Introduction

Older persons are gradually contributing to a greater share of the overall population due to increasing life expectancy. In fact, by 2050 older persons are predicted to account for 21% of the global population (United Nations, 2017). To meet the growing care needs among older persons, ageing in place policies are encouraged whereby support and resources are provided for people to continue living in their home and community (Pani-Harreman et al., 2021). While ageing in place is preferred by policymakers and many older persons, its success depends on the unpaid work of family carers who sometimes experience strain and burnout (Jacob et al., 2020; Sixsmith et al., 2014, 2017; Temple & Dow, 2018, Vreugdenhil, 2014). Issues such as time squeeze, role strain, poor health, and poverty are also reported (Chou et al., 2011; Lai, 2012; Longacre et al., 2017; Marino et al., 2020; Temple & Dow, 2018). To date, the perspective of family carers’ regarding their role in home care policies is underexamined. Additionally, while policies speak to the family as a unit in the provision of home care, in practice care is provided by individuals within the family and other informal networks (Barrett et al., 2016; Pani-Harreman et al., 2021), making the individual perspective more relevant. The subpopulation of older persons is not a homogenous group and care needs among its members vary. Importantly, the intersectionality between ageing and disability could result in specialised care needs (Watchman, 2018). Disability occurs more frequently in older ages due to functional limitations, frailty, and a higher incidence of falls, and Persons With Disabilities (PWD) are more likely to experience poorer health outcomes throughout their lifecourse (Cesari et al., 2017; Slaug et al., 2011; United Nations, 2019; van Schooten et al., 2015). Older Persons With Disabilities (OPWD) are therefore likely to require high amounts of care (Barrett et al., 2016).

This study measures care views towards OPWD in a Caribbean country, namely, Trinidad. Notably, by 2050 the predicted proportions of older persons in the Caribbean and Trinidad (and Tobago) are 25% and 28%, respectively, which is higher than the global average (United Nations, 2017). Care among OPWD in the Caribbean is of particular concern given the continued exclusion of PWD from education systems and the association between low education outcomes and greater care needs (ECLAC, 2012; Morris, 2018; Parey, 2021a, b; Yue et al., 2021). Investigating care views among Trinidadians could provide useful insights regarding care as a global issue since home and community care, which only became internationally prevalent within the last two decades, have been the traditional response to care in many developing countries and, moreover, within the context of limited public policy support (ECLAC, 2012; Ofori-Dua, 2014; Rawlins, 2015). Trinidad has a multi-ethnic composition, including African, East Indian, mixed, and other ethnicities (Central Statistical Office, 2012) which is reflective of Caribbean societies. As with other countries, ageing in the Caribbean is occurring in the context of changes in family structure and higher labour participation by women, which means the family model for social care is no longer reliable (Barrett et al., 2016; Destremau, 2021; ECLAC, 2012; Jones, 2016). Indeed, the Caribbean care literature focuses on intergenerational support in the context of changing family compositions, the imbalance regarding demand and supply of carers, and the fiscal challenges of caring for the older population (Destremau, 2021; Palloni & Pinto, 2014; Rawlins, 2015; Quashie, 2014; Quashie & Zimmer, 2013; Theodore et al., 2017). Additionally, among Trinidadians, the cost of caring is a concern and relatives are unable to provide adequate care to OPWD due to work and other family obligations (Parey, 2021a, b).

Care was initially seen as a private activity mainly performed by women and focused on specific dependents. The definition of care now covers a range of human activities which both men and women have the potential and responsibility to perform (Tronto, 2013). In the resulting ethic of care framework, care is seen as a universalist principle, and the rights to receive care, give care, self-care, and to reject care work are equally acknowledged (Tronto, 2013). The ethics of care framework is therefore relevant when investigating individual views. There are also concerns that carers risk experiencing the reduced circumstances and vulnerability of the persons whom they care for (Barrett et al., 2016). In this regard, consideration to the social and economic spheres within which care is embedded is required in order to accurately frame care and its relationship to public policy (Barrett et al., 2016). Care, initially defined as an emotion and a form of work, was additionally acknowledged as an activity of relationships within the ethics of care framework (Fine & Glendinning, 2005). A three-dimensional concept is therefore evident, consisting of emotion, form of labour, and social relationship (Fine & Glendinning, 2005). This three-dimensional concept is very similar to the one developed by Daly and Lewis (2000), where care is seen as a duty, labour, and cost. Since Daly and Lewis (2000) discuss their conceptualisation within the social and economic contexts of care, corresponding dimensions are useful in obtaining knowledge of care views.

Relevance and applicability of Daly and Lewis’ (2000) three dimensions to measure care views are supported by the literature. Care as labour considers the nature of the work and draws comparison with other types of work (Daly & Lewis, 2000). There is evidence that the time spent on care activities affects current work hours if carers are employed or carers are hindered from generally engaging in other forms of work (Carers UK, 2019; Longacre et al., 2017; Mudrazija, 2019). Care as a relationship embedded in obligation recognises the role of familial relations and responsibilities (Daly & Lewis, 2000). According to the literature, regardless of the level of support required, some carers view care as an obligation (Gill, 2020; Parveen et al., 2013). As a cost incurring activity, consideration is given to emotional and financial costs (Daly & Lewis, 2000). Care has led to emotional strains for carers and is likely to incur costs, such as meals, transportation, house modifications, medication, and assistive equipment (Parey, 2021b; Barrett et al., 2016; Longacre et al., 2017; Roy et al., 2018; Schulz et al., 2016). Factors driving care views also feature in the literature. Previous research reported on the influence of ethnicity shaping the experience of carers (Branin & Juarez, 2013; Burch et al., 2019; Parveen et al., 2013). Other factors regarding care views, such as education qualifications, and experiences with persons with disabilities, are less frequently examined. Regarding employment status, work participation negatively impacts on the willingness of persons to assume carer roles (Carmichael et al., 2019). Gender is also an essential factor to consider for several reasons, including the initial gendered nature of care and the persistence of gender imbalances in more developed countries (Pavalko & Wolfe, 2016). Scholars moreover argue that care policies relocate the responsibility of care from the state to women within households (Destremau, 2021; Pavalko & Wolfe, 2016). In the Caribbean context, women have historically assumed the role of carers (ECLAC, 2012; Rawlins, 2015). The explicit influence of gender on care views therefore warrants examination. Investigating the care views towards OPWD could provide recommendations to support the care needs of the ageing population, and to this end, the study seeks answers to the following research questions:

  1. 1.

    How is care of OPWD viewed among Trinidadians?

  2. 2.

    What is the prevalence of these care views among Trinidadians?

  3. 3.

    Does gender influence these care views among Trinidadians?

  4. 4.

    What are the factors associated with the various care views among Trinidadians?

Methodology

Ethical Considerations

Ethical approval was granted by the Campus Research Ethic Committee of the University of the West Indies. To encourage participation in the research, a waiver of formal consent was sought and granted. One should note that all participants were informed of the voluntary nature of the research, and anonymity and confidentiality were ensured.

Questionnaire

The study design was cross-sectional. The questionnaire was developed by the author and was based on descriptions of the care concepts proposed by Daly and Lewis (2000), the ethics of care framework, and the lived experiences of the author, namely a background in caring for an older person with a disability. Altogether, the questionnaire consisted of 15 statements (see Table 1) rated on a 6-point Likert scale (0-strongly disagree, 1-disagree, 2-somewhat disagree, 3-somewhat agree, 4-agree, 5-strongly agree). Care as duty was captured by the statement taking care of an older family member with a disability is my duty. Care as labour referred to interruption to work activities which could manifest as a disruption to work or job quit (Carers UK, 2019; Longacre et al., 2017; Pavalko & Wolfe, 2016). Statements pertaining to participation in social activities, achievement of personal goals, ability to spend time with other relatives, involvement in other family matters, and overall life progress were included as other aspects of life are also disrupted and since self-care is acknowledged within the ethics of care framework. Emotional costs (Daly & Lewis, 2000) and mental and physical strains (based on the author’s experiences also supported by the literature Branin & Juarez, 2013; Chou et al., 2011; Jacob et al., 2020; Jacob et al., 2020; Pavalko & Wolfe, 2016; Temple & Dow, 2018) were captured in statements for both the individual and other relatives. Financial strain (Lai, 2012; Pavalko & Wolfe, 2016; Temple & Dow, 2018) was captured by taking care of an older family member with a disability will be a financial burden.

The questionnaire also included some demographic variables: gender, household position, age, highest level of completed education, ethnicity, employment status, and experiences with PWD such as friends with disabilities, OPWD, other relatives with disabilities, and other PWD (see Table 2 for the categories of each variable). The rating of the experiences with OPWD and other PWD were also captured using a 4-point Likert Scale (1-very good, 2-good, 3-bad, 4-very bad).Footnote 1 To obtain accurate responses the following definitions were included at the start of the questionnaire:

  1. 1.

    Care as “the provision of what is necessary for the health, welfare, maintenance, and protection of someone (Oxford Dictionary).

  2. 2.

    An older person as someone 65 years or older (National Insurance Board, 2014).

  3. 3.

    Disability as any or a combination of physical, mental, intellectual, sensory impairments (United Nations, 2006).

Sampling

A minimum sample size (n = 887) was calculated using the formula, n = (z^2) * r * (1-r) * f * k / [ p * ñ * e^2], where z is the level of confidence, r is the estimate of a key indicator, f is the sample design effect, k is non-response multiplier, p is the target population proportion, ñ is the average household size and e is the margin of error (United Nations, 2008).Footnote 2 The Central Statistical Office used the Continuous Sample Survey of Population (CSSP) framework to provide the sample using a 2-stage sampling process. The CSSP sample frame for Trinidad consisted of fourteen administrative areas. At the first stage, enumeration districts were sampled from each administrative area with probability proportional to size. Notably, for each administrative area, enumeration districts were assigned a skill code in relation to the percentage of economically active persons. They were ordered in ascending order of their skill codes before a random sampling technique was applied to ensure representativeness. Overall, 156 were selected from a total of 2324 enumeration districts. The enumeration districts consisted of 150–200 households and varied in size. At the second stage, households within each selected enumeration district were randomly sampled with a probability inversely proportional to size. The resulting sample size was 893 households. Within each household, one person over the age of 18 years participated in the study. If a single person was present at the time of data collection, they provided responses to the questionnaire, and in cases where more than one household member was present, the person (most) responsible for the decision making within the household responded.

Data Collection and Analysis

A pilot study was conducted in July 2017. Data from 16 persons from 16 different households indicated a high reliability coefficient (Cronbach alpha = 0.89). The Central Statistical Office was hired to collect the data which occurred between October through December 2017. The response rate was 99.78% (n = 891). The data was cleaned and analysed using STATA version 15. Any observation with a missing value was deleted. The number of observations used in the analysis was 868, which corresponded to a 3.32% margin of error and a 95% confidence level for the population of Trinidadians aged 18 years and older. Factor analysis was carried out to determine the care dimensions. Overall, care dimensions contained either one item, three items, or five items. The reliability coefficient for the factors and the overall scale were determined. Subsequently, descriptive statistics for the various care dimensions were calculated. Corresponding to 6 ratings on the Likert scale, the range of scores for each dimension was divided into approximate thirds in order to determine the overall prevalence for each care dimension. More specifically, 0- < 2, 2- < 4, and 4–5 represented low, medium, and high prevalence for single-item dimensions, 0- < 5, 5- < 10, and 10–15 for three-item dimensions, 0- < 9, 9- < 17, and 17–25 for five-item dimensions.

In terms of the multivariable analysis, the scores for each dimension were used as the dependent variable in the regression analysis model. For the care factors, the score of the individual items were summed. Since these factor scores were synonymous with count data with an upper bound, a negative binomial quasi maximum likelihood estimator (NBQMLE) was employed (Wooldridge, 2002). Here, as is standard for non-linear regression models, the marginal effects were calculated at the means of the explanatory variables using the delta method, and these marginal effects, rather than the estimated coefficients are reported in the results section (Wooldridge, 2002). For the single-item care dimensions, ordered probit regression analysis was employed (Wooldridge, 2002). As with the NBQMLE, the sign of the coefficient in the ordered probit regression models indicates the direction of the effect of the explanatory variable in question on the probability of the care dimension, but does not determine the magnitude of the effect on the care dimension (Wooldridge, 2002). To provide information on the magnitude of the probability for each category of the various care dimensions, the marginal effects are calculated at the means of the other variables for the significant variables only.

Note that in order to allow for a possibly non-linear relationship of age and the various care views, squared age was also included as an alternative model for all regression models. Additionally, to determine the influence of the rating of experiences with OPWD and other PWD on the care dimensions, regression analysis was conducted for the subsample of respondents with the respective experiences. To explicitly examine any difference in terms of gender on the care dimensions, a likelihood ratio test was carried out for the benchmark regression model against the model containing the benchmark variables together with the interaction terms of those variables with gender. For the care dimensions where the benchmark model and the model inclusive of the interaction terms were statistically different, regression models were estimated separately for the male and female subsamples. Also, in order to roughly evaluate the predictive power of the regression models, the predicted score, as implied by the estimated coefficients, was compared to the actual score, allowing for a five-point margin of error. The percentage of correctly predicted scores was used as an indicator of the predictive power of the regression models. For all models, indicator variables were created for ethnicity and education. In terms of ethnicity, African ethnicity was the base category, and for education, primary school education was the base category.

Results

Descriptive Statistics

Table 1 displays the descriptive statistics regarding sample characteristics. Four hundred and sixty-eight women participated (53.92%), and about 57% of the respondents were household heads. Their ages averaged about 50 years. About 84% of them had secondary school education or less. Approximately 38% and around 40% identified themselves as having African or East Indian ethnicity, respectively, 20.51% identified themselves as mixed, and the remaining persons had other ethnicities. More respondents were employed (51.84%) than unemployed. About 3% identified as having a disability, almost 7% lived with an OPWD, approximately 13% had friends with disabilities, and around 24% had experiences with other PWD. For the subsamples of respondents who had experiences with OPWD and other PWD, about 80% of each subsample rated these experiences are very good or good.

Factor Analysis and Reliability Coefficient

Using factor analysis and a factor loading cut-off of 0.6 (Field, 2005), three factors were obtained as seen in Table 3: cost to the individual (three items), cost to other relatives (three items), and care as labour (five items). Care as obligation, care as a financial burden, and another care as labour dimension was measured with one item each. Altogether these 14 items had a reliability coefficient of 0.88, which indicated good reliability of the questionnaire to measure care views towards OPWD (Cronbach, 1951).Footnote 3 All three care factors also had reliability coefficients greater than 0.8 (Table 2), and all three one-item dimensions had uniqueness values greater than 0.6 (Table 3). Statistics for each dimension are given in Table 2 and the regression analysis results are presented in Tables 4–5. Notably, the estimated predictive power of each regression model is greater than 80%.

Care As a Cost to the Individual

Care as a cost to the individual refers to emotional, mental, and physical strains. As seen in Table 2, care as a cost to the individual had a mean score of 8.64 (σ = 3.69) indicating medium prevalence among Trinidadians. In terms of the regression analysis, as displayed in Column 1 of Table 1, age (α < 0.01) and employment status (α < 0.10) were the only two significant variables for the entire sample. As persons grew older, their view regarding individual cost increased but just very marginally. Persons who were employed regarded care as a cost to the individual more than persons who were unemployed, but the count difference was about 0.5. For the subsample of respondents living with OPWD (Column 2 of Table 1), as the rating of the experience became positive, their view regarding individual cost lessened by about 3.2 counts (α < 0.01). Ethnicity (α < 0.10) and experiences with other PWD (α < 0.01) were also significant among the subsample (Column 2 of Table 1). Respondents living with OPWD who identified with mixed ethnicity held the view regarding individual cost about 2.6 counts more than respondents living with OPWD of African ethnicity (base category). Those living with OPWD who had experiences with other PWD viewed care as a cost to the individual about 2.9 counts more than persons living with OPWD without other experiences with PWD. For the subsample of respondents who had experiences with other PWD (Column 3 of Table 1), the rating of the experience and age were significant predictors (α < 0.01). As the rating of the experience became positive, the individual cost view lowered by about 1.7 counts, and as age increased, the view increased very marginally.

Care As a Cost to Other Relatives

Other relatives could also experience mental, emotional, and physical strains. The mean score for care as a cost to other relatives was 7.52 (σ = 3.66) which corresponded to medium prevalence (Table 2). Regarding the regression results, gender was a significant variable (Column 4 of Table 1). Compared to women, men viewed care as a cost to other relatives more (α < 0.05) with a marginal effect of less than one.4 Age (α < 0.10), education (α < 0.01), and ethnicity (α < 0.10) were also significant predictors of the view regarding cost to other relatives (Column 4, Table 1). Respondents with post-secondary or tertiary education viewed care as a cost to other relatives more than respondents with primary school education (by 1.2 counts), and respondents of Caucasian, Syrian, and Chinese ethnicities viewed care as a cost to other relatives less than respondents of African ethnicity (by 2.2 counts). As age increased, the view of cost to other relatives increased with a marginal effect of less than 1 count.

Among the subsample of respondents living with OPWD, the rating of the experience was insignificant. The relationship for age was non-linear for the subsample, i.e., age and age squared were both significant (α = 0.05; Column 5, Table 4). The coefficients on these implies that from the age of 18, the view of cost to other relatives lessened until the age of 52 and then marginally increased afterwards (α < 0.05). Also, respondents living with OPWD who had other relatives with disabilities (α < 0.10) held the cost to other relatives view by two counts less than respondents living with OPWD who had no other relatives with disabilities. Respondents living with OPWD who had experiences with other PWD (α < 0.01) held the view more than respondents living with OPWD who had no experiences with other PWD by four counts. For the subsample of respondents who had experiences with other PWD, the rating of the experience was insignificant, while age (α < 0.10), education (α < 0.05), and experiences with OPWD (α < 0.10) were significant (Column 6, Table 1). The variable age had a marginal effect of less than 1, while respondents with post-secondary or tertiary education and those living with OPWD held the cost to other relatives view more than their respective base counterparts (by 1.7 and 1.5 counts, respectively).

Care as Labour I

Care as labour represents disruption to personal, family, social, and work matters, as well as overall progress in life. Its mean score of 11.01 (σ = 4.99) corresponded to medium prevalence (Table 2). The likelihood ratio test which compared the benchmark model and a model with the independent variables interacting with the gender variable produced a statistically significant chi-squared test statistic (α = 0.01). This indicated that the sample should be split into separate male and female samples and the benchmark model re-estimated on these. The results of the benchmark model for the entire sample are displayed in Columns 1–3 of Table 6, and the results for the male and female samples are shown in Table 7.

In the benchmark model only the variable experiences with OPWD variable was significant for the entire sample (α < 0.05), where respondents living with OPWD viewed care as labour less than their counterparts by one count (Column 1-Table 6). When care as labour was examined among men and women (Columns 1–2, Table 7), experiences with OPWD was only significant among women (women living with OPWD viewed care as labour 2.7 counts less than their female counterparts). Women with secondary school education also viewed care as labour 1.4 counts more than women with primary school or no education, and men of Caucasian, Syrian, and Chinese ethnicities thought care was labour 2.9 counts less than men of African ethnicity.

Among the subsample of respondents living with OPWD, as the rating of the experience became more positive, the care as labour view decreased by three counts (α < 0.01; Column 2 of Table 6). When looking across gender among the subsample (Columns 3–4 of Table 7), the significant effect was only present among women. Also, for the subsample, age, ethnicity and having friends with disabilities were significant predictors for care as labour (Column 2 of Table 6). These variables, however, were only significant among women (Columns 3–4 of Table 7). Women of mixed ethnicity living with OPWD had a higher score by four counts than women of African ethnicity living with OPWD (α < 0.01), and women living with OPWD who had friends with disabilities had a higher score by 4.1 counts (α < 0.01) than their female counterparts. Having experiences with other relatives with disabilities was also significant among women (α < 0.01); women living with OPWD with such experiences viewed care as labour three counts less compared to their female counterparts (Column 3, Table 7). Notably, in the subsample of men living with OPWD, men with secondary school education viewed care as labour about eight counts less than men with primary school or no education (α < 0.05; Column 4, Table 7).

For the subsample of respondents who had experiences with other PWD (Column 3 of Table 6), the rating of the experience was significant (α < 0.01), and further examination revealed significance among both men and women (Columns 5–6, Table 7). As the rating of the experience became positive, the care as labour view lessened by about 1.8 and 2.2 counts among women and men, respectively. Ethnicity also significantly predicted the score for the care as labour view (α < 0.10) among the subsample of respondents who had experiences with other PWD (Column 3 of Table 6), however, only among men (α < 0.05; Columns 5–6, Table 7). Men of Caucasian, Syrian, and Chinese ethnicities who had experiences with other PWD viewed care as labour 5.3 counts less than men of African ethnicity who had experiences with other PWD. As seen in Columns 5–6 of Table 7, employed women who had experiences with other PWD viewed care as labour about 1.8 counts less compared to their counterparts (α < 0.10), and men with disabilities who had experiences with other PWD thought care was labour about five counts less than men without disabilities who had no experiences with other PWD (α < 0.05).

Care as Labour II

The second care as labour dimension was represented by a single item and reflects the view that persons would have to quit their jobs to care for OPWD. The mean score was 1.49 (σ = 1.27) corresponding to low prevalence (Table 2). As seen in Column 4 of Table 6, gender (α < 0.10), education (α < 0.01), employment status (α < 0.01), and household position (α < 0.05) were significant variables for care as job quit. As seen in Table 8, the marginal effects indicate that compared to respondents with primary school or no education, those with post-secondary or tertiary education were 10.7% more likely to strongly disagree with the job quit view (Table 8). Also, employed respondents were 6% and household heads were 5.3% more likely to strongly disagree with the job quit view. Men were 4% more likely to strongly disagree with job quit compared to women. There were no significant variables among the subsample of respondents living with OPWD (Column 5 of Table 6). For the subsample of respondents who had experiences with other PWD, education and employment status had the same directional effect as with the entire sample (Column 6, Table 6).

Care as Duty

Care as an obligation had a mean score of 3.73 (σ = 1.12) corresponding to medium prevalence (Table 2). The regression coefficient and the marginal effects for the significant variables are presented in Columns 1–3 of Table 5 and Rows 12–21 of Table 8, respectively. Respondents of East Indian ethnicity were 6% (α < 0.05) and respondents living with OPWD were 14.1% (α < 0.01) more likely to strongly agree that care was an obligation compared to respective base counterparts. Persons with secondary school education were 5% less likely to strongly agree with care as an obligation compare to persons with primary school education or less (α < 0.10). As the rating of the experience with OPWD became more positive, respondents living with OPWD were 43.2% more likely to strongly agree that care was an obligation (α < 0.01). Respondents living with OPWD who had friends with disabilities were 50% less likely to strongly agree that care was a duty compared to respondents living with OPWD who had no friends with disabilities (α < 0.10). For the subsample of respondents who experiences with other PWD, as the rating of experiences became more positive, respondents in the subsample were 11.2% more likely to strongly agree that care was an obligation (α < 0.01).

Care as a Financial Burden

Caring for OPWD can incur a financial cost. The view had a mean score of 2.84 (σ = 1.51), indicating medium prevalence (Table 2). Columns 4–6 of Table 7 and Rows 22–34 of Table 8 display, respectively, the regression coefficients and marginal effects for the significant variables. Respondents of East Indian ethnicity were 5.9% (α < 0.01), respondents identifying with mixed ethnicity were 3.8% (α < 0.10), and respondents with other ethnicities were 12.1% (α < 0.10) less likely to strongly agree with the financial burden view compared to respondents of African ethnicity. Also, respondents living with OPWD were 10.8% less likely (α < 0.01), and respondents with disabilities were 14.6% more likely (α < 0.01) to strongly agree with the financial burden view compared to their respective counterparts. For the subsample of respondents living with OPWD, as the rating of the experience increased, respondents were 32% more likely to agree with the financial burden view (α < 0.01). Having friends with disabilities also increased the likelihood of the financial burden view (α < 0.10) among the respondents living with OPWD. For the subsample who had experiences with other PWD, as the rating of the experience increased respondents were 8.2% more likely to strongly hold the financial burden view (α < 0.01), and respondents living with OPWD were 11.6% less likely to think care was a financial burden when compared to respondents not living with OPWD (α < 0.05).

Discussion

A 15-item instrument was developed to measure care views towards OPWD. Bias in questionnaire development was possible as it was partially based on the author’s experience. The literature, however, supported these experiences, and the high Cronbach alpha value indicated a reliable instrument. There were six care views altogether based on 14 items. Care as labour where persons must quit their jobs was the least prevalent, and all other views including care as an obligation had medium prevalence. Trinidadians having a multidimensional view of care and simultaneously viewing care as an activity where carers do not necessarily have to quit their jobs means increasing challenges regarding the balance of care work and employment (Barret et al., 2016; Pavalko & Wolfe, 2016). Current policies should be revised to reflect the multiple care views. This is particularly important as support corresponding to the various care views means recognising the various rights within the ethics of care framework. The traditional model of social care which relies heavily on the unpaid work of women is no longer sustainable (Barrett et al., 2016; ECLAC, 2012). Gender equity should feature in policies as results hinted that traditional gender roles might still persist specifically regarding the view of care as a disruption to life activities where distinct implications for men and women were evident. Importantly, secondary school education seems to play an important role in balancing the care roles as women perhaps felt more empowered and more aware of their possibilities, while secondary school education coupled with experiences with OPWD made men feel more socially responsible.

Findings arguably indicate the importance of experiences with PWD since its positive effect was prevalent among both men and women. Currently in Trinidad, while there is some inclusion at schools and workplaces, the majority of working-age persons and children with disabilities are segregated from their peers or entirely denied the opportunity to employment and education (Parey, 2020, 2021a). It is possible that increasing inclusion of PWD in society could have a positive influence on people’s outlook regarding care of OPWD. As a matter of fact, some regression results indicated more positive outlooks of care among persons living with OPWD. However, the effect of external experiences coupled with experiences with OPWD led to less favourable care views. As such, one might argue that perhaps exposure to PWD in various contexts is critical to ensure balance among the rights to self-care, receive care, give care and to reject care work.

There might be differing levels of care among various ethnicities, with possible negative implications for OPWD in households of predominantly African and mixed ethnicities. Due to the different effects among ethnic groups, more personalised caregiver interventions and supports for various ethnic groups are recommended (Branin & Juarez, 2013). For predominantly African and mixed ethnicity households perhaps the option of home and/or community care services could offset the possible consequences of their views. Notably, the views of East Indian respondents were similar to the reported high familism among British South Asian carers, (Gill, 2020; Praveen et al., 2013). For the latter flexi-time and work-from-home options, singly or combined, seem more ideal as they were more positive towards care of OPWD. Ultimately, the choice of care services should be left up to OPWD and their families (both men and women), and hence a variety of options should be made available. Importantly, respite services, day care facilities and other coping interventions are needed to ensure sustainable care at home (Giles et al., 2015; Parveen et al., 2014; Rawlins, 2015).

Persons with higher levels of education and employed persons are less likely to assume the role of carer. Interestingly, the literature indicates that employed carers completely leave work or reduce work hours to provide care, which ultimately leads to poverty among them (Carers UK, 2019; Pavalko & Wolfe, 2016). Various care services which allow OPWD and their families to express choice are important. Notably, the question of who delivers these services arises. With the fiscal challenges reported in the literature, closer collaborations between the state, private sector and non-governmental agencies are recommended to meet the needs of older persons (Rawlins, 2015; Theodore et al., 2017). A balance between formal and informal care could also lessen the burden of informal carers (Barrett et al., 2016). Specifically in the context of Trinidad, a regulatory system is also recommended to ensure a high quality of care is provided due to a general distrust in care systems outside the family and variation in attitudes toward the care of OPWD. (Parey, 2021b).

Conclusion

The increasing proportion of OPWD means that understanding the care views towards them is essential. The use of a questionnaire based on Daly and Lewis’ (2000) conceptualisation provided considerable insight into care views for policy purposes. The study revealed six care dimensions among the sample, with care as a duty having similar prevalence as four other care views, and care as labour, where persons have to quit their jobs, as the least dominant. Care policies should reflect the various dimensions of care and not just the idea of care as an obligation to ensure OPWD and carers simultaneously receive the support they need. Education, employment status, ethnicity, and experiences with PWD influenced these care views. The findings indicate the important role of education among both men and women to disrupt traditional gender roles. There is also a need for a wide menu of public services to support ageing in place, such as flexi-time, work-from-home options, home care services, and community care services. Notwithstanding the provision of these services, a regulatory system is required to ensure the needs of both OPWD and carers are simultaneously met.