Maps are widely-known conventional models of the real world, serving as miniature representations of some spatial phenomena, trend, or relationship. The map’s role has been transformed from being an artifact of design and a navigational tool, to an interface that stores and communicates spatial information. Here, maps are decision-making tools, and are critical to policy-making by informing areas that require policy interventions. However, the production, analysis, and use of maps are prone to uncertainties that jeopardise either the maps’ accuracy, or the efficacy of policy-making. I will discuss how the modifiable areal unit problem (MAUP), varying geospatial data availability, and power relations within and produced by maps pose challenges to using maps in policy-making.
MAUP in maps causes statistical bias, resulting in inaccurate conclusions for policy-making when unaccounted for. MAUP occurs when spatial point data are aggregated into zones; the varying shape and scale of each aggregation unit influences the resultant summary values created. MAUP affects analysis outcomes because results are conditional on the spatial scale of analysis. MAUP is rather widespread because, to avoid questioning the validity of results, research involving aggregate data implicitly assumes that the MAUP does not pose a significant problem. Empirical findings biased by MAUP may lead to fallacious conclusions in analysis, leading to one drawing different or even the opposite policy implications. For example, in Canada, using Census Tracts (CTs) as neighbourhood proxies in smaller cities can produce misleading results. CTs are relatively large, and can span over both rural and urban areas, thus masking the heterogeneity of populations in the area. Aggregation by CTs can cause a large bias, which is especially pertinent to population health policy-making as it complicates the process of ‘statistically differentiating random illness patterns from deterministic ones’ (Swift et al, 2013). Parenteau & Sawada (2011) conclude, in an analysis of respiratory health and NO2 in Ottawa, that OLS regression results change significantly at different spatial representations. This may have caused policies to be misleadingly formulated against NO2 air pollution specifically. Failure to consider the impact of variable definitions of geographic units in analysis could engender inaccurate results and ineffective policies.
Maps in policy-making often suffer from uneven data availability from data handling, suppression, or collection problems, especially when data are collected by different organisations or scales of governance. Areas with missing data can jeopardise the efficacy of policy interventions when social or geographical trends are unevenly visualised. Whilst government organisations are stewards of massive datasets, subsequent key layers are produced and maintained by private organisations or non-governmental organisations (NGOs) that may use different, non-integrative databases (National Research Council, 2007). Private sector data are often shared under restrictive distribution agreements that heavily restrict access, even during emergency situations. Even government-owned data vary in availability and quality across scales of governance. Collected geospatialdata tend to emphasise formal areas, subsequently excluding informal or indigenous settlements which are significant parts of less developed countries. For example, Cape Town’s resilience maps require data from NGOs and the state, but data cannot be integrated without an interface that mediates the different technical systems and otherwise unrecognised formats in each dataset (Borie et al, 2019). Mapping efforts lack comprehensive datasets pertaining to informal neighbourhoods, which NGOs maintain but cannot integrate, leading to hazard maps that indicate city-level flood risk but not of smaller scales. This illustrates an incomplete, inconsistent picture of urban risk, causing gaps in resilience policy-making. However, if the missing data mechanism can be reliably ascertained, geo-imputation can be used to replace missing values in spatial data to alleviate missing data problems. Nevertheless, uneven data availability, caused by problems in integration of multiple datasets from different agencies and in restrictive data distribution agreements, can pose a challenge to usingmaps in policy-making.
Power is manufactured by and within maps (Harley, 1989); political power is inherent and imbued in the production and use of maps. Using maps in policy-making reflects the tensions between ‘scientific traditions’ and ‘the demands of…policy-makers’ who have to answer to the manifold and sometimes conflicting interests of the public (Haughton and White, 2018: 439). Mapping can be understood as ‘contested and contingent processes’ of power (ibid), and of ‘producing selective representations’ of landscapes (Gustafson, 2015: 146). The choice of data to omit or include can reflect how maps are always ‘selective’ and ‘political’ (Leuenberger, 2014 in Ruggeri, 2014). In southern Appalachia, risk hazard mapping has impacted property values, leading to political pressure to remove these maps from planning consideration, development ordinances and local government websites (Gustafson, 2015). The map’s contested politics arises from the economic interests conflicting with public health interests that necessitate fact-based maps to guide public safety management. Similarly, in Indonesia, government agencies used to collect geospatial information on the state-level, exacerbating existing issues with land conflicts. The confluence of varying data availability and power relations within cartography has manifested in territorial conflicts with communities of indigenious people: in the Kampar Regency, customary law communities have tried to gain official recognition of their territories on maps, for years, to no avail (Ridhwan et al, 2019). Today, policy-makers still use maps to enact power and ownership over lands and resources, often at the expense of natives (Miller, 2018), leading to exclusionary, misleading policy-making that disenfranchises entire populations of already marginalised peoples. The disuse and misuse of maps illustrate the power of those in economic and political dominance to sideline public interest and safety. Thus, while the map is ‘an instrument of state policy’ (Harley, 1989), it is equally a tool of power in mismanaging policy.
Maps are a model of reality, able to discern trends, relationships and spaces for policy intervention, but often incompletely represent reality’s myriad intricacies. While maps are powerful tools for visualising spatial data to guide policy-making, they misleadingly project a sense of ‘objectivity and certitude’ (Monmonier, 2006: 373). It is a desideratum to recognise how maps ‘legitimates…social dimensions’ insofar as it hides and denies them (Harley, 1989), and to strive towards quantifying uncertainties and accounting for possible biases, in order to produce accurate maps for equitable use and for public good.
(999 words)
References
Borie, M., Ziervogel, G., Taylor, F.E., Milington, J.D.A., Sitas, R. and Pelling, M. (2019) ‘Mapping (for) resilience across city scales: An opportunity to open-up conversations for more inclusive resilience policy?’ in Environmental Science and Policy, 99, 1–9.
Gustafson, S. (2015) ‘Maps and contradictions: Urban political ecology and cartographic expertise in southern Appalachia’ in Geoforum, 60, 143–152.
Harley, J. (1989) ‘Deconstructing the Map’ in Cartographica, 26, 2, 1–20.
Haughton, G. and White, I. (2018) ‘Risky spaces: Creating, contesting and communicating lines on environmental hazard maps’ in Transactions of the Institute of British Geographers, 43, 435–448.
Miller, G. (2018) ‘How mapmakers help indigenous people defend their lands’ (WWW) Washington, DC: National Geographic (Retrieved 24 May 2020 https://www.nationalgeographic.com/culture/2018/10/indigenous-cultures-mapping-projects-reclaim-lands-columbus/).
Monmonier, M. (2006) ‘Cartography: uncertainty, interventions, and dynamic display’ in Progress in Human Geography, 30, 3, 373–381.
National Research Council (2007). Successful Response Starts with a Map: Improving Geospatial Support for Disaster Management. Washington, DC: The National Academies Press.
Parenteau, M. and Sawada, M.C. (2011) ‘The modifiable areal unit problem (MAUP) in the relationship between exposure to NO2 and respiratory health’ in International Journal of Health Geographics, 10, 58.
Ridhwan, D., Fajar, M., Pradana, A., Hidayat, R. and Leatemia, R. (2019) ‘3 Examples of Land Conflict Cases and How One Map Policy Can Resolve Them’ (WWW) Indonesia: WRI, Indonesia (Retrieved 25 May 2020 https://wri-indonesia.org/en/blog/3-examples-land-conflict-cases-and-how-one-map-policy-can-resolve-them)
Ruggeri, A. (2014) ‘The politics of making maps’ (WWW) Westminster, London: BBC (Retrieved 24 May 2020 www.bbc.com/travel/story/20140602-the-politics-of-making-maps).
Swift, A., Liu, L. and Uber, J. (2013) ‘MAUP sensitivity analysis of ecological bias in health studies’ in GeoJournal, 79, 137–153.
