FIGURE 1.1 Types of comparison (Tilly 1984).

      Expressed differently, both types of comparison have their particular strengths and weaknesses – but only in relation to a specific strategy in making a theoretical argument. What counts is not the comparison as such, but the theoretical argument. It is in this field that both individualising and universalising approaches have their Achilles’ heel.

Concepts and Causation

Necessarily, this leads us to matters of causation and explanation. How can empirical facts be connected to theory? How can we determine that something we can see and document can meaningfully be called ‘gentrification’? While the popular image of gentrification, reflected in a now globally popular blend of artists and cappuccino places, renovation activities and rising housing costs, seems to suggest an easily recognisable picture on the surface; however, the underlying substance of gentrification is more difficult to track. The reason for this is the fundamental philosophical difference between the essence of a phenomenon, its manifestation and its description. For this reason, this section will discuss ontological and epistemological problems of causation and conceptualisation, thereby sharpening our understanding of the relationships between concrete phenomena, abstraction and causation, as well as laying out some basic ideas about the value and the pitfalls of connecting empirical phenomena and theory. In what follows, I base my arguments on what has become known as ‘critical realism’ in the philosophy of social sciences (Bhaskar 1975; Harré 1986; Stones 1996; Yeung 1997; Sayer 1992, 2000).

      One of the most common fallacies to be found in gentrification research is the treatment of gentrification as a ‘real’ phenomenon, instead of as a concept. More often than not, one encounters phrases like ‘gentrification has become a global phenomenon’, ‘gentrification has expanded beyond its places of origin’, ‘gentrification has reached into new neighbourhoods’, etc. What these phrases have in common is that they reify gentrification. They treat gentrification as something that has an equivalent in reality, i.e. as an objective fact that can be measured, described and assessed visually. If one examines the definition(s) of gentrification more closely, however, it soon becomes clear that what is usually referred to as gentrification is a bundle of empirically observable phenomena, rather than a singular object. Most importantly, these are:

In terms of observation, these are distinct empirical objects. What gentrification theories do is bundle together these objects under the term gentrification. The problem is that all these properties are also linked to social relations outside of gentrification. The immigration of middle‐class households has, thus, been linked to broader social, cultural and demographic shifts in Western societies (Ley 1996); the investment of money into housing has been connected to the growing financialisation (Aalbers 2012, 2016; Haila 2016) of the real estate sector; and the decline of working class housing occupants has been explained as an outcome of the professionalisation of occupational positions in ‘knowledge economies’ (Hamnett 2003), etc. Thus, what is usually coined gentrification is not only a wide umbrella, but the phenomena covered by it are co‐determined by factors that stand outside neighbourhood change. It is, thus, easily understandable why gentrification has been termed a ‘chaotic concept’ (Beauregard 1986) that lumps together unrelated phenomena and diverse explanations.

      But what then is the ontological status of gentrification? What is the relationship of empirically observable phenomena to the concept? Different ontological positions allow different answers. Thus, from a positivist point of view, a bundle of empirical phenomena could be labelled gentrification as long as the term accurately describes patterns of similarity and regularity with which phenomenon A (say the renovation of dilapidated houses) is followed by phenomenon B (say a decline in the number of poor households). From a hermeneutic standpoint, the term gentrification would make sense only in so far as it was able to reveal the meanings that are attributed to the actual actions underpinning the process. Calling something gentrification would, thus, be adequate when the term could be found in actual discourses and used by actual actors, e.g. to demarcate social demands discursively. A realist understanding of the question would look rather different and, under this view, putting together phenomenon A (a renovation of dilapidated houses) and B (displacement of poor households) under a concept X (like gentrification) would only be justified if one was able to establish a causal relation between the two. The major consequence of a realist approach to gentrification is that conceptualisation, not research methods, is put at the centre.

Yet, how then can causality be built and examined? At the most fundamental level, each concept or theory is an abstraction. It must abstract from the particular conditions of a specific case and select those constituents of the case that have a significant effect that might also be found in other cases. In other words, the multiple elements of a concrete object must be isolated from each other, examined one by one and then recombined to form concepts that reflect the concreteness of their objects. The difficult point about abstractions