DCUN Methodological Note

Should We Compare the Processes of Urbanization in Paris, London, and Shanghai, or Paris, Marseille, and Vesoul?

Strategies of Comparison in Urban Research

Joël Idt (Assistant Professor, University Paris Est Marne la Vallée, Lab’Urba)
Contact: joel.idt@u-pem.fr


People who do research on cities and urbanism spend their time making comparisons, just like their colleagues in other social science fields. Depending on the specific subjects of their research, they compare towns and cities, urban mobility and consumption, and urban projects. The question as to what is or is not comparable regularly comes up.

Download PDF

Full version

People who do research on cities and urbanism spend their time making comparisons,[1] just like their colleagues in other social science fields (Vigour 2015). Depending on the specific subjects of their research, they compare towns and cities, urban mobility and consumption, and urban projects. The question as to what is or is not comparable regularly comes up. Paris and Marseille are often cited as cities that are so unique, each in their own way, that they cannot be compared to any other. The size of the Paris conurbation means that its only parallel in Europe is London; for some people this immediately rules out any comparison with smaller municipalities in France, whether other large conurbations or small and medium-sized towns. Marseille is said to be characterized by its highly unusual forms of local governance, a situation which impedes any attempt to compare the processes of urban production taking place there even with those of cities similar in size.

Criticisms made by those who are sceptical of a comparative approach consist, for example, in pointing to differences in situation, or to specific local conditions which are too individualized to make any one case comparable to others. However, simply because two cities or two urban phenomena are radically different does not mean that they should not be compared. To the contrary, comparing such cases can sometimes be even more rewarding. What we see creeping in here is a misinterpretation of the verb ‘to compare’, which is sometimes used to express a similarity (‘The two cities are comparable in terms of size’; ‘The social composition of the two neighbourhoods is comparable’; etc.). But in a research context, this term refers first and foremost to the action of comparing, not the result of the comparison. Comparison must be understood here as a research method or analytical tool, or a scientific project (Bourdin 2015).

By sticking to this definition, we would like to argue for a radical position: everything can be compared to everything else (which does not mean that everything is similar), even the ‘incomparable’.[2] What matters is to know why and how, and of course to show the limits of comparison and to carry it out in accordance with the rules of the discipline (that is to say, methodological considerations, which we will address here only in passing). Now that urban allotment cultivation has become such a success, we may permit ourselves a vegetable metaphor: it is possible to compare cabbages and carrots – we see that they are certainly different in colour and taste but they can be grown on the same farm, and by mixing them together we end up with tasty coleslaw. If we replace our ingredients with urban projects or urban agglomerations, the argument still holds.

In this article, we propose to approach comparison in terms of the research strategy it underpins and the reasoning behind it. We describe several strategies for constructing a comparison, that is to say several ways to compare things: comparing areas which are similar but not entirely so, comparing areas which are different but not too different, and comparing in order to export or import research topics and questions. Our typology is by no means exhaustive; however, it is extensive enough to demonstrate that when it comes to comparison everything can be envisioned but not everything leads in the same intellectual directions or to the same results.

For our demonstration, we will rely on previous and current research which analyses processes of scattered urbanization, taking place outside urban centres on more or less distant peripheries. This research focuses on what happens outside major urban development projects, in the everyday dynamics of the kind of urbanization with a low media and political profile that Dominique Lorrain has called ‘urbanisme 1.0’ (Lorrain 2018): individuals who divide up their plots of land in order to build at the far end of the plot, farmers who sell land for subdivision, developers who combine two residential plots to build a small block of flats, and so on. An initial study has addressed the response of public authorities faced with ‘spontaneous’ densification, through a comparison between Paris and Rome.[3] Another study concerns the transformation of farmland at the interface between urban and rural areas, through a comparison of several sites in the Ile-de-France and another region nearby.[4] A third investigates illegal urbanization, through a comparison between Paris, Rome, and Hanoi.[5] We will refer only in passing to the results of this research, just enough to assess the comparative strategies at work and their potential contributions.

Taking what seem to be rather similar areas

– to confirm that the dynamics of urbanization are similar

This is probably the most common strategy used in comparative research. In the case of our research topic, it consists of choosing fairly similar areas in order to verify whether the processes of urbanization are similar. What we seek is to increase the level of generality: comparison shows that the description of the processes applies to more than one singular case, and confirms that there are correspondences between a particular type of area and the processes of urbanization taking place there. This type of argument can be used to strengthen existing categories of analysis (by refining or slightly modifying them) or to develop new ones. We are close to a Durkheimian form of reasoning,[6] establishing correspondences between several variables (in this case an area and processes of urbanization).

Our research on spontaneous densification provides an illustration of this. Specifically, we compared municipalities in the first and second ring of suburbs in the Ile-de-France in order to understand where ‘spontaneous’ densification of building stock was taking place, unconnected with public construction projects. Comparing the different cases we studied enabled us to increase the level of generality in our assessment of the sites and forms of these phenomena. For example, the largest operations are carried out by major developers, and are mainly located along major highways, close to public transport infrastructure, or in town centres. Smaller developers or individuals who want to make a property deal will undertake operations in other, more outlying neighbourhoods. This observation has been confirmed in towns such as Montreuil, Bagnolet, Vitry, and Sucy-en-Brie, among others. Such a level of increased generality can only be reached by comparing a fairly large number of cases.

– or to bring out differences

A situation in which the urban areas being compared are relatively similar but radically different processes of urbanization are observable can also be of interest heuristically. Here the strategy consists rather in disconfirming correspondences between urban situations and the processes of urbanization that go on in them, and/or in showing that extremely diverse processes can produce fairly similar outcomes. On occasion it is possible to identify a contextual variable which had not initially been thought of, and which would help to explain the variations. As in the previous case, the results obtained are general in scope, but reached through different reasoning.

Still on the subject of ‘spontaneous’ densification, in our research we have observed that communities whose geographical situation and social composition are fairly similar (briefly: formerly communist-majority towns in the inner ring of the Paris suburbs, now undergoing partial gentrification) can have sharply divided views on densification. Some are fiercely against it, while others look on it more favourably and even encourage it. Some have acquired the effective capacity to negotiate with developers, while others are relatively lacking in this respect, and so on. Without going into more detail, we see here that the forms that official public action takes are not over-determined by the social geography of these towns. Other variables must be brought in to explain these differences, such as the history of local institutions, the mayor’s interest in intervention or development, and so on.

Research on the transformation of farmland at the urban/rural interface provides another illustration of this type of reasoning. We have studied several peri-urban municipalities located within the boundaries of a regional nature park (PNR). One might have thought that they would be quite restrictive and ‘Malthusian’ (Charmes 2011) when it came to permitting further urbanization, since peri-urban PNRs have the reputation of encouraging such a strategy. In fact, our research shows diversity in the positions taken by elected representatives in this regard, including representatives from the same municipality: they are sometimes torn between restricting urbanization to preserve the rural way of life, and permitting it in order to attract a new, younger population or to diversify the residential opportunities available within the municipality.

Taking what seem to be rather different areas

– to show the common features of processes of urbanization

On the other hand, it can sometimes be appropriate to choose areas which are rather different from each other in order to show that the processes of urbanization observed in them have features in common. This may be a necessary stage before the development of more detailed classifications of the observed phenomenon, putting the emphasis on secondary differentiations. Such an approach makes it possible to broaden the scope of the phenomenon we observe, by showing that it can occur in very different kinds of area. Here again, comparison makes it possible to increase the level of generality: what might have been viewed as local peculiarities turn out to be common features. It is thus in our interest to take cases that seem in principle to be quite dissimilar if we want to be able to generalize.

The two urban agglomerations we chose for our research on the spontaneous densification of building stock (Rome and Paris) are structured administratively in significantly different ways. The city of Rome is very extensive, covering an area almost twelve times that of Paris proper (‘intra muros’). In contrast, around Paris the Ile-de-France is fragmented into a large number of municipalities. One might expect, looking at this situation from a strictly institutionalist perspective, that Rome would be better than the Ile-de-France at planning its urban development over a wide geographical area. Our research largely disproves this hypothesis. In Rome, the procedure for drawing up and approving planning documents alone takes so long that they are out of date even before they are ready to be executed. In the Ile-de-France, the fragmentation of the municipalities masks the extremely active development of plans at the local level, as new projects are constantly being implemented. In both cases, that is, planning mechanisms have shortcomings which get in the way of managing the plans. Another result of our research relates to a similar argument. Even though the two urban agglomerations we studied are very different (in their geographical, political, sociological, and other characteristics), they share observable features: the parties concerned play around with the regulations governing land use, and the local authorities and would-be developers negotiate building permits. The variety of situations enables us to classify the forms that these games can take (Idt and Pellegrino 2018).

A somewhat similar approach has been adopted for research on the transformation of farmland. In our comparisons of the peri-urban municipalities of the Ile-de-France, we have used the typology of the 2013 SDRIF (Schéma Directeur de la Région Île-de-France / Master Plan for the Île-de-France Region), which differentiates the approach to urbanization in the municipalities located on the ‘urban interface’, the ‘population hubs to be strengthened’, and the other ‘small towns, villages, and hamlets’, which are to be allowed ‘moderate expansion’: the rules designed to govern new urban development are significantly different in these cases. However, our research shows that very similar types of negotiation with the rules occur in all these areas, meaning that their degree of individuality is only relative. An interesting secondary result of our research is that the SDRIF’s capacity to manage urbanization should be called into question, since the actual processes of urbanization are indifferent to the distinctions it has established.

– or to accentuate differences

The situation in which very different areas undergo very different processes of urbanization is not without interest, contrary to what one might think at first glance. International comparisons very often address such configurations: there are too many contextual variables to make one-to-one comparisons, so other strategies have to be developed to take account of this situation. The approach may focus on emphasizing the most significant differences, possibly by accentuating or even caricaturing them, in order to construct broad analytical categories. This approach is along the lines of the Weberian method of constructing ideal types,[7] which is based on an examination of cases that are sufficiently different, and which can be modelled through comparison.

In our research projects on the densification and transformation of farmland, we analysed municipalities of very different sizes, ranging from a few hundred residents to tens of thousands. In both research projects, the size of the municipalities proved to be a significant variable determining their ability to negotiate with private developers who file permits for building on farmland. Size in effect determines the engineering resources available in the relevant municipal offices for analysing the permits and going over them with the permit filers. This finding is in line with a similar analysis by Olivier Morlet (1997), which demonstrates the extent to which the size of the municipality can affect the ability of elected officials and technicians to negotiate, either with private developers or with the state.

In a different register, our comparative research on illegal urbanization in Rome, Paris, and Hanoi leads us to emphasize to the point of caricature the differences linked to the existence of extremely strong growth in Asia and relative stagnation in Europe. Illegal urbanization obviously does not have the same meaning in the two cases; the thinking which leads to it and the games which underlie it are radically different. In the first case, illegality can be extensively practised and central to new processes of urbanization. In the second case, it is often more marginal and less overt, but its role as an adjustment variable can be crucial in regulating urban transformation (allowing a few illegal activities on the margins, for example, may be inevitable during a very serious housing crisis).

Applying research questions and topics to other areas

A final strategy for comparative research is the import and export of questions and topics, analytical frameworks, and research methodologies. For example, a question developed and tested in one area can be applied to other areas, in order to refine the analysis or query the initial results; alternatively, categories of analysis developed elsewhere can be imported for application to an area that we want to study in a different way, in order to make us look at it from another angle. In all these cases we are seeking to produce a different interpretation of the phenomena or to open up new avenues of analysis.

Research on the spontaneous densification of building stock is one example of the transfer of questions, topics, and categories of analysis. The urbanization of greater Rome is characterized by very extensive abusivismo edilizio, that is to say illegal construction, outside the areas authorized in the official plan or without a construction permit. Almost a third of the volume of construction in the city of Rome seems to have been done illegally (Nessi 2010). Abusivismo has been a public problem for a long time, and is addressed by specific action plans on the part of the authorities as well as being analysed by urbanists. This particularly Roman phenomenon led our research group to treat illegality as a particular category of analysis for understanding spontaneous densification in Rome. We have now applied this to the Ile-de-France region. While the phenomenon is far from being as extensive as in Rome, we have found that it does exist. In most of the municipalities we studied, violations of plans or building permits were reported to us, though generally on a small scale. The local authorities say they are aware of the problem but have few resources to deal with it.

Our current research on illegal urbanization in Rome, Paris, and Hanoi confirms our observation that in this case a North/South comparison is worthwhile even though the three sites may seem at first sight to be too dissimilar. In fact, illegal and informal urbanization has been extensively researched in the countries of the South, far more than in the North. Analytical frameworks have been developed and hypotheses tested. From this point of view, comparison opens up the possibility of applying existing research to other situations, as long as the limits of its transferability are clearly defined. For example, although the scale of the phenomenon is very different in the two cases, as we think more generally about the boundary between legality and illegality we may recognize very clear echoes from one to the other.


In each of the comparative research strategies described here, the usefulness of the comparison is that it enables us to increase the level of generality with respect to a single case, and to build, test, and modify categories of analysis. But the level of generality can be increased in several ways. In the first strategy, the aim is to construct correspondences between types of area and the processes of urbanization which take place there. In the second, in contrast, we seek to disconfirm these correspondences by showing that very different processes can take place in areas that we had thought were parallel. In the third strategy, we show that very different areas can give rise to very similar processes of urbanization. The fourth is that of the construction of ideal types, which are naturally different from one another. This strategy is based on the import/export of analytical frameworks constructed elsewhere and tested for validity on a new case.

But although in principle it is possible to compare everything, this does not exempt researchers from doing a substantial amount of work to avoid the methodological pitfalls inherent in comparison. We will not restate the rules governing this kind of work, which have already been largely set out elsewhere (in particular Vigour 2005). Differences in socio-geographical and cultural conditions should be taken into account. Access to data is not always parallel for the different areas studied. The categories of analysis and pre-defined terms (town or city, the urban, density, public interest, etc.) do not necessarily have the same meaning in all the situations studied; and so on. Everything can therefore be compared as long as the limitations of the exercise are clearly spelled out and the comparison is properly ‘constructed’. Otherwise, our scientific coleslaw will probably taste very strange.


Blanc M. and Chadouin O. (2015), ‘Editorial’, Espaces et sociétés, 2015/4, No. 163.

Bourdin A. (2015), ‘La comparaison telle qu’elle s’écrit’, Espaces et sociétés, 2015/4, No. 163.

Charmes E. (2011), La ville émiettée. Essai sur la clubbisation de la vie urbaine. Paris: Presses universitaires de France.

Detienne M. (2000), Comparer l’incomparable. Paris: Seuil.

Idt J. and Pellegrino M. (2018), ‘Les acteurs publics face aux phénomènes de densification spontanée. Une comparaison franco-italienne’, in Leger J.M. and Mariolle B. eds, Densifier/Dédensifier les campagnes urbaines. Marseille: Parenthèses.

Lorrain D. (2018), L’urbanisme 1.0. Enquête sur une commune du Grand Paris. Paris: Raisons d’agir.

Morlet O. (1997), ‘Les pratiques locales de la préemption’, Etudes Foncières No. 86, ADEF.

Nessi H. (2010), ‘Action publique et étalement urbain à Rome: une lecture par les services en réseau’, Flux, 2010/1, Nos. 79-80.

Vigour C. (2005), La comparaison dans les sciences sociales. Paris: La Découverte.



[1] See for example the issue of the journal Espaces et Sociétés on international comparisons in urban studies (Blanc and Chadouin 2015).

[2] Borrowing Marcel Destienne’s formula (2000) which argues for a similar approach from a different perspective, by showing the usefulness of a radical comparative approach to historical analysis.

[3] Research undertaken for the PUCA: ‘Les acteurs publics face à la densification spontanée du bâti: une comparaison franco-italienne’ (http://www.urbanisme-puca.gouv.fr/IMG/pdf/rapport_final_puca.pdf), by Joel Idt and Margot Pellegrino, carried out between 2014 and 2016 (Idt and Pellegrino 2018), with the assistance of Sarah Baudry.

[4] Research in progress, carried on within the framework of the PSDR CapIDF programme, segment 3 of the research project (with Roxane de Flore).

[5] Research in progress, carried on within the framework of the Réseau International sur l’urbanisation diffuse du Labex Futurs Urbains de l’Université Paris Est: https://dcun.hypotheses.org/ (research network coordinators: Adèle Esposito, Joel Idt, Clément Musil).

[6] For a more extensive analysis of the links between classical sociological theory and the comparative approach, see Vigour (2005).

[7] For a more extensive analysis of the links between classical sociological theory and the comparative approach, see Vigour (2005).

To cite this article:

Joël Idt, “Should We Compare the Processes of Urbanization in Paris, London, and Shanghai, or Paris, Marseille, and Vesoul? Strategies of Comparison in Urban Research”, DCUN (Diffuse Cities & Urbanization Network) Methodological Note, April 2019. URL: https://dcun.hypotheses.org/1492