Cycling and transport modelling

Cross-posted from my personal website.

This linked series of posts represents an initial attempt to think through a few ideas discussed at the second Modelling on the Move seminar on 15th February.

The topic was ‘Public Health Perspectives and Transport Modelling’, and different themes ran through the day, one being the perceived failure of transport modelling to deal adequately with walking and cycling, by comparison to car and public transport trips. Here I just focus on cycling, and from my own perspective. I’d like to thank Helen Bowkett, Tim Gent and John Parkin for helpful comments; of course these posts are solely my responsibility.

Go to Part 1: Where is Cycling in Transport Modelling?

Dynamic Modelling: reality, generality, and precision

At a time when discussions in public health and epidemiology in general  increasingly grapple with ‘complexity’, as well as focusing tightly on specific areas of interest (such of transport and health), it is likewise useful to step back to more abstract considerations. A general shift from the dominance of statistical-based models, which are arguably not well suited to represent change, towards dynamic modelling (Agent Based Modelling and System Dynamic Modelling have become prominent in the literature), and the possibility of including clearly specified local contextual details represent a major advance; however, such attention to technical and contextual issues may potentially move questions about modelling itself into the background. In any case, as someone who doesn’t specialise in work on transport and health, this is where my attention is drawn…

Many years ago (decades in fact), biologist Richard Levins considered the trade-offs involved when developing a modelling strategy. Unless you’re doing exceptionally badly to begin with, trade-offs are almost inevitable.  As ‘new’ modelling methods come into our view, including methods which may appear to offer only benefits, I think Levins’ principles potentially offer useful guidance.

Models can be thought of in terms of three axes: reality, generality, and precision (this typology is not intended to be comprehensive, but rather to capture some critical aspects). In general, gains in any one area are accompanied by losses elsewhere. Briefly, the three situations described by Levins are:

i.            Sacrifice realism for generality and precision

This strategy is very familiar to me as it’s typically used in global scale climate and health impact assessments. The aim to quantify impacts as precisely as possible using general relationships that are applicable to all world regions necessitates a loss of realism; in this case, what is happening nationally or below this level is lost.  An analogous case may be modelling the general impacts of transport policies at UK-level rather than specific impacts in a particular place.

But despite the apparent limitations of this approach compared to adopting a tighter, local focus, it is not without benefits. For instance, the system dynamics operating at a ‘high’ level are likely to play a role in shaping local conditions and possibilities; viewing the system only at the local level may well miss these dynamics. From this perspective, a loss of realism in terms of the local is an advantage.

ii.            Sacrifice generality for realism and precision

‘New’ modelling methods may naturally lead to this strategy as they potentially allow the inclusion of detailed local dynamics. But the degree of detail means generality is lost; the model is only useful for the very specific circumstances being considered. Viewed temporally, the lack of generality may limit the useful lifespan of such a model: as local conditions change, or as conditions at a higher level (e.g. national) change and potentially alter local dynamics, the model may no longer represent reality well.

Another limitation is the amount of data and knowledge required to construct and parameterise such a model: to make precise estimates of the outcome of interest, precise representation of relations and variables is necessary. Often the data or knowledge, particularly for a specific local setting, are not available. In the absence of the required inputs, the model, despite looking ‘fancy’, may tell us little of use about the situation of interest.  This leads on to the third strategy…

iii.            “Sacrifice precision for generality and realism”

At least in my experience, the goal of producing precise quantitative estimates is beyond discussion: of course precise quantification is the (or at least, a major) goal!

But is this sensible? In any case, it comes at a cost. If no data or quantification of a relation is available, it is either left out or crudely represented.  In many cases, so much is left out or represented by ‘best guesses’ (maybe named as ‘expert opinion’) that the resulting quantifications are accompanied by the qualification that they should only be interpreted in terms of magnitude; that  is, in a sense, qualitatively. The very sacrifices made to achieve ‘precision’ undermine the value of that precision.

Why not then – in some models, but of course not all – relax the relentless drive for precision and instead aim simply to understand system behaviour generally? Instead of asking, ‘How much will this increase if we do that’, you can ask, for example, ‘If we do this, will this increase of decrease?’. At the very least this type of strategy can be used to get a general understanding of the system of interest, and guide attention to specific aspects that appear to be critical to behaviour.

Because of the inevitable trade-offs, a mixture of strategies seems appropriate, with which is used at any moment being guided by what we know and want to know.

The final strategy is of particular interest to me.  We’ll be discussing and developing its potential for transport and health during the second event of the seminar series. It will no doubt require some imaginative input from those who come along…

“Stakeholder” involvement in transport modelling

Recently I finished a 5-year transport policy modelling exercise, attempting to engage a wide range of policy, public health, academic and community stakeholders – with only partial success. This has led me to reflect further on the notions of transdisciplinarity and participation in transport modelling. Rachel Aldred’s recent blog provoked me into re-visiting these challenges – her identified tension between small scale participatory narratives and Big Data modelling particularly struck a chord.

People from across disciplines emphasise notions of participation and transdisciplinarity as principles for decision-making that might improve outcomes for wellbeing, environmental sustainability and fairness (Dominique Charron and her colleagues at IDRC did a great job summarising this “ecosystem health” approach last year in a developing world context). Transdisciplinarity in this context tends to mean bringing knowledge from community, policy and academic stakeholders together on an equal footing. This involves the participation of community stakeholders in developing questions and decision-making, as well as increasing citizen control over solutions.

Community participation in urban planning has long been argued as a good thing – particularly if it is empowering and involves handing over some control over decisions to communities (see for example the work of Popay). A large number of health promotion and urban planning practitioners support the notion that empowering community participation in urban planning is required to achieve successful change for equity and sustainability, partly because of the limitations of representative democracy (giving a voice to the voiceless) and partly because successful joint learning can increase the support for resulting policies. In health promotion, empowering participation is also seen as a wellbeing end in itself – by increasing skill levels and a sense of agency over people’s lives (see for example the work of Glenn Laverack, Fikret Adaman and Pat Devine).

This is all sounds great, but very few attempts have been made to review the effectiveness of participation at achieving these outcomes. Beth Milton and her colleagues at the University of Liverpool systematically reviewed some of the evidence last year. Although participatory initiatives may have positively affected empowerment, information and relationships (intermediate capacity-building outcomes) there wasn’t much evidence that social or health outcomes were improved through changes to decisions.   Others argue that a focus only on empowerment as a positive outcome ignores a more complex set of roles and responsibilities on the part of individuals and authorities. For example, authorities may devolve the “work” of decision-making to community members without providing the resources to support this. There may be different levels of participation that are appropriate for different planning circumstances (Wilcox’s ideas about effective participation are getting old but are still useful).

A number of questions and challenges emerged from my own experience using a participatory modelling method to try and improve transport decision-making in Auckland.

  1. The priorities and knowledge of “community” stakeholders were best elicited through narratives about place – at a local level, yet to influence regional or national policy, there is a need to move from this local focus to regional/national level modelling. In my experience this transition resulted in losing the engagement of community stakeholders.
  2. It’s very difficult to keep any kind of transport model transparent and easily understood by the full range of stakeholders – there are methodological and skill (or art) questions here. Although it is necessary to understand system complexity, increasing model complexity can easily reduce transparency without improving decisions.
  3. It’s extremely easy for a participatory decision-making process to disempower and disappoint community stakeholders – especially when they are located within our current  neoliberal approaches to deliberative democracy (see a good recent critique in Paul Hanson’s recent Geoforum article, Toward a More Transformative Participation) – in other words, the priorities and solutions put forward by non-governance stakeholders are often in direct conflict with the limited set of possible policies within the governance framework.

I hope these challenges provoke interesting debate on the blog and during the seminars – particularly in one of the later seminars when we will focus specifically on participatory modelling.

Modelling Worlds

In May and July I attended two modelling conferences: Modelling for Policy and (one day of) Modelling World. The former was an academic conference, organised by British Academy postdoctoral fellow Erika Mansnerus, and mainly (but not exclusively) focusing on health modelling in relation to social science perspectives. The latter was the seventh annual conference of transport modelling professionals, organised by knowledge exchange company Landor.

In some ways the events were quite different. MW was primarily a practitioner event, with academics in the minority; MfP was the other way round. MW was about transport modelling; MfP mainly attracted people from the health disciplines. But reflecting back, it seems like the different modelling worlds are not so far apart in some respects, and lessons learnt in each are relevant for transport. In thinking about this, I want to focus briefly on two talks from MfP, and then draw links to some of the debates that took place at MW.

Although people at MfP felt that social science perspectives had traditionally been excluded from health modelling, there was a sense that this was beginning to change – that, in Helen Lambert‘s phrase, modellers were becoming more concerned about ‘socio-cultural plausibility’ : in other words, models should do more to represent and address social as well as medical relationships. Charlotte Watts‘ talk provided an example of this; titled ‘Using modelling to inform HIV policy in low & middle income countries’.

Charlotte leads the Social and Mathematical Epidemiology Group at the London School of Hygiene and Tropical Medicine’s Department of Global Health and Development. She spoke about their work modelling the spread of HIV in low and middle income settings, explaining that traditionally a simple infection model based around client-sex worker interactions had been used. This focused on those individuals likely to affect more than one person, leaving out key actors – specifically, men aside from clients involved in the industry.

Conversely, social science insights suggest analysing the sex industry as a business with market forces, key actors, and power inequalities. Drawing on this broader perspective the team produced a simplified ‘pimp model’. The model suggested that the role of pimps in the spread of HIV was particularly important where there are quite low rates of infection among sex workers, often when there is high turnover.

This example suggested that incorporating a more structural approach to sex work could both provide better social science knowledge and better knowledge about disease transmission processes. Charlotte argued that epidemiological models currently lagged behind understandings of health in its socio-economic and political contexts. Epidemiological models often focus on influencing individual behaviour however, she argued, ‘we know social, cultural, economic forces shape this – alcohol, poverty, violence, rape. But in modelling all we do is go from rates of partner change to HIV transmission.’ Including wider social determinants of health is a challenge but one that modellers should seek to address’. There are examples of organisational change affecting HIV transmission, but these have not been incorporated in modelling yet.

While Charlotte made a plea for social science perspectives in health modelling, Melissa Leach (who directs the ESRC STEPS programme: Social, Technological, and Environmental Pathways to Sustainability) argued more specifically for smaller-scale and participatory modelling. Melissa talked about how science is always wrapped up in stories; narratives of change, of fear, of invasion, of safety, of stasis. These narratives matter, and while they may seem a long way removed from models, they can be closer to them than we think. The rise of big data changes the model’s gaze; shaping how models view the individuals and systems they seek to know, predict, and control.

Melissa explained how the politics of modelling shape what is produced. Hierarchies, both within and between disciplines, affect who is invited onto panels and what kinds of knowledge are produced. The ways that people, objects and places are figured within models is shaped by, and shapes, particular worldviews. Melissa gave the example of Ebola models, where the conceptualisation of ‘forests’ often draws on, and helps to reinforce, a ‘fortress’ approach to conservation that can be highly problematic in low-income contexts where people rely on forests for survival. ‘Outbreak anthropology’ by contrast seeks to build on local knowledges, including local practices targeting disease prevention. Such practices may invoke non-medical explanations but may be more effective at slowing disease transmission than attempting to impose practices lacking local cultural legitimacy.

One issue raised by Melissa’s talk, and briefly discussed later, was the extent to which we should attempt to ensure that models incorporate ‘other’ knowledges, or whether we should insist that such knowledges are used to supplement models. Will anthropologically-informed models do justice to anthropology, or do they risk excluding knowledge that cannot be so easily incorporated within a modelling framework?

This had interesting echoes in a debate at Modelling World around ‘Big Data’, although here the talk was of newly available large scale quantitative datasets rather than qualitative anthropological worldviews. Yet in both cases proponents offer the promise of incorporating new knowledges, and new data sources, to transform theory and practices. Others express caution and worry that we cannot simply transfer theory or data into modelling and expect transformation: there will also be loss. (One of my personal concerns being that ‘Big Data’ about cycling is currently very limited, so the exclusion of cycling from transport modelling might be perpetuated in new ways). The sense of disciplines being pushed was present in both settings – Luis Willumsen said at Modelling World that it was not possible to stop changes in transport modelling that have been set in train by changing data. ‘Revolution in modelling will come despite our best efforts’.

Luis’s contribution was in response to Joan Serras‘ presentation on using Big Data in transport modelling. His later presentation addressed the conceptualisation of the individual, and individual behaviour, within transport modelling. Luis said that within transport demand modelling, ‘the ideal traveller is rational, selfish, and its tastes do not change, ever.’ (Cue image of Dr. Spock!) And yet what we know about human behaviour suggests all these assumptions are dubious, particularly the last one. As Keith Buchan pointed out later, people didn’t predict the massive rise of useful and/or interesting ‘things to do’ on the train which has radically changed experiences of travelling (and their implicit costs). I am typing these words while on a train up to Edinburgh to present my work to the Scottish government, connected to email and the Internet: the journey to Edinburgh and back takes nine hours of my time but I can to some extent take my office with me.

While Modelling for Policy brought out ways in which health modelling is struggling to incorporate better (and more social) representations of behaviour, Modelling World brought out related issues for transport modelling. Questions around social structure were also there, although differently posed. Luis said that using more sophisticated assumptions about behaviour can make it harder to read equilibrium states. ‘It’s not realistic – there is no equilibrium out there – but we need it to compare like with like.’ People discussed the risk that models could just become more and more complex, losing any hope of transparency but without fundamentally improving theory.

Both conferences discussed modelling imaginaries and narratives, although from different perspectives. Modelling for Policy addressed these issues as being inseparable from models, models even being seen as in themselves narratives. (There was a recurrent debate about how a ‘model’ could be defined and separated from other tools, techniques, and approaches: with Melissa arguing that we should see less formal explanations as also being models). Modelling World did not address this so explicitly, but it was there in the discussions around the politics of modelling. For example, in response to a question from the floor that mentioned Pol Pot, Keith asked why we only seem to be able to imagine either crisis or business as usual; he saw this as blocking change, because everything can either only be terrible or as it is now.

Near the end of the day I attended, Phil Goodwin made a related but more specific point on the politics of forecasting narratives. He asked: why is it that for the last 15 years, UK traffic forecasts have continued to predict rising motor traffic levels, despite evidence suggesting this might not be the case? Phil suggested that this is related to the politics of modelling: in the years when ‘New Realism’ (the recognition that it is impossible to grow the road network to met increasing demand) was becoming established, it was in everyone’s interests to predict rising traffic levels, both the traditionalists (for whom rising traffic justified road building) and the ‘realists’ (for whom rising traffic was a wake-up call). But now, Phil argued, when it is being clearer that the continuing predictions of rising traffic serve to block change, they are finally being questioned. (For Phil, the discrepancy between DfT and TfL traffic forecasts for London is indicative of the debate over the future of traffic).

Like Phil, many people who practice or engage with transport modelling are keenly aware that modelling is performative: it enacts that which it describes. In Phil’s presentation, this happens through the forecasts encouraging the belief that motorisation is unstoppable and ever-increasing, with policy responses fearing to stop it. Similar themes came up at the Modelling for Policy event, where there was a disagreement over whether epidemiological models could be seen as performative, by comparison to economic models. Angela McLean argued that economic models drive prices and so are very different from natural science models. Mary Morgan responded that issues of prevention and control are often involved in epidemiology – so, even though ‘the flu virus cannot hear what I say’ (in Angela’s words), what the model says will still shape what people do, and so what the flu virus does. The process is different but performativity is still there.

Our seminar series hopes to raise and develop some of these themes over the course of the six linked events. This will include fundamental questions around what kinds of data and theories we need, and how transport modellers can learn from debates in other fields, such as public health. Personally, I’m particularly looking forward to exploring some of the fascinating issues raised at both events around the politics and cultures of modelling; what kinds of knowledges and narratives are promoted and silenced, and how these can where necessary be changed.