Neighborhood-Built Environment and Ageism in Later Life

Determinants of Ageism Against Older People

Following Butler’s (1969) introduction of the ageism concept, researchers have since identified factors associated with ageism against older people. Existing studies show that certain sociodemographic factors such as being female (Allan et al., 2014), older (Officer et al., 2020), more highly educated (Officer et al., 2020), and knowledge about aging (Cooney et al., 2021) were associated with a lower level of ageism. In addition, other studies have found that certain interpersonal factors, including presenting older people positively (Marques et al., 2020) and having better quality contact with older people (Hale, 1998), were associated with a lower level of ageism.

Recently, studies have focused on the contextual determinants and social environment of ageism. For example, Vitman et al. (2014) found that higher levels of social integration in a neighborhood were negatively associated with ageism. Moreover, studies consistently report a negative association between ageism and countries with a higher proportion of older people participating in paid and volunteer work (Bowen & Skirbekk, 2013) or with more available economic resources (Marques et al., 2020). However, the proportion of the older population in a country has yielded inconsistent results due to the difference in ageism measurements and definitions of older people (e.g., Löckenhoff et al., 2009; Officer et al., 2020).

The neighborhood-built environment is crucial for older people because they tend to spend more time in their immediate neighborhoods (Besser et al., 2017) due to lifestyle changes and age-related mobility decline (Yen et al., 2009a). Therefore, a growing number of studies have focused on the association between the built environment and ageism. For example, Van Hoof et al. (2020) studied the manifestation of ageism in the built environment from the age-friendly city framework and found that poor accessibility in buildings and public transport reflected the explicit form of ageism. In addition, Gamliel (2000) found that older people living in sheltered housing exhibit ageism against their own age group. Noon and Ayalon (2018) found age and gender segregation in public open spaces. From a macro perspective, Phillipson and Grenier (2021) investigated the relationship between urbanization and ageism, and found that privatizing land and public assets (e.g., public space, leisure centers, parks, etc.) in an urbanized society may intensify ageism. Although extensive efforts have been made to identify different determinants of ageism, existing studies on ageism are limited in two ways (Noon & Ayalon, 2018).

Firstly, we know very little about ageism among older people themselves (John, 2013). Most empirical studies on ageism focused on young people (Minichiello et al., 2000), students (Wang & Chonody, 2013), healthcare professionals, and social workers (Levy & Macdonald, 2016). As life experience affects ageism (Marques et al., 2020), we cannot extend the existing findings from these populations to understand ageism among older people.

Moreover, most existing studies on the built environment and ageism were focused on the theoretical association supported by qualitative evidence. For example, some adopted the age-friendly city framework proposed by the World Health Organization to examine the manifestation of ageism through environmental design or facilities (e.g., Chui et al., 2019; Van Hoof et al., 2020). Other researchers explored ageism through the lens of urbanization (e.g., Phillipson & Grenier, 2021). Only limited studies provided quantitative evidence on the association between the built environment and ageism. For instance, Zhong et al.’s (2022) study on the community-built environment found that the presence of religious destinations is positively associated with older people’s intra-generational interaction, which could reduce ageism. Therefore, more quantitative studies are needed to understand ageism’s determinants comprehensively.

Theoretical Rationale of this Study

A community-built environment can be associated with ageism through two pathways. Firstly, a well-designed built environment can enable higher levels of social interactions, which in turn, can reduce ageism by increasing perspective-taking and empathy, as suggested by the intergroup contact theory (Pettigrew et al., 2011). Intergroup contact theory could be applied to study intra-generation ageism, where older people disassociate themselves from being old and declining by perceiving the frail, retired, and inactive older people as an outgroup (Kydd et al., 2018). For example, Kwong and Yan (2021) found that people with quality face-to-face contact with older people displayed lower levels of ageism.

From this perspective, a built environment that facilitates social interaction in the community can help reduce ageism. For example, public parks have been identified as places with frequent intergroup contact (Neal et al., 2015). Public parks in Hong Kong are open to people of any age group with no specific restriction. Therefore, visiting public parks could bring people together and facilitate repeated social interaction (Neal et al., 2015; Peters et al., 2010). Similarly, community centers provide activity spaces for older people to engage in various hobbies or volunteer activities, promoting healthy behavior, and facilitating social participation and interactions (Hosokawa et al., 2019). Finally, older people are frequent users of public libraries in Hong Kong (Leisure & Cultural Services Department, 2015). The universal design with barrier-free facilities in Hong Kong public libraries ensures older people easy access (Leisure & Cultural Services Department, 2022), which creates opportunities for them to interact with others and offers opportunities for intergenerational activity (Institute of Museum and Library Services, 2012). As a result, the built environments which promote positive social interactions could be associated with lower ageism (Zhong et al., 2022).

Secondly, the stereotype embodiment theory proposes that individuals internalize ageism by repeatedly being exposed to negative stereotypes of older people in their life course (Levy, 2009). These stereotypes could be reinforced by the negative presentation of older people in daily life (Donlon et al., 2005). Previous research has confirmed that negative representation of older people generates a poor impression of older people and increases ageism (Gekoski & Knox, 1990), whereas positively presenting older people can lower ageism (Marques et al., 2020).

People may be exposed to different environmental features that promote negative aging stereotypes in their community. For example, nursing homes were perceived as a place of abandonment and a stop before death (Gamliel, 2000; Villeneuve et al., 2022). Furthermore, nursing home residents are often portrayed as frail, dependent, desperate, and lonely (Villeneuve et al., 2022). Similarly, clinics and geriatric wards represent the ill and sickness in aging. Thus, it is plausible that individuals who have continuous exposure to these built environments may develop a higher level of ageism (Davidovic et al., 2007).

Based on the intergroup group contact theory and the stereotype embodiment theory, we hypothesize that older people’s ageism against their own age group is negatively associated with the number of public facilities that enhances social interactions. Therefore, parks, libraries, and community centers, which allow interactions, would associate with lower ageism. We further hypothesize that ageism is positively associated with the number of clinics and nursing homes, which present older people as frail, weak, and dependent (Figure 1).OPEN IN VIEWER

Finally, different socioeconomic status (SES) groups might have different interaction patterns with the environment and be affected differently by environmental attributes. For example, some studies show that lower SES groups may rely more on their immediate environment because they tend to travel across a smaller area, walk shorter distances for recreation, and visit fewer activity places than people with higher income (Tao et al., 2020). More importantly, a small but growing number of studies have found that the SES of a community is associated with ageism. For example, Marques et al. (2020) found that communities with more available economic resources recorded lower levels of ageism among their residents. In addition, despite no existing research on the impact of community-level education on ageism, the strong linkage between an individual’s education and ageism (e.g., Chopik & Giasson, 2017; Maurya et al., 2022) may hint at the potential association between community-level education and ageism. As a result, the impact of the community’s SES (e.g., income and education) on ageism warrants further examination.

To fulfill the knowledge gaps mentioned above, the objectives of this study were to examine: (1) the association between the objectively measured built environment features and ageism among community-dwelling older people; and (2) variations in such associations between neighborhoods with different levels of income or education.

The Study Site: Hong Kong, China

Hong Kong is a Chinese city with high levels of economic development, urbanization, and population density. In 2021, 20% of its population were 65 years or older and the figure will increase to about 33% by 2039 (Census and Statistics Department, 2020; 2022b). Although many believe that Hong Kong people share the traditional filial culture that respects older people and have a more positive attitude toward them, studies found otherwise (Lou & Dai, 2017). In fact, a recent study found higher ageism in Hong Kong than in other Western countries (Löckenhoff et al., 2015). With this context, Hong Kong provides a suitable place to understand ageism in rapidly developing Asian cities.

Study Design and Respondents

This cross-sectional study comprised 1333 people aged 60 years and above recruited by purposive and snowball sampling. We recruited participants from 44 centers for older people in three districts of Hong Kong. These centers are funded by the Government and provide healthy aging and leisure activities to older residents in the district (Social Welfare Department, 2022). Although membership is open to all district residents, most members were from the lower to middle socioeconomic classes. We therefore supplemented the sample by snowball sampling to increase the number of participants from upper middle socioeconomic status. The snowball participants came from volunteers in aging programs of the center-on-aging of the university. We successfully recruited a minimum of 400 participants in each district. Most older people were members of multiple centers, and we were not able to associate them with a specific center. Therefore, we did not use center characteristics as control variables in our multivariate analysis.

The inclusion criteria were (1) people aged 60 or above; and (2) able to understand and communicate in Cantonese. Domestic helpers or individuals with a medical history of dementia diagnosed by medical professionals were excluded. Domestic workers were excluded as they were not permanent residents of Hong Kong. People with a diagnosis of dementia by a medical professional were excluded because of their memory challenge as we asked participants to answer survey questions retrospectively. Data were collected via self-administered questionnaires with the assistance of trained research assistants. Face-to-face interviews using the same questionnaire were conducted with respondents who were illiterate. Participants had given informed written consent before the interview began. A total of 55 (4.0%) people were excluded from the statistical analysis because of missing data. The final sample comprised 1278 individuals. The study was approved by the Human Research Ethics Committee of The University of Hong Kong [Reference Number: EA1510033].

Dependent Variable—Ageism

Our study focused on older people’s ageism against their age group (intra-generation ageism) and we used the short version of Kogan’s Attitudes towards Older People Scale (KAOP) to measure it (Kogan, 1961b). KAOP has been used in various ageism studies to assess older people’s ageism against their own age group (e.g., Kogan, 1961a; Salguero et al., 2019).

The original KAOP has 34-items consisting of 17 matched pairs. Each pair has one negatively phased statement (KAOP-) and one positively phased statement (KAOP+) on older people (see Kogan, 1961b). Respondents are asked to rate their agreement with the statement using a six-point Likert scale ranging from 1 (strongly disagree) to 6 (strongly agree). To obtain the total score, the 17 KAOP-items are reverse coded and summed with the KAOP + score. The possible scores range from 34 to 204, and a higher score indicates a more positive attitude toward older people (i.e., a lower level of ageism) (Kogan, 1961b). The scale has been validated in a Chinese population (Yen et al., 2009a). The short version of KAOP used in this study included 10 items selected from the original KAOP based on a factor analysis conducted by a team with another larger dataset (Sun et al., 2019). Cronbach’s alpha of the short-version scale was 0.80 in this dataset, demonstrating good reliability. The first nine negative items were reverse coded, and the score of these 10 items was summed to generate the total scores. The possible score ranged from 10 to 60, with a higher total score indicating a lower level of ageism.

Independent Variable—Neighborhood-Built Environment

The neighborhood-built environment variables included the number of (1) community centers, (2) libraries, (3) parks, (4) nursing homes, and (5) clinics within the 200-meter buffer area of the respondent’s residential address. They were selected because they could facilitate social interactions (from the intergroup group contact theory) or promote negative aging stereotypes in the community (the stereotype embodiment theory). Although public transportation affects older people’s well-being and opportunity for social interaction (Van Hoof et al., 2020), we did not include it in our study as all districts were served by extensive public transportation services and older people enjoy a significantly discount on bus ($0.25 US dollar per trip) and on subway (half priced) services in Hong Kong.

Each respondent’s residential address was geocoded to latitude and longitude automatically using Map Location (https://maplocation.sjfkai.com/). We used Google Map to geocode the addresses manually should the address could not be geocoded automatically using Map Location. The neighborhood-built environment was assessed within a 200-meter road network and slope-adjusted buffer area of the respondent’s residential address to ensure maximum closeness to reality, thus significantly improving the precision of objective measurement. The 200-meter buffer area was chosen to capture the neighborhood-built environment within a five-minute walk of older people’s residences (Gehl, 2010). The 2014 GeoReference Database (scale 1:1000) (including the road network data), the 2014 digital topographic data (Scale 1:1000) (showing the topography of terrain), and the 2014 GeoCommunity Database (including exact location points of various types of facilities) provided by the Hong Kong Government’s Lands Department were used to create the objective measures. ArcGIS version 10.7.1 was used for processing the geographical data.

Covariates

Control variables included age (years), gender (male and female), education (no education, primary school, junior high school, senior high school, and post-secondary education), individual monthly income (HKD <$2000, $2000-3,999, $4000-5,999, $6000-$7,999, $8000-$9,999, and ≥$10,000) ¹ , physical health, mental health, and percentage of older people in the sub-district. Physical health and mental health were measured by the Physical Component Summary (PCS) and Mental Component Summary (MCS) scale of the 12-item Short-form Health Survey (SF-12). The SF-12 is a shortened version of the 36-item Short-form Health Survey (SF-36). It extracted 12 questions from the SF-36 with the same eight domains in physical health (i.e., physical functioning, role-physical, bodily pain, and general health) and mental health (i.e., vitality, social functioning, role-emotional, and mental health) (Ware et al., 1996). Twelve items were weighted separately and summated to generate the standardized PCS-12 and MCS-12 scores (Ware et al., 1996). The norm-based standardized PCS and MCS score were used in this study, in which higher scores indicated better physical and mental health. The SF-12 has been validated in a Chinese population (Lam et al., 2005). The percentage of older people in the sub-district was retrieved from the district profiles generated by the Hong Kong Census Department based on the Hong Kong Population By-census 2016 (https://www.bycensus2016.gov.hk/en/bc-dp.html).

Statistical Analysis

Firstly, we divided each sub-district based on whether their average income/education level was higher or lower than the district median income/educational level (from 74 sub-districts in this study). The division of the sub-district allowed us to test the potential moderation effect of district-level socioeconomic status on the association between environmental variables and ageism. The district and sub-district median incomes and educational levels were obtained from the Census and Statistics Department of the Government (Census and Statistics Department, 2017).

Secondly, we reported sample characteristics and examined the differences between groups using chi-square and t-test. Moreover, we used the intra-class correlation coefficient (ICC) to test the clustering effect between different sub-districts. The result of ICC was close to zero (<0.001), indicating that no significant clustering effect was found among sub-districts. Therefore, we used multivariable linear regression modeling to examine the association of our environmental variables with ageism after controlling for the covariates (including individuals’ income and education). The variance inflation factor (VIF) ranged from 1.04 to 1.62, and all were lower than 10, indicating that multicollinearity was not a concern (Hair et al., 1995). Finally, we repeated the main regression in each subgroup of income and education to identify the moderation effect of district-level income and education on the association between environmental variables and ageism. Statistical analysis was undertaken using SPSS 26.