Seminar 3 – Qualitative data and transport modelling
This seminar was split into two halves. The morning saw presentations and the afternoon involved exercises using real data extracts. This write-up summarises some of the points made in the morning; material from both halves can be found here.
• If transport modelling is to make more use of qualitative data, often this will be secondary data, available through sources such as the UK Data Archive. Niamh Moore spoke about debates within the qualitative research community about the status of secondary data: for example, not all researchers agree that it is possible to analyse qualitative data isolated from the original research context. Niamh talked about potential relevant sources, such as the Mass Observations Archive where a panel of participants are regularly asked to write about specific topics, which included driving in Autumn/Winter 1987. She also introduced the STEP-CHANGE project which is using the ‘mobility biographies’ approach to explore changing experiences of transport over time.
- One particularly new area is combining datasets from different projects for analysis. What specific benefits does this bring, and what new issues does it raise?
• Mike Yearworth spoke about the potential for using qualitative data within system dynamics (SD) modelling. This approach takes a holistic view of complex systems and focuses on understanding feedback loops (reinforcing or balancing), understood as underlying outcomes of interest. Qualitative data can be used to structure SD models, by coding for, and hence theorising, relationships between key variables and through this defining the boundaries of the system. The resulting models can be qualitative or quantitative. One problem is that existing qualitative software packages encourage hierarchical coding, and hence hierarchical thinking, whereas a networked approach may actually be more appropriate.
- What is the status of the data used? Usually it’s interview data – do we assume interviewees have accurate knowledge of the system structure? And that there is ‘one’ system view?
• Edmund Chattoe-Brown introduced agent-based modelling (ABM) and when (and when not) to use it. ABM models agents (which can be people, but don’t have to be) as heterogeneous rule-followers who can learn and develop new patterns of behaviour. So when we have good evidence about micro processes (for example, from ethnographic observation), and where systems are non-linear, ABM may be useful. Challenges include many traditional methodological problems for qualitative research, including how we observe environments and again, what the status of our data is (does it really tell us about the world we’re modelling?)
- For both ABM and SD: qualitative data can often tell us new things about system structures: but how do we get this into transport modelling, where very often model structures are well established already and taken for granted (e.g. based on economic theory)?
• Paul Rosen talked about his experience of using data (from the Workplace Cycling Cultures study) within a simple model. Interview data had been coded in NVivo, and included analysis of barriers and facilitators for cycle commuting, such as ownership and type of bicycle and car, job role, household structure, distance to work, and attitudes to different transport modes. This was turned into a spreadsheet to create a decision tree that could be modelled, although because the data wasn’t collected for this purpose there were a number of missing values. This was used to develop an algorithm with the outcome variables related to cycling sometimes, always, or never. Although this did not happen at the time, the plan was to go further and develop an ABM.
- Should traditional transport modelling approaches be doing more to incorporate the kinds of factors identified by Paul as important for cycling?
- Modelling on the Move 6: cycling & transport modelling
- Seminar 5 – Participatory Modelling
- Seminar 4 – Social Theory, Transport and Energy Modelling, Friday 13th September
- Seminar 3 – Qualitative Data and Transport Modelling: Friday 12th April
- Public Health Perspectives: 15th Feb, LSHTM
- Launch Event 7th December