Don’t panic! The need for change and for curricular pluralism

Spatial science: How much respect would be enough respect?

Along with Johnston et al. (2014), I am concerned that quantitative approaches in geography have been somewhat written out of the discipline, although I am unconvinced that a brief review of a small number of geography textbooks is a sufficient basis on which to make this argument. It would be surprising if proponents of any of the many philosophical perspectives on geography did not find fault with how they are treated in the survey text (Cresswell, 2013) cited as the primary evidence for the argument. Cresswell is entitled to his opinion on the merits of spatial science, and Johnston et al. (2014) are equally entitled to disagree, as they do here.

But clearly, there is something else at work here. Johnston et al. later go on to say,

Because of such mis-representations of quantification among human geographers, however, it may be that although introductory courses in quantitative analysis are available in most degree programmes, they are not given equal intellectual credence to other components. (2014: 10)

Inside geography this is truer now than it would have been even a decade ago. The complicated ups and downs of geography’s entanglement with quantification demand some explanation. Trevor Barnes’s excellent, sustained effort to document the history of the quantitative revolution (see e.g. Barnes, 2001, 2003, 2004, 2009) is a substantial and often sympathetic contribution, which yields a highly nuanced understanding. Any responsible instructor of a contemporary survey course on geographical thought surely ought not to omit this work, nor recent related developments (see e.g. Sui, 2004; Wyly, 2009, 2011). Indeed, things have changed sufficiently – at any rate they seem to have done so outside the United Kingdom ¹ – that it is difficult now to understand how we went so quickly from the serious engagement with the value (or not) of quantitative approaches in Ideology and Human Geography (Gregory, 1979) to offhanded dismissals, such as Sayer’s, ‘One of the least interesting ways of looking at society is by demographic analysis’ (1984: 160), or the accusatory hostility of Smith’s, ‘The war against Iraq in 1990–91 was the first full-scale GIS war’ (1992: 257). It also bears saying that earlier attempts to document the quantitative revolution reveal an awkward truth that even some apparently enthusiastic ‘quantifiers’ had their doubts about the value of the enterprise (see Johnston, 1984!). There is unmistakeable relief in the autobiographical accounts of some 1960s adopters of quantitative methods when Harvey’s Damascene conversion (Harvey, 1973) allowed a return with honour to the more familiar qualitative methods of the humanities (see Gould and Pitts, 2002). In short, geography (at any rate human geography) was never entirely comfortable with quantitative methods – as we might expect given its uncertain (perhaps privileged?) status somewhere between science and the humanities. Thus, the ‘writing out’ of quantitative methods from the discipline is not a recent development, attributable to contemporary textbooks: It is much more firmly embedded in the disciplinary DNA acid than that. It makes sense to think of the quantitative revolution in geography as an unfinished revolution. ² But that it was, and remains, unfinished, is a very good thing, given the nature of geography.

Which theoretical geography, which geographical concepts? ³

I have already welcomed the emphasis that Johnston et al. (2014) place on using quantitative methods to understand localised effects on people and populations in the places where they live, work and act. This is a necessary corrective to misrepresentations of contemporary quantitative methods as retreads of 1960s attempts to construct a nomothetic, scientific ‘geometric geography’. But, in their enthusiasm to distance themselves from such a naive geography Johnston et al. (2014) commit two (lesser) misrepresentations of their own.

The first of these is to imply that geometry plays little part in their favoured analytical approaches. This implication cannot be sustained. As soon as we deploy space as a frame for data and use it to make associations either among data within a data set (perhaps by aggregation or by estimating spatial autocorrelation, see p. 13) or between data sets to relate one aspect of a population to another (health or educational outcomes and socioeconomic status, say), then geometry is unavoidably involved. Johnston et al. argue that seeing geography as an ‘ever-changing set of spatially-defined locales, often in nested structures’ rather than as a ‘fixed, geometric entity’ (2014: p. 6) represents progress. It certainly does, but it is still geometry that underpins this way of representing the world. Spatially located data are still related to one another via geometry of one kind or another. As Gould puts it,

we start with the idea that this strange no-thing [space] is structured by other things, which we relate in various ways to each other, and which we measure as various distances to each other as the fancy takes us according to our purpose of utility, curiosity, or ambition. (1997: 128)

The second minor misrepresentation is to underplay the importance of theoretical models to the original enterprise of quantitative geography. Here, Johnston et al. (2014) seem too eager to endorse the negative commentary of critics of spatial science, who point to the self-evident wrongness of theoretical models such as central place theory, diffusion models, gravity models and so on. As the epigraph at the outset of the paper points out, in all model building there is a trade-off to be made between accuracy and usability. Box puts it more succinctly, ‘Models, of course, are never true, but fortunately it is only necessary that they be useful’ (1979: 2). Given the commitment (already cited above) to understanding ‘ever-changing’ geographies, it is a pity that this facet of the spatial sciences is also neglected in the proposed curriculum, ⁴ particularly given the considerable progress in this direction in the intervening decades, under the guise of complexity science and its computational methods (Batty, 2005; Manson and O’Sullivan, 2006; Mitchell, 2008; O’Sullivan, 2004). These methods allow for truly dynamic ‘ever-changing’ representations whose behaviour can be explored as an alternative way to understand the world around us. Admittedly, the scientific community as a whole is still feeling its way towards an understanding of how to progress from ‘animations to science’ (see Railsback and Grimm, 2012), but these approaches hold promise, and unlike inferential statistics directly represent processes, not just snapshots of outcomes.

Arguably, the best ‘way in’ to these other aspects of spatial science is getting to grips with how computational tools can be used to represent geography, and how not only place can be represented but other key geographical concepts besides, such as scale, relationality, distance (in its many guises), accessibility, networks and so on. Perhaps this kind of approach to computational representations in geography would be more attractive in Johnston et al.’s (2014) vision of the geography curriculum than their relative sidelining of geographic information science/GIScience suggests. I agree with their assessment that North American geography overemphasizes GIS (and a skills-based training approach to it), while UK geography underemphasizes it. In any case, overall, I see a greater role in the curriculum for properly understanding computational representations of the world (call that ‘GIScience’ if you must!) and for exploring the implications of those representations for understanding geography, for spatial analysis and for everyday life. Such developments seem particularly crucial to a truly critical ⁵ geography faced with the ‘new quantitative revolution’ of Elvin Wyly’s commentary.

Conclusion

As I have suggested, curriculum design is rarely many steps removed from an ugly street fight. Nevertheless, contributions such as Johnston et al.’s (2014) paper are a vital part of the process. It is important to step back and reflect on changes in the social and political context and re-evaluate long-held (perhaps hard won) positions. As Elvin Wyly’s commentary makes clear, the geospatial landscape is changing rapidly, in ways that point to significant reorientations for geographical research and education. It is certainly hard to believe that geographers can adequately respond to this ‘new quantitative revolution’ without some grounding in basic aspects of computation, statistics and their various manifestations in geography under the guise of ‘spatial science’. It is equally clear that other theoretical and epistemological tools will be needed besides quantitative methods.

It is also not clear that the current situation is as dire as Johnston et al. (2014) suggest. Their own curriculum in Bristol seems in robust good health, so much so that they are proposing it as a model for others to follow. My view is that other departments and other national and international geographical communities will arrive at their own versions of these changes. At no time during my time teaching in two academic programmes (I am soon to become acquainted with a third) has everyone (or even anyone!) been happy with the curriculum. Painful as the process is, curricula are constantly under review and being adjusted and reformed. As Johnston et al. (2014) suggest other particular sets of methods and approaches may be preferred in other places (footnote 25, p. 19). I have offered some thoughts on what would ‘work for me’. Not all departments will pursue these quantitative paths, although many may do so, partially in response to ‘the new quantitative revolution’. But it would be unfortunate if a too-urgent insistence on one particular ‘model curriculum’ were to reawaken bitter disputes and so prevent other, possibly more holistic approaches (Barnes, 2009; Kwan and Schwanen, 2009; Wyly, 2009, 2011) from finding their expression in alternative curricular developments, at their own pace. Either way, I am more optimistic today about the prospects for quantitative approaches in geography that can ‘say [something] really meaningful about events as they unfold around us’ (to paraphrase Harvey, 1972: 6) than I have been at any time since I first encountered the hostility that ‘spatial science’ occasionally engenders. ⁶