ArticlePDF Available

Editorial—A World Without Data? The Unintended Consequences of Fashion in Geography

Full Terms & Conditions of access and use can be found at
Download by: [McGill University Library] Date: 12 October 2016, At: 20:00
Urban Geography
ISSN: 0272-3638 (Print) 1938-2847 (Online) Journal homepage:
Editorial—A World Without Data? The Unintended
Consequences of Fashion in Geography
Richard Shearmur
To cite this article: Richard Shearmur (2010) Editorial—A World Without Data? The
Unintended Consequences of Fashion in Geography, Urban Geography, 31:8, 1009-1017, DOI:
To link to this article:
Published online: 16 May 2013.
Submit your article to this journal
Article views: 93
View related articles
Citing articles: 8 View citing articles
Urban Geography, 2010, 31, 8, pp. 1009–1017. DOI: 10.2747/0272-3638.31.8.1009
Copyright © 2010 by Bellwether Publishing, Ltd. All rights reserved.
Richard Shearmur2
INRS Urbanisation Culture et Société
Université du Québec
Montréal, Canada
The issue raised in this essay—the abrupt abolition, without consultation, of the
census’s mandatory social questionnaire, by a minority government with a strong ideo-
logical bent—is primarily a Canadian one.3 However, in light of news from the UK and
debates (of an admittedly different nature) concerning the U.S. census, it casts light on
wider issues such as the nature of statistics, their role in constructing a shared social imagi-
nary, and the role that academic fashions may inadvertently play in paving the way for
such destructive political decisions.
The outline of the story is simple. On 10 July 2010, articles ran in the Financial Times
and the Telegraph (Hope, 2010; Pickard, 2010) announcing that the British government
is going to abolish the national census. It is considered too expensive and intrusive, and
the data are out-of-date before they can be compiled (data more than one year old are
considered to be of no use, according to Britain’s Cabinet Office minister). Instead, it is
proposed that administrative data and private data (such as credit ratings) can be relied
upon to gather a quasi-instantaneous picture of the British people and society. Similarly, in
late June 2010 during the G-20 riots, the Canadian Minister for Industry quietly announced
that the Canadian census’s long form—the form distributed to 20% of Canadians and from
which detailed income, housing, language, employment, occupational, family, and ethnic-
ity information is gathered, all at a fine spatial scale—will be made voluntary (Proudfoot,
2010). The reason given is that an (unspecified) number of Canadians have complained
that it intrudes on their privacy—probably while using Facebook and purchasing goods by
credit card over the Internet.
This should come as a relief to some geographers. Ever since David Harvey (1973)
seminally put the first nail into quantitative geography’s coffin, a number of radical, post-
modern, cultural, and other geographers have been hammering away, as indeed have
social scientists in other fields. Thirty years of academic bludgeoning seem finally to
have borne fruit: a generation of innumerate students, some of whom are now politicians,
1I would like to thank Susan Hanson, John Adams, and Elvin Wyly for comments made on earlier versions of this
essay. It is much improved as a consequence, although its contents remain solely my responsibility.
2Correspondence concerning this essay should be sent to the author at INRS-Urbanisation Culture et Société,
Université du Québec, 385 Sherbrooke East, Montréal, H2X 1E3, Québec, Canada; telephone: 514-499-4052;
fax: 514-499-4065; email:
3The questionnaire will henceforth be voluntary. The unanimous view taken by statisticians and data users is that
a voluntary survey will incorporate unknown and unknowable spatial and social biases, and that response rates
will be far lower than for the mandatory questionnaire.
has understood the truth of the oft-quoted adage “Lies, damned lies, and statistics.The
message has even traveled across the political spectrum, because it is coalitions led by
right-wing parties that are in the vanguard of this final burial of authoritative numbers!
It is true that some geographers (e.g., Tickell et al, 2007; Kwan and Schwanen, 2009)
have recently offered counter-arguments that numbers can be important for critical and
radical geographies, and it is also true that numbers have continued to be used by many
human geographers since 1980. However, there is a feeling, expressed by a variety of
researchers (Markusen, 1999; Fortheringham et al., 2000; Shearmur and Charron, 2004;
Tickell et al., 2007; Kwan and Schwanen, 2009) that quantification has, since the 1980s,
been on the defensive compared to qualitative, case-based, culturally informed, and radical
geographies. Influential thinkers have damned numbers with faint praise: Lynch (1994, p.
330) deplored the fact that quantitative analyses of cities are “flavorless” and “tedious to
read” and Soja (2000, p. 190) lamented that the broad social geometries revealed by quan-
titative analyses “mislead not because there is disagreement over their degree of fit … [but
because] geographical covariance in the form of empirico-statistical regularity is elevated
to causation and frozen into place without history.” These sentiments seem to have become
pervasive throughout much of human geography, and although numbers are still used by
many researchers, the status of quantitative analysis as a worthwhile intellectual pursuit (as
opposed to a background technical occupation) is questioned.
There is increasing realization, however, that radical and critical geographers have in
fact relied on numbers to elaborate their critiques (Kwan, 2009), and that numbers, there-
fore, may need to be taken more seriously than critics such as Lynch or Soja suggest.4
How better to explain the importance of work on feminism than to demonstrate, using
census data collected in a transparent and rigorous way by census statisticians, that women
are systematically underpaid for similar work as men? How better to draw attention to
inequalities in urban areas than to have recourse to authoritative income and employment
data at the census-tract level? And how better to demonstrate that the benefits of economic
growth are concentrated geographically and socially into fewer and fewer hands than to
have income data cross-matched with place of residence and ethnic origin—data that only
the census can provide?
Unfortunately, numerate geographers who are today attempting to reassert the intellec-
tual value of numbers and their relevance for all brands of human geography may be too
late. The current whittling away of the census’s authority is consistent with arguments dili-
gently constructed over the past 30 years. To the extent that these arguments have served
to demote statistical analysis as a worthwhile intellectual pursuit within human geography,
it can be argued that, as a discipline, we are to some extent reaping what we have sowed.
Even if the British and Canadian censuses survive, they will survive weakened because
it is now up for debate whether or not geographically and socially representative data on
social trends are worth gathering. Powerful coalitions, relying on populist arguments, have
already decided that they are not.
4It should be noted that neither Lynch nor Soja, nor indeed most other critics, deny that numbers have some uses.
But they are seen as rather uninteresting background information for the truly important radical, critical, and per-
son-centered work that occurs elsewhere. There has been a tendency to downplay the idea that numerical analysis
may be just as important to our understanding of social processes as these other ways of comprehending them.
But what is the census, and why is it important? The census, and the data it collects,
is a vocabulary. Just like words, each number in the census represents a concept, and that
concept can of course be debated and deconstructed, as can the associated measurement
techniques. However, even Derrida managed to express himself: the words and sentences
he so skillfully deconstructed still carry meaning because words and grammar are the
results of long histories of cultural refinement. So even though deconstruction can some-
times be useful and illuminating, communication is only possible because most of the time
we choose to accept the consensual and historically constructed meaning of words and
word combinations, as good, if approximate, starting points.
The census can be viewed in precisely the same light. Each concept (income, family,
place of work, occupation, ethnicity, etc.) has been arrived at after slow refinement (over
nearly 200 years in the case of the Canadian and modern British censuses), and each
concept is further refined, usually at the margin, in each census through ongoing debates
in academia, government, business, and society at large (Alonso and Starr, 1987; Skerry,
2000). In order to make sense of the world, most statisticians—and indeed most research-
ers who rely, however indirectly, on census-derived information—choose to accept the
consensual meaning of the data, as a good, if approximate, starting point.
It is because census concepts and measurement techniques have been arrived at through
slow cultural consensus, and because the sampling framework, survey methods, questions,
and definitions are all in the public domain (and thus open for debate), that the census has
authority unlike other statistics (in a similar way that the Oxford Dictionary does in the
domain of word definitions). The census is not just numbers, but includes the institutions
and history that lend credibility, weight, and meaning to the numbers. Doing away with the
census is, for a society, the equivalent of abolishing dictionaries for a language. Without
any authoritative definitions and etymology, whether for numbers or words, no one will
be able to ascertain what is being said or what is being disagreed upon, and each group or
subgroup will begin to believe what they choose in isolation from the rest of society, with
whom no common words, or statistical reference points, will exist.
Of course, I do not mean to imply that people are not free to believe what they choose.
But anarchy and totalitarianism are avoided in free societies at least in part because some
minimal elements are agreed upon as “facts”:5
where statistical collecting and reporting agencies enjoy a reputation for pro-
fessionalism (as they generally do in our society), their findings are commonly
presented—and accepted—as neutral observations (Alonso and Starr, 1987, p. 1).
For example, sector and occupational category of employment, jobless rates, commuting
patterns, and income distribution—all of which are derived from the census, particularly
for subnational spatial units—are widely accepted and relied on as facts. Their signifi-
cance and what should be done about them can be hotly debated, but so far—in Canada at
5For a more complete and nuanced discussion of the connection between democracy and census statistics, see
Prewitt (1987).
least—their status as facts upon which there is broad agreement has not been a matter for
discussion or ideological distortion because of the professionalism, neutrality, and open-
ness of Statistics Canada.6 Henceforth, there will be no such facts in Canada.
The absence of authoritative statistics is profoundly regressive. Policies will no longer
even pretend to be evidence-based, and all that will be left to guide them are impressions
of social trends seen through lenses of ideology. Who will benefit from this numerical
polyvocality? Surely not the underprivileged and powerless. Indeed, as has been amply
demonstrated by obfuscation over the count of Iraqi civilian deaths, a lack of authorita-
tive statistics allows the powerful to (literally) get away with murder (Ellis, 2009). And
those among the powerful who seek to use their power humanely will, in all but parochial
spheres, be unable to do so, because even they will be bereft of the comprehensive and
authoritative guidance that the census provides.
Without an authoritative source of concepts and data that allow us to track inequalities,
women’s participation, incomes in peripheral areas, and longevity, how will social problems
be identified, interventions targeted, or policies assessed? Of course, there will be wrench-
ing stories of inequality and oppression, and there will also be happy stories of prosperity
and growth; but that is all that will be left, disjointed stories with no common thread.
As Benedict Anderson (2006) has eloquently argued, the communities that we live in,
our nation-states in particular, are built of stories, but not of disjointed stories. When they
thread together into a coherent narrative they are an important way of imagining society,
of explaining where it comes from and of projecting it into the future. But as Anderson
himself points out, and as Desrosières (1998) and Curtis (2002) convincingly demonstrate,
stories and imagination are built from numbers as well as words, and specifically from the
census that carries the history and authority to allow for the building of a shared narrative
(Anderson, 2006, pp. 164 ff.). The census enables people to imagine the materiality of
entities, be it countries, cities, or social groups, which stretch well beyond their ability to
perceive. Indeed, if one were asked to describe the United States in a few words, almost
certainly these words would either include numbers or be based on certain numbers: how
else could its size, diversity, and dynamism (for example) be fully imparted without at least
some numerical concepts?
It is argued (Hope, 2010) that there are many other sources of data, particularly admin-
istrative data, that can be used to track social changes and feed the imagination. These
data do exist. They are even backed by state authority. But they are not authoritative—
at least in countries such as Canada and the UK—in the manner of the census. Each
government ministry department collects information specific to its function, using popu-
lations and concepts specific to current needs, and using classifications, formats, and data-
gathering methods adapted to a particular use. There is little need to ensure the historical
or geographical consistency of such data, which are collected for short-term purposes.
6Prewitt (1987) emphasizes that professionalism and peer review are important bulwarks against political
intervention in the data gathering and dissemination process. Of course, professionalism does not resolve issues
surrounding measurement error, underlying concepts, and the like—these are objects of slow cultural refinement
and ongoing debate.
Crucially, the concepts and methods that underlie these data are not open to public debate,
critique, and appraisal, and the surveyed populations are not designed to be representative
of society at large. Furthermore, these data are usually not accessible to researchers or to
the wider population, and certainly not in any clearly organized and aggregated format
amenable to cross-tabulations. This is because of privacy concerns, because each admin-
istration guards its data, and because no one has the mandate (or possibly even the power)
to harmonize these data and use them to provide society with the means to forge identities
that build upon the past and project into the future.
It should be noted, however, that a census per se is not necessary for a shared statisti-
cal language and numerical imaginary to exist. In Sweden, for instance, social data are
gathered through government departments: except for a head count, there is no compre-
hensive data-gathering exercise. However, Statistics Sweden coordinates data-gathering
across government departments and ensures that concepts, questions, and methodologies
are consistent (Statistics Sweden, 2006). Thus what is required for census-like data to exist
are a set of institutions and an authoritative coordinating agency that ensure that a coherent
body of data, comparable over time and space, is gathered and organized. In Canada and
the UK, this has led to the development of the census apparatus, in Sweden to a strong
coordinating agency for official data-gathering. It is these institutions, rather than any
particular form that they may take, which are crucial.
It is also argued—if for a moment we grant the census some importance—that it costs
too much. There is no doubt that good data cost money, but at $567 million every five years
(Bonoguore, 2007) the Canadian census is a bargain. The recent G-20 meeting in Toronto
cost $1.2 billion to host, and each of the 65 F-35 fighter jets ordered by the Canadian gov-
ernment in the summer of 2010 will cost $250 million (including maintenance expenses;
Canadian Press, 2010). For $16.65 per person7 spread over five years (i.e., for one cup
of coffee a year) the census provides authoritative information on the very societies that
the G-20 leaders claim to lead. It assists researchers, policy makers and businesses in
understanding, governing and enriching the communities that the fighter jets purportedly
protect, and it helps tell the story of what society is, where it comes from and where it may
be heading. If the G-20 leaders no longer have the means of knowing the societies they
govern, if these societies fail to have some shared and authoritative information with which
to imagine themselves, what are all these outlays for? What, exactly, will the fighter jets
be protecting?
One of the underlying problems behind the Canadian and British politicians’ easy will-
ingness to forgo authoritative statistical knowledge about their societies and their evolution
7$567 million divided by 34 million. The cost is comparable (actually slightly less per capita) in Britain; the 2001
census, for a population about twice the size of Canada’s, cost £254 million (i.e., about 600 million Canadian
dollars at 2001 exchange rates, or $700 million allowing for inflation between 2001 and 2006) (BBC, 2002). At
£487 million (Hope, 2010; $780 million at current exchange rates), Britain’s census costs remain comparable
to Canada’s. The cost of the U.S. census, which was less than $5.00 per head in 1980 (Gauthier, 2002), was
projected to be more than $46.00 for 2010 (GAO, 2009), three times the cost of the Canadian or British census.
Clearly, if these cost estimates are comparable between countries, cost is an important issue in the U.S. but one
that is under control in Canada and the UK.
across time and space is that numeracy is undervalued. In 1959, C. P. Snow (1960) decried
the increasing gap between the humanities and the sciences, the breakdown in commu-
nications between words and numbers. This rift has evolved but has not healed, shaping
disciplines such as geography: ‘‘[C]lear lines of demarcation are today still widely rec-
ognized as separating quantitative geographers who count, calibrate, map, and model the
thing-world from qualitative geographers who converse, consort, engage, and empathize
with the people world’’ (Philo et al., 1998, p. 191).
This rift means that techniques necessary for manipulating and using numbers are
rarely taught outside of technical disciplines (such as economics), where numbers are often
unquestioningly used (they measure the “thing” world) and skill is judged by the ability to
perform complex numerical manipulations (“models”). But mathematics is a language, and
technique is not sufficient for meaning to be derived from numbers (McCloskey, 1998);
indeed technique can become obfuscating, just as pedantic grammatical contortions can
stifle an argument expressed in English.
Of course, the rhetoric of words is recognized and valued in human geography (and in
the social sciences and humanities more widely), so no one seriously thinks that special-
ists in grammar are the best creators and interpreters of text. Language is seen as a tool for
expressing ideas and meaning. In contrast, particularly in the more recent “turns” of the
discipline of geography, the ideas and meaning that can be derived from numbers (which
are just another type of language) have been downplayed: numerical and statistical imagi-
nations are devalued, sometimes, it must be said, by quantitative analysts themselves who
place technique before substance. Criticism, which has rightly been leveled by human
geographers at this overemphasis on numerical techniques,8 has slowly extended to all
research based on numbers, including that which attempts to focus on their meaning and
interpretation, and (more prosaically but just as importantly) on their descriptive capacity.
Words are valued, as they rightly should be when assembled into coherent arguments, as
descriptors and as rhetoric. But sometimes on the weakest of grounds (they are tedious to
read!), numbers—even those carefully analyzed and interpreted—are relegated to back-
ground information at best, or irrelevance at worst.
This attitude has smoothed the way for the Canadian and British governments to put
forward cost considerations and privacy concerns—both of which are important, but have
easily been surmounted in the past—as reasons to subvert an institution that has served
to imagine, understand, track, and critique society for more than two centuries.9 The easy
assurance that all numbers are the same, and that other numbers will therefore serve the
same purpose, displays these governments’ (deliberate?) failure to grasp what a census
is, not only as a collection of numbers but as an institution. This will be a great loss to
the people and social groups whose only way of being heard is through their presence in
8Good technique, like good grammar, is important, but is not an end in itself. I am not making an apology for
shoddy quantitative analysis, but am suggesting that there exist many simple and robust methods of analysis and
presentation, and that it is probably better to rely on these (wedded to good data and concepts) rather than on the
most elaborate technique (wedded to approximate or poorly conceptualized data). Furthermore, if a simple tech-
nique can provide an important insight, then relying on a more complex technique may merely make the insight
less accessible, even if it may impress colleagues more.
9The modern census is about two centuries old. Of course, according to the Bible, Jesus Christ was born in a
manger because of Herod’s census more than 2,000 years ago—so censuses are in fact much older. Old censuses
(such as the 1086 Domesday Book) are still vital sources of information millennia later.
authoritative, diligently gathered, and widely available numbers. It will also be a loss to
any serious government—whatever its leaning—that seeks to understand the effect that
policies have on people and to design better ones. Finally, over the longer term it will alter
society because the census is one of the institutions that creates and maintains our imagi-
nary communities.
Within human geography, and the social sciences more generally, the role and impor-
tance of numbers as vehicles for intellectual debate have weakened over the past three
decades, and quantitative analysts have been put on the defensive. Some of the arguments
and attitudes that now lend credibility to the dissolution of the census have been rehearsed
within our discipline, and certain geographers may indeed applaud the end of this impe-
rialist, nationalist, patriarchal bean-counting institution. But I suspect that many geogra-
phers, like myself, will be driven to think deeply about what we will lose if the census (or
census-type institutional data-gathering) disappears.
We may wonder to what extent the demotion of numerical imaginations that has
strongly influenced our discipline has contributed to the current climate in which popu-
list rhetoric (possibly motivated, at least in Canada, by the government’s all too accurate
assessment that statistics which are fairly consensual can undermine ideologically driven
assertions) outweighs reasoned arguments, from academia, business leaders, municipali-
ties, and think-tanks, relating to evidence-based decision-making and to the importance of
finely scaled social information.10 We may also begin to wonder about the slash-and-burn
dynamics of academic fashion, whereby proponents of new ideas, methods, and episte-
mologies often feel compelled to fully reject the old. Maybe we should learn to tread more
carefully around institutions and approaches that do not happen to coincide with our own
current view of the world, and to exercise constructive criticism rather than wholesale
Alonso, W. and Starr, P., 1987, Introduction. In W. Alonso and P. Starr, editors, The Politics
of Numbers. New York, NY: Russell Sage Foundation, 1–6.
Anderson, B., 2006, Imagined Communities, 2nd revised edition. London, UK: Verso.
BBC, 2002, Census should be reviewed. March 6. Retrieved August 17, 2010 from http://
Bonoguore, T., 2007, Canadian census sees cities surge. Globe and Mail, March 13.
Retrieved July 16, 2010 from
Canadian Press, 2010, Ottawa to spend $16 billion on fighter jets. Globe and Mail, July 16.
Retrieved July 16, 2010 from
10This summarizes the Canadian debate, and the UK debate seems to be similarly ill-informed, except that cost
is put forward more frequently there (despite costs that are no higher than those reported in Canada). The US
census is considerably more expensive (footnote 2), and the debate about the US census has been somewhat less
ideological and better informed than in Canada (for instance a voluntary survey was tested there in 2003 and
found to be wanting, rather than imposed without consultation as it has been in Canada).
Curtis, B., 2002, The Politics of Population: State Formation, Statistics, and the Census of
Canada, 1840–1875. Toronto, Canada: University of Toronto Press.
Desrosières, A., 1998, The Politics of Large Numbers: A History of Statistical Reasoning.
Cambridge, MA: Harvard University Press.
Ellis, M., 2009, Vital statistics. The Professional Geographer, Vol. 61, 301–309.
Fotheringham, A., Brunsdon, C., and Charlton, M., 2000, Quantitative Geography.
London, UK: Sage.
GAO (U.S. Government Accountability Office), 2009, 2010 census: Little time remains to
address operational issues. March 5. Retrieved August 17, 2010 from http://www.gao
Gauthier, J., 2002, Measuring America: The Decennial Censuses from 1790–2000.
Washington, DC: U.S. Bureau of the Censsus/U.S. Department of Commerce, 2002,
Appendix A, Page A-1.
Harvey, D., 1973, Social Justice and the City. London, UK: Edward Arnold.
Hope, C., 2010, National census to be axed after 200 years. Daily Telegraph, July
9. Retrieved July 13, 2010 from
Kwan, M.-P. and Schwanen, T., editors, 2009, Special Issue on the Quantitative Revolu-
tion, The Professional Geographer, Vol. 60, Nos. 3 and 4.
Kwan, M.-P., 2009, Guest editorial. Environment and Planning A, Vol. 41, 261–264.
Lynch, K., 1994, Good City Form, 2nd revised edition. Cambridge, MA: The MIT Press.
Markusen, A., 1999, Fuzzy concepts, scanty evidence, policy distance: The case for rigor
and policy relevance in critical regional studies. Regional Studies, Vol. 33, 869–884.
McCloskey, D., 1998, The Rhetoric of Economics, 2nd revised edition. Madison, WI:
University of Wisconsin Press.
Philo, C., Mitchell, R., and More, A., 1998, Reconsidering quantitative geography. Envi-
ronment and Planning A, Vol. 30, 191–201.
Pickard, J., 2010, UK census to be scrapped. Financial Times, July 10. Retrieved July 13,
2010 from
Prewitt, K., 1987, Public statistics and democratic politics. In W. Alonso and P. Starr, edi-
tors, The Politics of Numbers. New York, NY: Russell Sage Foundation, 261–274.
Proudfoot, S., 2010, Clement says census decision stands despite data concerns. Montreal
Gazette, July 7. Retrieved July 13, 2010 from
Shearmur, R. and Charron, M., 2004, From Chicago to L.A. and back again: A Chicago-
inspired quantitative analysis of income distribution in Montreal. The Professional
Geographer, Vol. 56, 109–126.
Skerry, P., 2000, Counting on the Census? Race, Group Identity, and the Evasion of Poli-
tics. Washington, DC: Brookings Institution Press.
Snow, C., 1960, The Two Cultures. Cambridge, UK: Cambridge University Press.
Soja, E., 2000, Taking Los Angeles apart: Towards a postmodern geography. In R. Le Gates
and F. Stout, editors, The City Reader, 2nd revised edition. New York, NY: Routledge,
Statistics Sweden, 2006, Sufficient quality and criteria for official statistics. Stockholm,
Sweden: Statistics Sweden. Last accessed August 31, 2010 from
Tickell, A., Sheppard, E., Peck, J., and Barnes, T., editors, 2007, Politics and Practice in
Economic Geography. London, UK: Sage.
... Censuses provide a 'gold standard' population map by surveying every individual household nationally. However, this is an extremely expensive endeavor, and therefore they are typically conducted every ten years [21]. During the intercensal period projections of the population are available at a coarser enumeration level that is estimated either using birth, death, and migration rates [20], or through combining sample household surveys using small area estimation [21]. ...
... However, this is an extremely expensive endeavor, and therefore they are typically conducted every ten years [21]. During the intercensal period projections of the population are available at a coarser enumeration level that is estimated either using birth, death, and migration rates [20], or through combining sample household surveys using small area estimation [21]. Estimating finer-resolution intercensal population, e.g., over a 100 m grid, has received significant interest in the last decade, and several population maps, e.g. ...
Full-text available
Knowledge of population distribution is critical for building infrastructure, distributing resources, and monitoring the progress of sustainable development goals. Although censuses can provide this information, they are typically conducted every ten years with some countries having forgone the process for several decades. Population can change in the intercensal period due to rapid migration, development, urbanisation, natural disasters, and conflicts. Census-independent population estimation approaches using alternative data sources, such as satellite imagery, have shown promise in providing frequent and reliable population estimates locally. Existing approaches, however, require significant human supervision, for example annotating buildings and accessing various public datasets, and therefore, are not easily reproducible. We explore recent representation learning approaches, and assess the transferability of representations to population estimation in Mozambique. Using representation learning reduces required human supervision, since features are extracted automatically, making the process of population estimation more sustainable and likely to be transferable to other regions or countries. We compare the resulting population estimates to existing population products from GRID3, Facebook (HRSL) and WorldPop. We observe that our approach matches the most accurate of these maps, and is interpretable in the sense that it recognises built-up areas to be an informative indicator of population.
... Physical access to basic services such as clinics, schools and water sources are often not known (Nesbitt et al., 2014) (Dougherty, 2019). It can take years to find out about the true impact of policies and interventions if we rely on census and survey data alone (Martin, 2007) (Shearmur, 2015) (Shearmur, 2010). Focusing on smaller administrative units can have a significant impact on poverty alleviation. ...
... This is often the largest administrative undertaking in any country as every person should be enumerated. Reflecting this huge undertaking, in most countries a census is conducted once every 10 years (Shearmur 2010). The gap between surveys means that if we rely solely on this information we can often see that a demographic or socioeconomic change occurred but we may not necessarily understand what was driving this change or when it started to occur during the intermediate years between enumeration (Rigg 2006). ...
Assessing the sustainability of natural resource management choices for agricultural and forest lands requires quantification of potential changes to a set of environmental and socioeconomic indicators selected to characterize reference scenarios relative to projected future scenarios. Correctly framing the questions with local stakeholders is a critical first step in the sustainability assessment, and the questions that can be addressed are often limited by data availability. Selecting and prioritizing indicators with stakeholders to address their needs and concerns improves the likelihood of investment in monitoring and evaluation of those indicators over time. Computational techniques for analyzing interactions between the selected indicators are inherently affected by the scales and formats of the assembled indicator datasets. Data analytics have the potential to improve understanding of the potential synergies and tradeoffs involved with meeting multiple environmental and socioeconomic goals simultaneously, but timely and appropriate indicator datasets are not always available—even in this new era of “big data.” Continued improvements in data science and data analytics are needed to broaden understanding and acceptance of problems and to provide valuable information for natural resource management. Advances in these areas will enable society to design future landscapes that meet multiple objectives, including the provisioning of agricultural and forest resources along with a variety of ecosystem services (e.g., clean water and healthy soils).
... Figure 4 shows the difference in population maps produced by these two approaches. We observe that, (1) the estimated population of these approaches matched well quantitatively (Spearman's ρ 0.70, Pearson's ρ 0.79), (2) there are regions where SCIPE underestimated population compared to GRID3, and these are areas where microcensus was not available, and (3) there are regions where GRID3 underestimated population, and they usually coincided with regions where microcensus was available and SCIPE could potentially provide better estimates. Therefore, there is a high level of agreement between the two products and they provide similar estimates, and discrepancies appear in regions that lack microcensus for training. ...
Full-text available
Knowledge of population distribution is critical for building infrastructure, distributing resources, and monitoring the progress of sustainable development goals. Although censuses can provide this information, they are typically conducted every 10 years with some countries having forgone the process for several decades. Population can change in the intercensal period due to rapid migration, development, urbanisation, natural disasters, and conflicts. Census-independent population estimation approaches using alternative data sources, such as satellite imagery, have shown promise in providing frequent and reliable population estimates locally. Existing approaches, however, require significant human supervision, for example annotating buildings and accessing various public datasets, and therefore, are not easily reproducible. We explore recent representation learning approaches, and assess the transferability of representations to population estimation in Mozambique. Using representation learning reduces required human supervision, since features are extracted automatically, making the process of population estimation more sustainable and likely to be transferable to other regions or countries. We compare the resulting population estimates to existing population products from GRID3, Facebook (HRSL) and WorldPop. We observe that our approach matches the most accurate of these maps, and is interpretable in the sense that it recognises built-up areas to be an informative indicator of population.
... Thus, more direct measures of population well-being are preferable if the question of interest is local development (as opposed to system-wide productivity): local incomes and local employment are better, if imperfect, measures of local development. They are preferable to local GDP because they can generally be measured directly by way of the census (though this is no longer the case in Canada, Shearmur, 2010) and are tied to local outcomes: a community with rising personal incomes, or in which employment is rising, can be said to be developing to the extent that the population's economic wellbeing is improving. ...
Full-text available
In this chapter I examine the belief that local innovation leads to local employment and income growth. Indeed, the vast majority of research on the connection between entrepreneurial innovation and regions has focused on understanding how different geographic contexts and locations are more or less conducive to firm-level innovation. This has recently expanded into attempts to understand how firms reach out, along virtual and physical networks, to tap into resources in more distant locations. However, this does not address the impact that innovation has on localities. I therefore propose an explanation for why it is assumed that local innovation and local growth are connected: whilst I do not wish to suggest that sustainable (in the sense of long-term) local development is possible without some degree of local innovation, I argue that it is possible - and indeed likely – that in many circumstances (i.e. in most smaller cities and regions) local firms can be innovative without engendering local growth or development.
... Consequently, it is frequently difficult to understand the temporal and causal processes and details of human capital generation. Further, the census and related files have been seen by critics as being intrusive and out-of-date (Shearmur 2010), notions that led, for instance, to the cancellation of the US long-form census in 2000. Likewise, the UK government suggested that their census could be dropped, with a mix of administrative and survey data used in its place. ...
Human capital has been an important source of growth for cities and regions and a driver of differences in wage levels, with a large and growing body of literature exploring its sources, growth, migration, and implications. Increasingly, however, the questions are not what the returns to investment in human capital are and where those returns are maximized, but what factors are associated with decisions to invest in human capital. How does where we grow up, where we live, where we move to, and where we work matter? Although we are increasingly close to having the data to build a picture of those decisions, they bring with them a potentially larger number of theoretical and econometric issues. How do we meaningfully identify these relationships? Can we use big data as a means to shed light on what drives the accumulation of human capital? This chapter will provide some initial thoughts and insights into understanding our investment in human capital in the era of big data.
Transformation of the Earth's social and ecological systems is occurring at a rate and magnitude unparalleled in human experience. Data science is a revolutionary new way to understand human-environment relationships at the heart of pressing challenges like climate change and sustainable development. However, data science faces serious shortcomings when it comes to human-environment research. There are challenges with social and environmental data, the methods that manipulate and analyze the information, and the theory underlying the data science itself; as well as significant legal, ethical and policy concerns. This timely book offers a comprehensive, balanced, and accessible account of the promise and problems of this work in terms of data, methods, theory, and policy. It demonstrates the need for data scientists to work with human-environment scholars to tackle pressing real-world problems, making it ideal for researchers and graduate students in Earth and environmental science, data science and the environmental social sciences.
The UN has called for a ‘data revolution’ to help overcome the low quality and lack of regularly updated statistical data available in developing countries. But how do we achieve this with limited financial resources and insufficient capacity in national statistical offices around the world? Recent studies have demonstrated how information captured by satellite imagery can be combined with social datasets to increase our understanding of socioeconomic systems. Thus, in the future, satellite data may offer a cost-effective way to reliably measure and monitor progress towards development goals. We examine how satellite data can be linked with household and census datasets to provide information on socioeconomic conditions. We suggest that the Sustainable Livelihoods Approach provides an appropriate framework for which to develop remotely sensed earth observation (EO) data proxies for key socioeconomic conditions because it will allow the linking of data in a way that reflects more the way in which populations interact with landscapes. The aim of using EO data for mapping and predicting socioeconomic conditions is not to replace survey data but to provide more frequent information on likely socioeconomic conditions between census and survey enumeration. Timely recalibration of models predicting poverty from EO data would be necessary to reflect often rapid social, economic and political changes. However, if we are to acheive the SDGs more frequent data at finer spatial scales will be required and EO data provides a cos effective solution.
Full-text available
The recent surge in populist movements sweeping many countries has brought into focus the issue of regional inequality. In this paper, we develop a panel dataset for Canada that includes information on 284 regions observed at 5-year intervals (from 1981 to 2011) and estimate a series of spatial econometric models to study the causes and consequences of regional inequality. Our results draw attention to the fact that the rise in inequality at the national-level has been accompanied by greater cross-regional inequality. Differences in the level of economic development, precariousness of labour market conditions, socioeconomic factors are among the key drivers of these regional patterns of inequality. We also find that the industrial mix of a region plays an important role in shaping its distribution of income: regions with high concentrations of manufacturing activities typically have lower levels of inequality whereas regions with high concentrations of tertiary services, arts and entertainment as well as knowledge intensive business services tend to have higher levels of inequality. In terms of the consequences of inequality, the growth/equity trade-off across Canadian regions varies significantly over the short-vs. medium-term horizons. In the short-run, our results suggest that inequality is positively related to regional economic growth. This response changes as we move to a medium-term horizon which suggests that as inequality persists over longer periods of time it has a negative and significant impact on regional growth trajectories. Panel vector autoregressive models are also used to further explore the direction of causality of the growth-inequality relationship.
Perhaps one of the mostly hotly debated topics in recent years has been the question of "GIS and Big Data". Much of the discussion has been about the data: huge volumes of 2D and 3D spatial data and spatio-temporal data are now being collected and stored; so how they can be accessed? and how can we map and interpret massive datasets in an effective manner? Less attention has been paid to questions regarding the analysis of Big Data, although this has risen up the agenda in recent times. Examples include the use of density analysis to represent map request events, with Esri demonstrating that (given sufficient resources) they can process and analyze large numbers of data point events using kernel density techniques within a very short timeframe (under a minute); data filtering (to extract subsets of data that are of particular interest); and data mining (broader than simple filtering). For real-time data, sequential analysis has also been successfully applied; in this case the data are received as a stream and are used to build up a dynamic map or to cumulatively generate statistical values that may be mapped and/or used to trigger events or alarms. To this extent the analysis is similar to that conducted on smaller datasets, but with data and processing architectures that are specifically designed to cope with the data volumes involved and with a focus on data exploration as a key mechanism for discovery. Miller and Goodchild (2014) have argued that considerable care is required when working with Big Data significant issues arise from each of the "four Vs of Big Data": the sheer Volume of data; the Velocity of data arrival; the Variety of forms of data and their origins; and the Veracity of such data. As such, geospatial research has had to adapt to harness new forms of data to validly represent real-world phenomena.
This article questions the idea that quantitative methods, in particular the analysis of social statistics, is at odds with critical approaches to geography. It argues that numbers-based research is vital to highlight social injustice and oppression and that quantitative research can meet the requirements of critical geography to be reflexive, politically conscious, and activist. The article highlights two issues of pressing interest for research and activism by critically inspired quantitative researchers. First, there should be vigilance about the retrenchment of data collection and releases by the state, which will end up obscuring the ability to see social inequalities. Second, there should be a vigorous challenge to the political and judicial undermining of the results of statistical sampling and inference. These are vital tools for estimating hard-to-count populations and inferring inequalities between groups. The article gives examples of where these estimation issues are critical, which include civilian deaths in Iraq, counting minority populations in the United States, and the detection and remediation of structural racism.
This review discusses three books by Benedict Anderson, David D. Laitin, and Anthony D. Smith in light of the perceived rivalry between constructivist and modernist theories of nations and nationalism and primordialist and perennialist ones. The concept of the imagined community is inadequate, and Anderson's contribution to theory is limited and possibly contradictory. Laitin's use of the tipping game and his version of rational choice theory are flawed, and his theory of identity change in the former Soviet Union is unsupported by his own data. Finally, Smith's notion of the ethnie, while conceptually unpersuasive, serves the theoretical purpose of reconciling modernism with perennialism.
Inspired by recent developments in social theory and based on extensive archival research, this book provides the first systematic analysis of the developing knowledge capacities of the state in Victorian Canada. No government can intensively administer citizens about whom it knows nothing. The centralization of knowledge in the form of official statistics was an important dimension of state formation. The census of population was the leading project for the production of social intelligence. .The Politics of Population. provides a detailed account of the political and social context in which census-making developed in Canada. It deals with census-making as a political project, investigating its place in and impact on party politics and ethnic, religious, and sectional struggles. It also looks closely at census-making as an administrative practice, identifying the main census managers and outlining the organization of five attempts at census-making between 1842 and 1850, before following in detail how census-making finally unfolded between 1852 and 1871. Curtis examines parliamentary debate and governmental reports, but he also follows census enumerators into the field and traces how what they saw was worked up into 'official statistics.' Theoretically, the manuscript engages in a critical dialogue with work in the history of statistics, studies of state formation, social studies of scientific knowledge, and work in the field of 'governmentality.' Winner of the Sir John A. Macdonald Prize, awarded by the Canadian Historical Association, and the John Porter Prize, awarded by the Canadian Sociology and Anthropology Association.
This publication provides census questionnaires and instructions for decennial censuses beginning in 1790 and ending in 2000. It also presents a history of the decennial censuses from 1790 to 2000 and offers individual histories of the U.S. census from the first (1790) to the sixteenth (1940). The publication discusses the seventeenth (1950), eighteenth (1960), nineteenth (1970), twentieth (1980), twenty-first (1990), and twenty-second (2000) censuses in more detail by examining the procedures to improve coverage, use of sampling, field enumeration, publicity, and Census 2000 advertising campaign. Three appendixes contain: (1) "U.S. Population and Census Cost"; (2) "National Archives and Records Administration Headquarters and Regional Branches"; and (3) "Availability of Records for the Eleventh Census of the United States." (Contains 34 references.) (BT)