• No results found

and those that cycle at least once a week at either end of the scale. Within the attitudinal analysis the data is divided into three levels. There are:

1. Frequent cyclists (F) who reported cycling at least once a week 2. Occasional cyclists (O) who reported having cycled in the past year 3. Non-cyclists (N) who did not report cycling

These definitions were chosen for the analysis in Chapter 4 and 5 to help draw out the differences between these groups.

As the measure of cycling frequency included both leisure and utility cycling while the measure of the intention to cycle focusses only on utility cycling analysis of those that cycle at least once a month examined which constructs appear to influence an increase in frequency, from cycling ‘occasionally for any purpose’ to cycling ‘regularly for utility purposes’.

3.2.3 Demographic Questions

Respondents were also asked questions intended to assess: Physical Activity; Socio-economic Group; Place of Residence; Cycling Behaviour; Age and Gender; Ethnicity; Children living at home; Mobility; Bicycle ownership; Car ownership; Travel Behaviour; Job type; Commute.

Details of the questions used are given in Chapter 4, where the results are also analysed.

3.3 Pilot Survey

The purpose of the pilot within the formative stage of the project was to develop both the questionnaire design and to choose an analysis method. As such the PLS-SEM method was compared against another method which would also allow the creation of an Importance-Satisfaction matrix.

For comparison the ISA method was chosen as a suitable method with which the research team had experience (Yahya, 2013) and that has been used previously to monitor the Taipei YouBike scheme (Yang, 2013). The chosen method directly asks for a level of importance and satisfaction with each factor. A gap analysis is performed

on the data, highlighting the factors with the biggest gap between importance and satisfaction. This formed the basis of one version of the questionnaire.

The ISA method was compared with PLS-SEM. Within the PLS-SEM questionnaire agree-disagree scales would be used as a proxy for satisfaction within the PLS-SEM allowing topics to be raised more naturally in a way that should make sense to non-cyclists. This formed the basis of the second version of the questionnaire. If similar patterns are seen in both surveys then this would suggest that these statements make a suitable proxy.

3.3.1 Pilot Survey Launch

A draft questionnaire was tested in a seminar of approximately 15 Academic Staff and PhD students from a Transport Operations Research Group at Newcastle University. Based on the feedback from this the draft was amended and launched as a Pilot Survey conducted in winter with cyclists at Newcastle University in November and December 2014.

Over the period of the launch N=99 responses were obtained to the online version. Additional responses (N=13) were obtained when paper versions of the survey was also distributed to staff and students from Newcastle University at an event on winter cycling. These respondents were allocated one of the two questionnaires alternately. The main changes from the draft were:

1. Statements about an individual’s specific current or potential cycling route were dropped to be replaced by more general statements about the local cycling environment as non-cyclists found them difficult to relate to.

2. The decision was taken to test two different methods in the Pilot Survey (see below). Respondents were directed randomly to one of the two questionnaires from a central website, the link to which was distributed by email to staff and students from Newcastle University and Northumbria University that had expressed an interest in cycling and sustainability.

3.3.2 Pilot Survey Results

The survey respondents were all staff or students from Newcastle University and Northumbria University who were interested in cycling strongly limiting the

3.3. Pilot Survey

applicability of the findings to the research aim, however, the pilot study provided a useful method for testing the questionnaire design and potential distribution and analysis method.

Two versions of the Pilot Survey was designed and launched. The first version was intended to pilot ISA methodology while the other was intended to pilot PLS-SEM methodology.

Due to the small number of responses it was not considered useful to test the data against national statistics. However, even without these tests, it was obvious that, due to the nature of the mailing lists and event, almost all respondents were regular

cyclists and from a similar cohort (university staff and students). While this similarity limits the ability of the pilot to explain preferences across different groups it does allow decisions to be made about the methods with fewer complications than would be possible with an equally sized sample from a broader cohort.

3.3.3 Lessons Learned from the Pilot

The main outcomes from the pilot survey related to the information gained about the suitability of the survey distribution and analysis methods. The decision on which analysis method to take forward was informed by exploring whether the results aligned with the qualitative evidence and existing literature alongside a visual comparison of the spread demonstrated between indicators across both methods. Supporting data such as relating to the survey completion such as average response time and drop-out rate were also considered.

Primary Analysis Methods:The decision was taken to focus on PLS-SEM as the primary analysis method (Section 3.8) as the pilot survey results suggested that this would both help reduce the questionnaire length required and better represent the importance of potential factors.

The Impact of Summer and Winter on Responses: Results suggested that the gritting of cycle paths in winter weather was associated with satisfaction with the local

cycling environment. It was felt that this highlighted the potential importance of temporal differences across seasons, with respondents potentially being more aware of the issues which they had experienced most recently. Thus, to reduce the influence of timing on the results, it was decided that the survey would be delivered in two waves. One wave representing autumn/winter and a second wave representing spring/summer. Weather data and automatic cycle tracker data were consulted to

determine which dates would be suitable for data collection alongside other

restrictions on timings, such as school holidays which were to be avoided because of the impact of different schedules and traffic levels during school holidays on

behaviour and perceptions.

The first wave was collected between 08/09/2015 and 28/09/2015 and the second wave was collected between 11/02/2016 and 23/02/2016.

Accessing Non-cyclists: It was obvious from the responses that the distribution methods used were only effective in targeting active cyclists. This informed the choice of distribution method used within the main study, it was decided to use an On-line Access Panel (Section 4.2.4) which allows for non-cyclists to be reached. A potential limitation within this study which arose from the difficulty in obtaining responses from non-cyclists for the pilot survey is that this limited the level to which the statements used within the main study were tested within non-cyclists potentially leading to statements which were confusing to non-cyclists not being addressed at this stage.