and euphoric (α = .89); four items measured daily unactivated positive affect: relaxed, content, at rest, and calm (α = .91); four items measured daily activated negative affect: distressed, annoyed, fearful, and nervous (α = .81); and four items measured daily unactivated negative affect: dull, tired, drowsy, and sluggish (α = .89). Chauffeurs were instructed to indicate the extent to which they felt each item on a normal day by using a 5-‐point Likert Scale from 1 (not at all) to 5 (extremely).
Transformational Leadership/CLIO: A questionnaire often used to measure transformational, transactional and
example, participation, intellectual stimulation and individual consideration. Charismatic and empowerment based leadership (combined) can be considered to measure the same as transformational leadership. As mentioned before, transformational leadership involves idealized influence, inspirational motivation, intellectual stimulation, and individualized consideration (Robertson & Barling, 2013), whereas charismatic and empowerment based leadership involves the formulation of a clear and attractive vision, give meaning to an employee’s job, be a good example, participation, intellectual stimulation and individual consideration. Since it is assumed that transformational leadership will positively influence eco-‐driving behaviour, it is also assumed that charismatic and empowerment based leadership will positively influence eco-‐driving behaviour (due to the similar descriptions). Another scale of the CLIO questionnaire is ‘transactional leadership’, which includes 6 items and mirrors behavioural aspects by which the leader provides a fair agreement to its employees. These items are centred on the social exchange process between leaders and employee. The autocratic leadership scale includes also 6 items and is focused on maintaining and protecting a leaders’ own position, i.e. making independent decisions without consulting or informing employees first. Finally, the passive leadership scale includes 4 items, which refers to leaders who try to outrun their leadership responsibilities and wait for others to take initiative. I computed a factor per leadership style by averaging the items that specifically measure this leadership style.
Descriptive Norms: I will follow Robertson et al (2013) in measuring descriptive norms with the help of 5
questions. With the five questions chauffeurs were asked whether family, friends or co-‐workers practice pro-‐ environmental behaviours, endorse environmental programs, or work for environmental organizations. Sample questions include ‘Do your co-‐workers endorse environmentally friendly programs?’ and ‘Do your co-‐workers practice environmentally friendly behaviours that you know about or have seen?’ Again, these questions are translated to Dutch and back translated with the help of an English Professional. Chauffeurs answered these questions with a ‘yes’ or ‘no’. In addition, I added two similar questions that specifically involve eco-‐driving behaviour of similar others (friend, family and colleagues).
OCB: As mentioned above, OCB is measured with 5 dimensions, namely altruism (ALTR), conscientiousness
(CONSC), sportsmanship (SPORT), courtesy (COURT), and civic virtue (CIVIC) (Organ, 1988) (for definitions, see table 1). Podsakoff, MacKenzie, Moorman and Fetter (1990) developed a 24-‐item questionnaire based on the five dimensions of Organ. Chauffeurs answered the questions on a 5-‐point Likert scale, ranging from 1 (totally disagree) to 5 (completely agree). The questionnaire has been translated and back translated from English to Dutch and Dutch to English subsequently.
Gamification: Because of a lack of existing measures, we created scales specifically for this study to measure
questions include: ‘I think the Drive Me Challenge is a great initiative’ or ‘I am more aware of my own driving behaviour due to the Drive Me Challenge’.
4.3 Procedure
Participant recruitment took place by attending depot meetings at which the chauffeurs were asked to fill in the questionnaire. In addition, at some depots chauffeurs were asked to come in half an hour earlier (before their shift started), so they could fill in the questionnaire. Chauffeurs who needed to come in half an our before their shift started were paid for this half an hour. All attending chauffeurs received the questionnaire and an envelope, which they could stick down after completing the questionnaire (to guarantee anonymity). A short introduction was given on the topic of the research and it was specifically stressed that some parts might not be directly (visibly) related to eco-‐driving behaviour. During filling in the questionnaire, the chauffeurs could ask questions if they did not understood a question.
5. PRELIMINARY ANALYSIS
In this section I will present the preliminary analysis. The aim is to validate the questionnaires that measure eco-‐driving behaviour, OCB and CLIO. I performed both a non-‐linear principal component analysis (CATPCA) and a Principal Component Analysis (PCA). The comparison between the outcomes of PCA and CATPCA are given. The test with the best outcome is explained in more detail. I will start this section by explaining the reason for performing both PCA and CATPCA. I will also describe why the validated questionnaires OCB and CLIO are validated a second time.
5.1 Why comparing PCA and CATPCA?
I performed both a PCA and CATPCA to test the construct validity and reliability of the eco-‐driving questionnaire. PCA is considered to be an appropriate analysis to perform data reduction (Fabrigar, Wegener, MacCallum, & Strahan, 1999). CATPCA is a similar but non-‐parametric analysis. In literature, there is no consensus on whether to use non-‐parametric or parametric tests to analyze Likert type data. Although Likert scales fall within the ordinal level of measurement, that is, the response categories have a rank order but the intervals between values cannot be presumed equal (Susan Jamieson, 2004), researchers frequently assume that the intervals are equal (Blaikie, 2003). According to Kuzon Jr. et al. (1996), using parametric analysis for ordinal data is the first of the seven deadly sins of statistical analysis. Kapp (1990) on the other hand finds the question whether parametric or non-‐parametric tests should be used subordinate to the sample size and distribution. Norman (2010) argues that many studies, dating back to 1930s consistently show that parametric statistics are robust with respect to violations of these assumptions. He argues that parametric methods can be used without concern for ‘getting the wrong answer’. Linting & Van Der Kooi (2012) argue that CATPCA is specifically suitable as opposed to PCA to analyze nominal (qualitative) and ordinal (e.g., Likert-‐type) data, possibly combined with numeric data. Based on the contradictory arguments, I did both PCA and CATPCA and compared the results. The test with the best results is explained in more detail (for the SPSS output, please look at the Appendix). Before I compare PCA and CATPCA, I determined the right number of components to retain with Parallel Analysis1.
1
Normally, the well-‐known Kaiser’s criterion is used as a guide to determine the number of components to retain in PCA. The criterion suggest to retain any component with an eigenvalue > 1. In general, Kaiser’s criterion overestimates the number of factors to retain but there is evidence that it is accurate when the number of variables is less than 30 and the resulting communalities (after extraction) are all greater than 0.7. The Kaiser’s Criterion is also accurate when sample size exceeds 150 and communalities after extraction is on average 0.6 (Field, 2009). In this study, the sample does not exceed 150 (sample = 63). A more complex way to determine how many components to retain is parallel analysis (Field, 2009). Horn firstly introduced Parallel Analysis in 1965. PA is a Monte Carlo simulation technique that aids researchers in determining the number of factors to retain in Principal Component and Exploratory Factor Analysis. This method provides a superior alternative to other techniques that are commonly used for the same purpose, such as the Scree test or the Kaiser’s eigenvalue-‐ greater-‐than-‐one rule (Ledesma & Valero-‐Mora, 2007). However, parallel analysis is not a well-‐known test as it is not included in standard packages such as SPSS. O’Connor (2000) explains how to use parallel analysis in SPSS with the help of a syntax and Ledesam & Valero-‐Mora (2004) explain how to interpret PA. In essence, each eigenvalue (which represents the size of the factor) is compared against an eigenvalue for the corresponding factor in many randomly generated data sets that have the same characteristics as the data being analyzed. In doing so, each eigenvalue is being compared to an eigenvalue from a data set that has no underlying factor. Factors (components) that are bigger than their ‘random’ counterparts are retained (Field, 2009). To determine the number of components to retain, I used PA.
5.2 Why validating OCB and CLIO?
Although the OCB and CLIO scales are validated and have high internal reliability, I will validate it again. There are several reasons for this. In this study I want to test whether validating the questionnaire will lead to similar constructs and thus reliable output. Both CLIO and OCB are validated with a substantially different group of people (scientists in the case of OCB and managers in the case of CLIO). Large differences between the samples on e.g. IQ level and interests might lead to a biased output. It might be the case that chauffeurs do not understand the questions well enough to provide a good and reliable answer. Therefore it is important to re-‐ validate the questionnaires. In addition, a part of the chauffeurs were ‘asked’ to come in half an hour before their shift started. Some of them were reluctant (even though they got paid), even angry, to fill in the questionnaire. They believed it was nonsense to come in half an hour earlier to fill in a questionnaire on eco-‐ driving. Finally, chauffeurs in general are reluctant to fill in the questionnaire. They do not see the point in it and some are reluctant to do anything that falls outside their daily tasks. A possible result would be that they randomly filled in the questionnaire. Therefore, it is important to re-‐check the validity and reliability of the questionnaires. Only OCB and CLIO are validated. Descriptive Norms, Self-‐Concordance and Daily Positive and Negative Affect are excluded from this analysis, as they do not consist of multiple scales (variables). However, the Cronbach’s α are given for all questionnaires to measure its internal reliability.
In the next section I will firstly compare the result of PCA and CATPCA. The test with the better results is described in more detail. This is done for eco-‐driving, OCB and CLIO. The output of SPSS is given in the Appendix.
5.3 Eco-‐driving: PCA vs. CATPCA
(For SPSS output, see Appendix Part A). To determine the right number of components, I performed a Parallel Analysis, which resulted in retaining 1 component (see table 1). Then, if CATPCA and PCA are compared, PCA leads to VAF = 43.29% (1 components, 4 items, see table 2 and 3). The reliability of the component (scales) is α = .704 (see table 4). CATPCA leads to VAF = 47,93% with a reliability of α = .78 (see table 7) In total, 6 items are used (see table 8). Based on these results, I believe CATPCA results in a better solution, due to its higher VAF and reliability of the scale. The CATPCA will be explained in more detail in the next section.
5.3.1 Detailed analysis of CATPCA
