CAPTURING DISPLACEMENT: A Dialectical Mixed-Methods Approach to the Study of Renoviction—A Case from Sweden

A dialectical mixed-methods approach to the study of displacement

Our methodological approach is connected to the field of mixed methods, of which scholars in urban geography and related disciplines have elaborated several variations (Plano Clark and Creswell, 2008; Sweetman et al., 2010; DeLyser and Sui, 2014; Finney, 2021). According to Guetterman et al. (2024), mixed methods consist of a combination of quantitative and qualitative methods. As Tashakkori and Creswell (2007) point out, some studies take the issue of integration of methods more seriously than others, and whereas some researchers use a multiplicity of methods in one or more phases, others mix methods throughout the research process. Triangulation, a methodological approach close to mixed methods, intended to allow for verifying the results from one to another type of method (Rothbauer, 2008), has been criticized for ignoring the ‘different and incommensurate ontological and epistemological assumptions associated with various theories and methods’ (Blaikie, 1991: 115).

In this study, we move further in two directions. Firstly, our integrated methodological proposal entails a continuous dialogue throughout the entire research process between researchers who work with different methods but share similar ontological and epistemological standpoints. In a dialectical interaction between methods, the researchers should consider all the theories identified as relevant to their study and maintain a permanent dialogue throughout the course of the research (Guetterman et al., 2024). Thus, we attempt to interrogate the variegated sources of data analyzed from a particular epistemological standpoint through what we designate as a dialectical mixed-methods approach (Figure 1).

In essence, we conceive of the phenomena of renoviction and displacement as socially constructed and fully embedded in the power relations that structure the housing market and society as a whole. Our collaboration is reflected throughout the entire research process: in the analysis and writing and in the procedure of selecting which period to focus on and what geographical districts, cohorts and variables to include in the study. For instance, both previous research and our own qualitative data demonstrate that the renovations in Kvarngärdet and Gränby caused displacement pressure long before the physical alteration of the buildings commenced, and the renovations set in motion processes and changes which continued to unfold several years after the renovations were formally completed. The design of the statistical study was heavily informed by insights such as these. Moreover, our methodological choices were fundamentally shaped by the voices of those coping with the threat and effects of renoviction. To take one example, the narratives of the first tenants who received notice from their housing company about the upcoming renovations informed our choice of time frame for the statistical analyses.

The dialogue within the research group was anchored in a social constructivist approach through which an intersectional understanding of the power relations embedded within capitalism, racism and patriarchy (Hill Collins, 2000) informs the methods used, and the particular combination used in each part of the analysis. As outlined above, displacement is a process of profound economic, social and spatial restructuring, and the study of this issue is imbued with inherent methodological challenges. Similarly to feminist researchers Garcia and Ramirez (2021), our research is guided by a desire for social justice for the populations at stake, in our case tenants struggling for housing justice in the context of displacement pressure. Garcia and Ramirez call this variation of mixed-methods approach ‘a transformative’ one. Hence, we argue that the ontological differences between statistical and place-based methods, within a housing struggle context, should be anchored in a social constructivist epistemology, bridged by a dialectical approach and characterized by a transformative desire.

The methodological procedure we forward in this article seeks to combine methods in a social constructivist relational and transformative approach (McNamee and Hosking, 2012; Garcia and Ramirez, 2021), which sets statistical data and qualitative studies in a dialectic relation throughout the research process. While qualitative studies can provide important insights regarding emotional dimensions and the complexity of socio-spatial relations, they are limited in scale and can offer only a partial picture of such processes. Yet at the same time, statistical analyses that do not consider lived experiences and the complexity of displacement risk giving rise to hasty assumptions and simplified conclusions. Rather than making speculative assumptions, it is crucial to pay close attention to the people who live in the houses and allow their actual experiences of renoviction to inform the research process. Our use of quantitative methods is intended not only to accomplish a ‘counting of numbers’, but also to deepen the understanding of the structure and complexity of place transformation.

In what follows, we outline the variegated ways in which the different methods we employ in this article are put in dialogue with each other.

There are people in the house!

The qualitative data in this study were collected throughout the renovation processes in Kvarngärdet and Gränby. The study builds on more than 60 interviews with residents, materials from local tenant groups who opposed the renovations such as leaflets and posters, recorded meetings, posts on social media and newspaper articles, as well as written, recorded and visual material produced by tenants, including letters, poems and photographs. The collection of the qualitative data has been fundamentally shaped by how one of the authors (Richard) lived in one of the two neighborhoods for several years before and throughout the renovations. Her position at the time was that of tenant and scholar activist (Nevárez Martiínez, 2016), as she engaged in the resistance that unfolded. From 2010 onwards, she resisted, observed, documented and analyzed the processes in close collaboration with her neighbors.

The ethnographic data and the interviews were collected and carried out between 2010 and 2016. Interviewees were selected by means of what we designate as an ‘opportunity-based selection’, a terminology inspired by similar methodological practices developed by activist researchers engaged in housing justice projects (Briata et al., 2020; Roy and Rolnik, 2020), and the data were analyzed using Ethnographic Content Analysis (Althiede, 2008). A crucial aspect of displacement research is trust, as access to data that otherwise tend to remain obscure for research depends on the willingness of people to share their experiences. Following Hill Collins (2000), we find connectedness with a community to be essential for the validation of any knowledge process. For this article, the local situatedness and the personal experience of renoviction and local resistance opened up for a more finely tuned collection of data. The mutual building of trust throughout the process provided access to interviews and materials that would otherwise not necessarily have been made available. This kind of specific place-based engagement is important, as it opens up dimensions called for by displacement researchers regarding the temporality of these processes, their emotional impact and their implications for social justice (Fullilove, 2005; Roy and Rolnik, 2020).

The interviews were recorded and transcribed, and the respondents have given their consent. Of these interviews, 48 were held in the street, at local meetings or as part of door-knocking campaigns. These interviews ranged from 10 to 30 minutes and focused primarily on practical and everyday tensions of the renoviction process. Such encounters opened up for further inquiry and 14 in-depth interviews were conducted in the home of the tenant, structured as conversations of mutual learning, with the interviewee regarded as an expert rather than as a victim (Thambinathan and Kinsella, 2021). Furthermore, three semi-structured discussions with groups of tenants (three to five people) were conducted, covering themes of affordability, resistance and tenants’ approaches to practical issues related to the changes in the neighborhood. In total, 60% of the interviewees were of migrant background.4 The age span ranged between 20 and 85, and the gender division was 50/50.

Numbers speaking

Approaching the quantitative methodological ground in research on displacement from a relational perspective (Figure 1) implies a dialectical, permanent dialogue between methods throughout the entire research process. Statistics do not talk by themselves; rather, their use in research is conditioned by decisions and considerations that require a critical attitude from the outset of the research process.

In close dialogue with the qualitative findings, the statistical analyses in this article focus on two primary dimensions: how the renovations impacted the mobility of the tenants who lived in Kvarngärdet and Gränby before these events on the one hand, and the changes to the demographic composition of the two neighborhoods following the renovations on the other. To this end, we draw on data retrieved from GeoSweden, a database compiled by Statistics Sweden that encompasses annually updated geocoded register data on the entire Swedish population for the years 1990–2017. GeoSweden features a range of demographic and socio-economic variables available on individual level, as well as data on the Swedish housing stock, including information about tenure. All of these variables are available on a grid with a resolution of 100 meters. The longitudinal nature of the data allows us to track each member of the population over the entire period our study is concerned with. We used GIS software to manually define shapefiles that encompass the rental apartments in the two respective neighborhoods. These neighborhood definitions served as the basis for all of the statistical analyses.

In the first part of our analysis, of how the renovations impacted the tenants in Kvarngärdet and Gränby, we measure the total degree of mobility from the neighborhoods in the years prior to, during and after the renovations. Specifically, we calculate the share of residents who left their neighborhood each year between 1997 and 2017. Moreover, to assess what factors were associated with staying or leaving following the renovations, we fitted two binary logistic regression models focused on the mobility of two cohorts: the tenants who resided in Kvarngärdet in 2007 (n = 1534) and those who resided in Gränby in 2010 (n = 2161). The response variable in these models denotes whether a member of one of the two cohorts remained in or had left the neighborhood at any point during the five subsequent years (2007–2012 and 2010–2015 respectively). Descriptive statistics for these cohorts are presented in Table A1 in the Appendix.

Our choice of these two cohorts was grounded in insights from both the qualitative data and the literature discussed above on the challenges of determining from which point in time the residents of a neighborhood undergoing transformation begin to experience displacement pressure. While the renovations in Gränby commenced in 2012 and those in Kvarngärdet by around 2008–2009, they were preceded by information and rumors which circulated for several years prior to the initiation of the physical alteration of the buildings. Already at an early stage, tenants were aware that the renovations would probably be followed by major rent hikes. The two cohorts we focus on include tenants who resided in Kvarngärdet and Gränby shortly before the point when official information about the upcoming renovations was first disseminated.

The predictors in our models include standard variables such as gender, age, country of birth5 and household type. We also include two variables indicative of SES: disposable income and education level. The variable for disposable income was derived by first splitting the entire Swedish adult population into deciles based on their disposable income. The deciles were subsequently used to divide these individuals into three income groups: low (deciles 1–3), middle (deciles 4–7) and high (deciles 8–10). This categorization follows the approach used in a number of similar studies (e.g. Alm Fjellborg, 2021). The models also include a variable that indicates if an individual was a recipient of housing allowance, a type of financial support for housing granted to citizens who meet certain requirements. Diagnostics for the models indicated that no variable had a variance inflation factor above 2.

The second part of the quantitative analysis focuses on how and to what extent the renovations were followed by changes to the demographic composition of the neighborhoods. To this end, we track the changes to the demographic composition in Kvarngärdet and Gränby prior to and after the renovations. Understanding the socio-spatial changes caused by renoviction processes from an intersectional perspective that combines class, gender and race as the main factors structuring power relations (Hill Collins, 2000; de los Reyes and Mulinari, 2005), we focus on three variables: level of disposable income, education level and country of birth.

One of the moments in the research process where the qualitative results were shown to be contradictory to the statistical data was when we realized that the register data from GeoSweden provide no information on short-term contracts. Nevertheless, the ethnographic study had revealed how an increased use of these contracts had become a common practice in the neighborhoods prior to and during the renovations, and that the use of these contracts had an impact on the process. This challenged us to reconsider our previous decisions and demarcations. Against this background, we combined the qualitative accounts with an analysis of the use of short-term contracts, drawing on data from the Swedish Rent Tribunal.6

‘You have to feel good where you live, not only live somewhere’

At this meeting, held in large room full to the brim, representatives of the housing company bluntly advised tenants who could not afford the increased rents to start looking for somewhere else to stay. Several tenants who attended the same meeting similarly recalled how they experienced sleep deprivation, dizziness and even panic attacks in its wake.

While the upcoming renovictions gave rise to bewilderment, anxiety, anger and grief, they also provoked a will to resist among tenants who reacted to the disturbances inflicted upon them (Polanska and Richard, 2019; 2021). Traces and voids of those who had already left surface in several ways, here represented by Ciwan, who made a photomontage of the empty apartments in her street (Figure 2).

In her analysis, Tanihaa draws attention to the injustice and inhumane character of the situation. Tenants affected by the disruption of renovations witness how stress and processes of mourning are intertwined with demanding practical decisions concerning home, personal finances and work. Her experiences reflect the classic essay by the urban planner, social scientist and housing activist Chester Hartman (1984) on the harmful dimensions of displacement and the need to limit property rights in favor of the tenants’ right to stay put. Tanihaa, reflecting upon the impact of renovictions on her daily routines and what she perceives as a huge gap between her own position and those in power, insists that ‘the lords’ (the landlords) should take responsibility and acknowledge her pain.

References to the uneven power relations which underpin the renovations, eloquently expressed in the poem by Tanihaa, surface frequently in the material produced by tenants reflecting on their situation.7 For instance, during a casual conversation in the street, Joha, a long-term resident of Kvarngärdet approaching retirement, explained the difference between a house and a home as follows: ‘You have to feel good where you live, not only live somewhere’. This way of understanding the home is important to recognize, given that the profound relations between the residents, the houses and the neighborhood have not yet received adequate attention in housing studies (Kemeny, [1992] 2015; Molina 1997).

‘These will be social welfare neighborhoods’

A number of tenants, like Gunnel, expressed doubts already at an early stage about a possible ‘upgrading’ of their neighborhood, as well as the visions espoused by the housing companies of a ‘green’ and prosperous future. Gunnel and others argued that this part of town would remain poor and that the social services would most probably have to provide additional financial support to the residents, predictions that align with both the findings presented in a municipal report published in 2014 (Sinisalo Åström, 2014) and the statistical analyses we present further below. In a phenomenological study of the renovictions in Kvarngärdet and Gränby, Pull argues that these are neighbourhoods in decline and proposes that, through the process of renoviction, we might be witnessing a paradigm shift where ‘the housing regime of Sweden is exploited in ways in which rent extraction can increase without a corresponding increase in affluent gentrifiers’ (2020a: 153).

The narratives shared by residents and the proposal put forward by Pull are indicative of the complexity and volatility inherent to displacement pressure. Recognizing the economic, social and spatial dimensions discussed in previous research, our statistical study was fundamentally informed by the restructuring processes set in motion by the renovations. For the most part, as we will outline in what follows, our analyses of the quantitative data do not by themselves suggest that the renovations caused displacement. In this regard, the results of our analysis of the mobility of tenants prior, during and after the renovations constitute an exception of sorts. Figure 3 illustrates the percentage of the population who left their neighborhood the following year. For example, the data point for 2009 designates the share of those who had lived in the neighborhood the year before (2008) who moved from the neighborhood at some point during the next year (2009).

Figure 3 indicates that the renovations appear to have had a considerable impact on residential mobility, which sharply increased in Gränby after the start of the renovation process. While the pattern in Kvarngärdet is less outstanding, the renovations may well have increased the degree of mobility also from this neighborhood. It needs to be stressed that these numbers do not by themselves provide evidence of displacement, since the quantitative data do not contain any information about residents’ reasons for leaving. However, combined with the insights from ethnographic study, these trends strongly indicate that the renovations were a driving factor behind the increased mobility.

In contrast, Figures 4-6 demonstrate that the renovations were not followed by any major demographic changes, with the exception of an increase in the share of the population with higher education, and of residents born in the global South (Figure 6). The increase in the number of tenants leaving their neighborhood during or after the renovations was thus not combined with any major transformation of the already racialized class composition of these areas. In line with what Gunnel and other tenants argued already at the outset of the renovations, these findings point to something akin to a ‘post-gentrification’ scenario (Pull, 2020b), where tenants who have indeed been displaced are not replaced by new residents who are comparatively substantially better off. Critically, however, the renovictions accentuated the process of racialization of Uppsala which has been observed for many years (Molina, 1997; Andersson and Molina, 2003). While the share of residents with higher education increased in Kvarngärdet and Gränby, both retained their character as racialized working-class neighborhoods.8

Notably, the outcomes of the logistic regressions (Tables A2 and A3 in the Appendix) do not provide evidence suggesting that the renovations resulted in displacement. To begin with, the regressions show that higher age was associated with staying in the neighborhood, although the effect in this regard is relatively small.9 There was also a substantial effect associated with living together with a spouse and children. This category of residents had a high probability of staying in the neighborhood. Most notably, individuals with high SES were more likely to move out. In contrast, low SES was thus associated with staying, with the same being true for recipients of housing allowance.

In essence, the models indicate that, similarly to our qualitative data and the findings of Pull (2020a; 2020b), what has taken place in Kvarngärdet and Gränby, despite the already low socio-economic status of the neighborhoods, is not primarily a process where poorer residents have been the most prone to move out following major rent hikes. Rather, the residents who were most likely to leave had a comparatively higher income.

On the one hand, the quantitative explorations outlined above provide some indications that the renovations resulted in displacement, insofar as the mobility of tenants in Kvarngärdet and Gränby increased during and after the renovations. Yet, on the other hand, the demographic composition of the two neighborhoods did not significantly change, and residents with lower socio-economic status were comparatively more likely to remain.

Crucially, the findings from our analysis of the qualitative data make it clear that there is even more complexity to these dynamics. In what follows, we will show how it is plausible that the results from the statistical analyses also reflect the extent to which residents with lower incomes or parents with children in the local school go to great lengths in order to be able to stay in their neighborhood. Accordingly, it would be incorrect to jump to the conclusion that displacement did not occur. Even a ‘short move’ within the neighborhood can have detrimental effects on the people affected, and we suggest that these kinds of moves may also constitute a form of displacement. Furthermore, the extensive use of short-term contracts, addressed further below, is also important to consider in this regard.

Moving within the neighborhood

In a qualitative study on Gränby, Hallberg and Heldebro-Lantto (2018: 26) found that common strategies for the most vulnerable residents were to move to a smaller flat in the same area, sublet a cheaper apartment scheduled to be renovated at a later stage or open up their home for lodgers. Our qualitative material features a range of similar accounts.

In order not to unroot Rahwa's brother from his social networks, she and her mother gave up their security of tenure in order to be able to remain a few more years in the neighborhood, as the apartment they moved to would be renovated at a later stage. Previous reports indicate that Swedish tenants who ‘solve’ the demand for higher rent by moving to an apartment that has not yet been renovated are likely to move again in a few years (Boverket, 2014). Besides sacrificing their security of tenure, the family also gave up personal space, as in the new and smaller apartment, 20-year-old Rahwa and her mother had to share a bedroom. While the family managed to stay in Gränby, they had to cope with pressing practical issues as well as lasting stress and anxiety following the short-term nature of their new situation.

Elderly residents, families of migrant background and single household mothers go to great lengths in their struggle to stay, sharing narratives that resemble the poem by Tanihaa. ‘Knowing the place’, concerns for the children and local social networks emerge as important for their well-being. Divorced parents express concerns that their children will prefer staying with the other parent as a consequence of downscaling, while elderly residents explained that even moving just a few blocks within the neighborhood has made them feel isolated and lonely.

The trickiness of short-term contracts

Security of tenure is an important pillar of Swedish housing policy. Outright evictions, although existing and highly problematic, are still uncommon (Samzelius, 2020), and in the case of renovictions, tenants facing displacement pressure formally leave of their own accord. However, housing companies are allowed to set aside tenants’ legal protection to stay put by offering short-term contracts to residents moving into the neighborhood during a period of four years prior to a renovation.

Figure 7 depicts data retrieved from the Swedish Rent Tribunal on the number of short-term contracts solicited in Kvarngärdet between 2006 and 2022 and in Gränby between 2008 and 2022. The gray and black lines visualize the number of contracts issued each year and in total, respectively. Throughout this period, 296 short-term contracts (of a total of 1746 apartments) were issued in Kvarngärdet and 1096 short-term contracts (of a total of 1201 apartments) were issued in Gränby, indicating a rising relocation of residents. However, the remarkably high number of short-term contracts does not correspond to the number of original residents who have left, since residents who sign up for these contracts tend to stay for only shorter periods in the neighborhood. Several rounds of these insecure contracts can thus be registered on the same apartment over the years.

The extensive use of short-term contracts depicted in Figure 7 is reflected ‘on the ground’ by insecure housing conditions for the short-term tenants, and by the narratives of permanent residents who describe how the increased in- and outflux of tenants who stay in the neighborhood for only a short period of time results in unrest in stairwells, backyards and streets. A construction worker living in the area explained that ‘it's chaos in the building and many are forced to move’. The situation creates tension, as long-term residents associate their experiences of dispossession in the neighborhood and the perceived disorder with the arrival of new residents, claiming that ‘if you stay just briefly, you don't care’.

The extensive use of short-term contracts divides the tenants in neighborhoods undergoing renoviction and hinders attempts at collective resistance (Polanska and Richard, 2021). Furthermore, this practice underscores the importance of temporal and scalar sensitivity in research on displacement as, if these short-term contracts are not included in statistical analyses, or if the voices of short-term residents are not present in the data, the complexity and socio-spatial tensions of the renoviction process will fly under the radar of research.