Archives of Design Research

Previous research has investigated personalization in three respects, including its function, the type of employed elements, and its setting. As this behavior is largely considered to be a basic mechanism of territoriality, its main function would be to boost defense and regulate social interactions (Altman, 1975; Edney, 1976; Gifford, 1987). Greenbaum and Greenbaum (1981) made a distinction between two almost opposite functions of territorial personalization; preventing unwanted social contact and facilitating social interaction. The preventive function is strongly evident in measures taken to prevent robbery and invasion, provide a sense of security (Newman, 1973; Brown & Altman, 1983), express ownership and legitimate occupation (Sommer & Becker, 1969), and provide privacy, especially visually (Pastalan, 1970; Tames, 2004). These would be accomplished in a way that simultaneously serves the facilitative function, too. For example, the decoration of an entrance space might both welcome friends and warn burglars (Brown, 1985). However, we might distinguish those personalization acts which are mainly performed to increase the beauty and attractiveness of a place and, at the same time, convey resident’s tastes, preferences, and interests. These self-expressive instances elicit individual and communal identity, with the primary concern of winning others’ approval. Comparing a standard door guardrail, installed to prevent robbery, with a handmade wreath flower on the front door to welcome visitors can show this distinction between preventive and facilitative functions of personalization.

Personalization, despite its territorial function gaining more prominence in previous studies, inherently involves instrumental and structural interventions to adjust a given space for use, too. This third function acquires considerable significance in affordable housing as its inhabitants might not be able to relocate if the current dwelling was not congruent with their lifestyle and requirements (Tipple, 2000). For example, the way to deal with housing pressure in mass housing is sometimes to extend a house to its adjacent outdoor area instead of moving to a larger dwelling (Brand, 1995; Tipple, 2000; Tames, 2004). Therefore, affordable housing experiences such interventions as extension, transformation, subdivision, rearrangement, and the like, for its practicality to improve.

Beside function(s), each act of personalization is performed through some type of element, and in a physical setting. Built on the concept of the three features of elements of space by Hall (1969), including fixed-feature, semifixed-feature, and informal/nonfixed-feature elements, personalization elements are mainly placed under the category of semifixed ones (Rapoport, 1982; Brown & Altman, 1983; Kopek, 2006). This is because while collective sociocultural meanings can be better conveyed by fixed elements, semifixed ones serve to express personal meanings, as well, which is significant in the study of personalization (Ahrentzen, 2002). Nonfixed elements, on the other hand, are bound up with the non-verbal behavior of occupants in determination of distance and proximity (Hall, 1969; Rapoport, 1982) rather than physical markers of personalization. Therefore, we deem them inapplicable in the study of personalization and focus on fixed and semi-fixed elements.

Concerning the setting for personalization, we refer to the definition of three types of territories by Altman (1975), including primary, secondary, and public. As primary territory is permanently owned and expresses a close link among privacy monitoring, mechanism of territoriality, and self-identity (Altman, 1975), it could be more susceptible to personalization in comparison with the other two. As an example, Abu-Ghazzeh (2010), comparing flat housing with detached ones in Jordan, indicated that the residents showed a tendency to mark either owned areas or places over which they had legitimate control more than public territories. Secondary territory, also, witnesses several acts of personalization, especially in the proximity of entrance space, such as greenery, embellishment, paving, etc. (Abu-Ghazzeh, 2010; Tames, 2004; Werner, Peterson-Lewis & Brown, 1989). Types of personalization in the public territory, though, are somehow different from the instances recited in this introduction, as they generally involve short-term occupation of space (Becker, 1973) without any lasting influence on the environment.

In sum, personalization is expected to vary according to its three aspects, including function (preventive, facilitative, and instrumental), element (fixed and semi-fixed), and setting (primary, secondary, and public territories). This paper, thus, takes all three into account during investigation and discussion.

The general aim of the current study is to investigate the possible relationships between the three aspects of personalization in mass housing and residents’ housing statuses, dwelling’s physical characteristics, and household demographics. Based on the findings of sections 1.2.1 to 1.2.3, the following correlations may be hypothesized.

First, preventive personalization may correlate with those housing statuses which involve the length of residence, past or future, as well as with physical characteristics that increase the likelihood of visual and/or physical intrusion into a dwelling, including floor, presence of a private outdoor space especially adjacent to the dwelling’s entrance.

Second, facilitative personalization may be correlated with housing statuses, such as ownership type and length of residence, that promote the feeling of identity and belonging and with physical characteristics, like indoor and outdoor space size, number of rooms, and presence of open space adjacent to a dwelling’s entrance, that provides space and opportunity for self-expression.

Third, instrumental personalization, based on the housing pressure theory (Tipple, 2000), may correlate with dwelling’s physical characteristics and household demographics which prompt a family to transform, subdivide or enlarge their habitable spaces, including indoor and outdoor space size, number of bedrooms, presence of private outdoor space, number of children, age, and gender.

Fourth, the majority of personalization is expected to be employed with semifixed elements; however, the use of fixed elements may correlate with housing pressure and therefore, become affected by the same predictors as of instrumental personalization.

Fifth, the primary territory is anticipated to be the main setting of personalization, whenever a factor affects it, though the interventions in the secondary territory may be correlated with those physical characteristics which indicate the immediate adjacency of a dwelling’s primary territory to the public one, such as the floor number.

Based on our five hypotheses, we developed a set of factors of personalization and used their measurements as predictors of its three aspects in multiple regression analyses. There are, also, some intervening factors, such as residents’ personality, preferences, life experience, and the like, which might affect the personalization behavior, though they are not included in this cross-sectional study.

Of the subsidized affordable mass housing developments in Mashhad, Iran is Armes complex whose construction was concluded in 2005 and released for occupation till 2007. It is selected mainly because it includes households with higher variation in the length of residence (up to 11 years) comparing with other local instances. As no pre-registration and assignment process was followed by its funding source at the time, the ready-made flats were offered with low-interest loans to those families who proved not to own another property. These families were at liberty to choose a flat most suited to their resources and requirements, due to the variety of flats’ physical specifications, such as size, number of rooms, level, and type of private outdoor space. We found this variation beneficial for this study as it allowed us to examine the effects of a dwelling’s physical characteristics on personalization in combination with residents’ housing statuses and household demographics.

Armes is comprised of 756 flats divided into four identical clusters (see supplementary appendix 1 for the location and the aerial view of the Armes complex in Mashhad). We surveyed personalization instances in one of these clusters, comprising 189 households out of whom 133 families could be reached. Other inhabitants either did not answer the doorbell or refused to fully participate in the research. Each of the four clusters consisted of eight blocks ranging from three to five floors. Ground-floor flats enjoyed a separate front door, followed by a front yard, while other floors each consisted of two to seven flats being accessed by common stairs and a hallway. The flats, with areas ranging from 43 to 106 square meters (mean=72.36 m²), had between one to three bedrooms. Of surveyed flats, 85.7 percent had a private outdoor space in the form of a front yard, balcony, or rooftop terrace, with a mean area of 32.86, 6.11, and 41.25 m² respectively (Figures 1 and 2).

Figure 1 A general view of the surveyed area.

Figure 2 A three-story block comprising 16 apartments, with either a front yard, a terrace, or a rooftop terrace.

The interviews were conducted with the female head of each households², who provided us with their demographic data. Due to the different number of children in different households, we used the age of each interviewee as an indicator of the age of her children, as older mothers normally have older children. The variation in their financial ability, due to their reluctance to provide accurate information, was recorded using the same method Evans et al. (2003) employed, using housing value as its index. These values, estimated by the help of a local realtor, were categorized in ten levels and used as dwellers’ financial ability (Table 1).

Concerning independent variables, the data on housing statuses were collected through interviews, similar to demographic data. It included ownership type as two categories of owners (54.9 percent) and renters (45.1 percent); length of residence as a continuous variable (from one month to 11 years; mean= 4.15 years) and anticipated residence as an ordinal variable with five levels, including near-term (16.5 percent), short-term (44.4 percent), medium-term (15.0 percent), long-term (17.3 percent), and life-time residence (6.8 percent).³ Furthermore, each of the dwelling’s physical characteristics was collected based on architectural documents of the complex. They were measured as continuous variables, using the metric system when applicable, excluding the adjacency of a private outdoor space to the dwelling’s entrance which was registered as a dichotomous variable (0 as no, 1 as yes). As this condition only applied to the ground floor flats with front yards adjacent to their entrances, we named it the front yard entrance. Also, instead of the number of habitable rooms in each flat, the number of bedrooms was registered, since all households possessed one great room (living and dining room).

In registering dependent variables, what we called an instance of personalization was according to the type of each intervention, or simply its name, such as flowerpot, tree, guardrail, etc. (see supplementary appendix 2 for a list of instances of personalization and their frequency).

Initially, if each of these was recognized as a post-occupation intervention, it was counted as one instance, regardless of its number, volume, and the like. Afterward, for each instance, three measurements were made. Its category in the elements as well as settings of personalization was checked and counted as dichotomous variables, such as one fixed element and one primary territory. Then, the function of each instance was determined based on what the interviewee stated as her or his primary motive for implementing it, such as providing visual privacy which was registered as the preventive function of personalization. Although having observed that some personalization elements served more than one purpose, e.g. climbing plants both beautified the place and provided privacy, we recorded the one the dweller announced as the primary motive for employment. Afterward, through observation, we measured this function on a 5-degree ordinal scale from very low to very high according to the extent to which that instance affected and changed the character of its surroundings (Figure 3).

Figure 3 Each of these two examples of personalization with climbing plants in front yards was recorded as one primary territory, one architectural semi-fixed element, and facilitative personalization as its function. However, this function for the one on the right was measured five and the one on the left, two.

During the investigation, we found out that the category of semifixed elements included two subcategories. One consisted of those personalized items whose employment either demanded any kind of construction work and/or eventually became a part of the architecture of a house, like paving, painting, a replaced door, a tree, etc. The instances of the other subcategory, though, could be easily carried around the house or moved to another flat, such as flowerpots, shoe rack, free-standing cabinets, and wooden partitions. Therefore, we divided semifixed elements into two subcategories of immovable semifixed and movable semifixed ones.

Data were collected through observation and, at the same time, interview. Primarily, we prepared a form for recording and measuring the variables. A team of two trained female research assistants, then, went door to door, and following explaining the research objective and acquiring the permission of the dweller, filled a separate form for each household and took photos, if allowed, of personalization instances both inside and outside each flat. Female assistants were chosen as their same-sex with the interviewees could be instrumental in promoting trust and inclination to participate in the research. Also, their working together as a team prevented the error in the subjective measurement of personalization functions. They visited the complex and collected data on weekends so that they could contact both housewives and women with jobs. In each interview, following demographic questions, they asked the interviewee about the interventions the family had done in their flat, as well as their secondary and public territories, after occupation. They, then, observed and recorded these instances, according to the measurement guidelines described in section 2.2.

Overall, 49 types of personalization with Cronbach's alpha of 0.711 were recorded. Each type included three aspects concerning its function, element, and setting. As each of these three had their own three variables, nine sets of numbers were determined as dependent variables. Also, for the total amount of personalization performed in each flat to be determined, the three measurements of personalization functions were summed which produced the tenth dependent variable. But since no act of personalization was performed in the public territory, nine dependent variables remained, for each of which we performed nine separate stepwise regression analysis. There were a total of 12 independent variables in each test, including housing statuses (ownership type, length of residence, and anticipated residence), dwelling’s physical characteristics (indoor space size, outdoor space size, number of bedrooms, floor, front yard entrance), and household demographics (age of the female head of a family, household size, number of children, and financial ability) (table 2). For a regression analysis with 12 independent variables when a medium-size effect is expected, our sample size (n=133) was just above the minimum required number, based on Field’s estimation (2013).

During the analysis, we saw that some independent variables that had shown a significant bivariate correlation with the dependent one at the first stage of the test had been omitted from the final results. For example, although ownership type showed a 0.307 correlation (p<0.001) with the overall personalization, it did not appear in the results as a significant predictor because of its interrelationship with anticipated residence, a stronger predictor. Therefore, we repeated the same analysis twice more, but each time, we removed one significant predictor and observed what new predictors would appear. This process enabled us to determine which variables had been concealed behind the stronger influence of another predictor, which we called substitute predictors (see supplementary appendix 3 for substitute predictors).

We expected generally medium-size effects, though finally, this only applied to some first row predictors while the majority of significant correlations showed low-size effects. As this low-size effect calls for larger sample size, the possibility of type 2 error in our results increased. To mitigate this problem and raise the probability of the reported effects, we showed that the P-value for most correlations was lower than 0.001. Also, those correlations with the effect size less than 0.2 and p>0.01 were excluded from our discussion.