Ethnicity and Support for Parents in Malaysia

Ethnicity and Support for Parents in Malaysia

Adult daughters are important in the caregiving of aging parents. There is enough evidence that adult daughters are more likely than sons to provide support for their older parents or parents‐in‐law (Allen, Blieszner, & Roberto, 2000; Chesley & Poppie, 2009; Ingersoll‐Dayton, Starrels, & Dowler, 1996; Sarkisian & Gerstel, 2004a). Further comparison of parental care provided by daughters and daughters‐in‐law revealed that the former are more likely than the latter to provide more care and support (Allen et al., 2000; Lee, Spitze, & Logan, 2003; Merrill, 1993). Findings also revealed that couples give priority to the wife’s parents rather than the husband’s (Shuey & Hardy, 2003). These conditions may be the norm in more developed western societies, but may not be so elsewhere. In certain parts of the world, some people are inclined to differing sets of beliefs and practices in supporting parents and parents‐in‐law (Datta, Poortinga, & Marcoen, 2003).

This study examines ethnic diversity among adult women (married or previously married) in their propensity to provide tangible support to either set of parents. Malaysia, the country considered in this study, is multi‐ethnic and multicultural. The Malaysian Malays, Chinese, and Indians are the main ethnic groups, and these groups have differing cultural practices, even with regards to supporting parents and parents‐inlaw. Thus, married women are driven by culturally perceived obligations, and some are bound by traditional norms to oblige parents‐in‐law above parents (DaVanzo & Chan, 1994; Wu & Rudkin, 2000). We extend research regarding support to parents and parents‐in‐law by examining ethnic differences within a multi‐ethnic, developing Southeast Asian nation. Other factors influencing support are also examined. These are the increased needs in parents or parents‐in‐law, and the constraints on resources such as competing demands for support from other family members. As such, adult daughters have been found to be more affected by these demands than sons (Grundy & Henretta, 2006).

Conceptual Framework

Silverstein, Gans, and Yang (2006) built on intergenerational solidarity theory and conceptualized filial responsibility as a facet of social capital. Increased needs in either parent (declining health) evoked filial obligations that converted to support from adult children, and daughters had stronger filial obligations than sons towards their mothers. Other researchers (Fingerman, VanderDrift, Dotterer, Birditt, & Zarit, 2011) extended their model from contingency theory and found racial differences between American families in caring for younger and older generations. They included cultural beliefs that drove obligations towards support. Between White and Black Americans, the latter have stronger beliefs in caring for older parents. Fingerman et al. (2011) also included constraints on resources, such as income and competing demands from other family members. Family members compete for support from adult women, and having more offspring decreases the likelihood of providing for a parent (Concepcion & Perez, 2006; Grundy & Henretta, 2006).

Our model extends from these previous works. The three main ethnic groups in Malaysia have different culturally‐defined family structures. These family structures, in turn, influence the pattern of parental support. Cultural norms and beliefs in certain ethnic groups prioritize obligations towards the woman’s parents‐in‐law, above her own parents (DaVanzo & Chan, 1994).

This study relies on contingency theory, which predicts that family members respond to each other’s needs in times of crisis or recent events (Eggebeen & Davey, 1998; Silverstein et al., 2006). A contrary dimension to contingency theory is that of constraints on resources. The amount of support provided is contingent upon the availability of resources and one’s socioeconomic status (SES) (Fingerman et al., 2011; Sarkisian & Gerstel, 2004b). The likelihood of support also depends on combined demands for the resources, as other family members may be simultaneously competing for it (Concepcion & Perez, 2006; Grundy & Henretta, 2006).

Malaysia’s Ethnic Groups and Differences in Cultural Practices

Malaysia is an independent nation in Southeast Asia and consists of thirteen states in two major regions (i.e., Peninsular Malaysia, and East Malaysia on the island of Borneo). The Malaysian society is multi‐ethnic and multicultural, consisting of mainly Malays, Chinese, and Indians (United Nations Country Team & Economic Planning Unit, 2005, pp. 2‐4). According to the latest census, the composition of ethnic groups is 67.4% for Malays (also known as Bumiputeras, literally sons of the soil), 24.6% for Chinese, 7.3% for Indians and 0.7% for others (Malaysian Department of Statistics, 2011).

The Malays are the main ethnic group and are generally known to have a bilateral family structure. In some parts of the country, matriarchal structures are more dominant in practices of post‐marital residence and inheritance (Kling, 1995). By contrast, the Chinese and Indians generally have patriarchal family structures and their elders rely on sons for old‐age support. When daughters marry, they leave their family to become part of the husband’s family (DaVanzo & Chan, 1994; Hsu, Lew‐Ting, & Wu, 2001; Lillard & Willis, 1997; Wu & Rudkin, 2000). Thus, this study examines ethnic diversity among Malay, Chinese, and Indian women in providing support. We expected that Malay women would be more likely than Chinese and Indian women to support their own parents. Conversely, we expected that Chinese and Indian women would be more likely than the Malays to support parents‐in‐law.

Tangible Forms of Support

The primary form of support for aging parents is coresidence with adult children. Coresidence is a common practice among Asians and is viewed by Asian governments as an essential form of informal support (Chan, 2007; Jones, 2007). This form of living arrangement enables companionship, emotional, and financial support for aging parents. It also helps both parties share cost of living and domestic chores. Adult children who live together with their parents may purchase goods and services for their parents (Attias‐Donfut, 2000; DaVanzo & Chan, 1994). Past studies have shown that Malaysian Indian elders were more likely to coreside with adult children, whereas the Malays were the least likely. Many adult Malay children lived in separate houses from their parents, though often in proximity. A plausible reason for this is that the Malays are traditionally a rural population, and it is cheaper to own houses in rural areas. By contrast, Chinese and Indians in Malaysia are more urban (DaVanzo & Chan, 1994; Noor Laily, Tan, Tey, & Rohani, 1985; Wu & Rudkin, 2000).

The flow of intergenerational support in Malaysia is typically from the younger to older generations. Findings revealed high incidences of intergenerational transfers between non‐coresidential adult children and elderly parents. Chinese elders were more likely than the Malays to receive financial assistance and in larger amounts. Indian elders were less likely to receive financial help, but when it happened the amount was considerably larger (Lillard & Willis, 1997). This study examines parental support as either coresidence with adult children or assistance from noncoresidential children.

Predictors of Support

Elsewhere, proponents of contingency theory found that adult children provided support when aging parents were in times of need, rather than through normal forms of familial support or recurring intergenerational exchanges (Eggebeen & Davey, 1998; Silverstein et al., 2006). Therefore, there may be more support from adult women in situations of increased needs by parents or parents‐in‐law. Other researchers had used declining health or older age as proxies for increased needs, and found them to significantly influence support to aging parents (Fingerman et al., 2011; Silverstein et al., 2006).

On the other hand, constraints on resources may negatively affect support. Women’s SES, as in education level and income, will likely determine the amount of resources available for support (Concepcion & Perez, 2006; Fingerman et al., 2011). Thus, higher SES in women may lead to support. However, combined demands from other family members may decrease support. Adult children with more offspring have been found to decrease support to parents (Concepcion & Perez, 2006; Grundy & Henretta, 2006). Furthermore, there may be more competition for support if both parents and parents‐in‐law are alive. Therefore, women with more offspring or who have both sets of parents may reduce support.

There may be differences in attitudes towards parental support among the younger and older age groups (Hsu et al., 2001). Findings from Taiwan indicated that older children, aged above 40 years old, experienced many stresses in their careers, marriages, and bringing up offspring, and were less agreeable to multigenerational living than younger ones. Location of residence (rural or urban) may also influence support to elders (Concepcion & Perez, 2006). Besides that, the marital status of women is also of interest. It provides a comparison of the degree of support between still married and previously married (widowed, divorced or separated) women.

Data

This research uses data from the Women sample of the 4th Malaysian Population and Family Survey (MPFS‐4) in Peninsular Malaysia. The MPFS‐4 sample was selected through a stratified multistage sampling design and was fielded between July 2004 and September 2004. This national survey was conducted by the National Population and Family Development Board (NPFDB) of Malaysia and has been conducted every 10 years since 1974. It is guided by a study advisory committee comprising of officials from key departments and academics in Malaysia (Malaysian National Population and Family Development Board, 2004). We were given written consent to use this data for this research.

The MPFS‐4 Women sample consists of ever‐married women aged 15 to 49 years from all ethnic groups, i.e. Malaysian Malays, Chinese, Indians and other non‐Malay indigenous groups. Only the Malays, Chinese, and Indians were included because our primary focus is on the comparison among the three main ethnic groups. Besides, only respondents who were at least 21 years old were included in the study. We defined them as “adults,” and assumed that they were a source of support to both sets of parents. Others had used a rather similar age cutoff point in a study of living arrangements among older Malaysians with adult children (DaVanzo & Chan, 1994).

Two separate analyses were conducted: the first for support to parents and the second for support to parents‐in‐law. In examining support to parents, a subsample was created and included respondents who had at least one living parent (n = 2912; 2079 Malays, 546 Chinese and 287 Indians). Likewise in examining support to parents‐in‐law, respondents who had at least one living parent‐in‐law were selected (n = 2558; 1831 Malays, 484 Chinese and 243 Indians). Both subsamples were weighted according to the population distribution of the Malays, Chinese, and Indians in Peninsular Malaysia. Statistics were obtained from the 2000 Malaysian census (Malaysian Economic Planning Unit & United Nations Country Team, 2005, p. 11).

Measures

Dependent variable. In the MPFS‐4 survey questionnaire, respondents reported if they were coresiding with parents/parents‐in‐law. If they were not, they were asked if they had provided financial or material help in the past three months. Therefore, this study defined tangible support as either coresidence with parents/parents‐in‐law or financial/material provision by non‐coresidential children. In the analysis for parents, the dichotomous dependent variable indicated support to at least one parent. Likewise for parents‐in‐law, the dependent variable indicated support to at least one parent‐in‐law.

Independent variables in support to parents. Respondent’s ethnicity was included into the model (Malay, Chinese, Indian). Two measures of SES of the respondent were also added to indicate constraints on their resources: education (no schooling, primary, secondary, pre‐university/college/ tertiary) and monthly income (no income/not working, RM 1–1,000, more than RM 1,000). Other socio‐demographic variables were respondent’s age group (21‐29, 30s, 40s), residence (urban, rural) and marital status (currently married, widowed, divorced/ separated). A dichotomous measure indicating at least one elderly parent (aged 65 and above) was added, which served as a proxy for increased needs in parents. Variables representing competing demands for support from within the respondent’s family were number of offspring (“0 or 1,” “2 or 3,” “4 or more”) and a dichotomous indicator of having at least one living parent‐in‐law.

Independent variables in support to parents‐in‐law. All but three variables were similar to the variables in support to parents. The marital status variable had only two categories (currently married, widowed), as women divorced or separated did not report any support to parents‐in‐law. A dichotomous variable indicating at least one elderly parent‐in‐law was used to measure increased needs in parents‐in‐law. Among indicators representing competing demands, a dichotomous indicator of having at least one living parent was added.

Analytic Strategy

First, simple crosstab analyses were conducted on support provided to parents and parents‐in‐law according to ethnic groups. This was followed by separate analyses on support using hierarchical logistic regression methods (Hosmer & Lemeshow, 2000). The ethnic variable was added in the first step and the rest of the variables in the second step. The models with all nine variables were our main effects models. In the third step, two‐way multiplicative terms were added.

Through the estimation with interaction terms, true relationships between ethnic, other socio‐demographic factors, competing demands, and parental needs were analyzed. As there were nine independent variables in each model, there were therefore 36 (⁹C2) possible two‐way interactions. Each of these 36 interactions was added one at a time to the main effects models. The significance of each of these interactions was assessed through likelihood ratio test at an alpha level of 0.05. Various combinations of two‐way interactions were also added into both main effects (parents/parents‐in‐law) models. Finally, the fitted models with significant interaction terms were confirmed using Wald statistics. These models are the final models.

Support to Parents

We hypothesized that Malay women would be more likely to support their own parents compared to Chinese and Indian women. Our preliminary findings through various crosstab analyses (See Table 1) revealed that Malay women had the highest percentage (73%) in providing overall support (either coresidence or financial/material support) to parents. Chinese and Indian women also reported rather high percentages of support, with 63.4% and 63.7% respectively. With ethnicity as the only predictor in the logistic regression analysis (See Table 2), Malay women were more than 1.5 times more likely to support parents than were Chinese (OR = 0.64) and Indian (OR = 0.65) women. This was still true after including other predictors in the main effects model (See Table 3) and significant two‐way interactions in the final model (See Table 4). The odds that Malay women were more likely to support parents had also increased.

In the main effects model (See Table 3), income, residence, marital status, elderly parents (proxy for increased parental needs) and number of offspring were significant after controlling for other variables in the model. Women with higher income and women who stayed in urban areas were more likely to support their own parents. In terms of marital status, being widowed lowered the odds. Married women were about 1.7 times more likely to support than were those who were widowed (OR = 0.58). Surprisingly, divorced/separated (OR = 2.07) respondents were more likely to provide support than married ones. In the case of increased parental needs, respondents were more motivated to support (OR = 1.33) when parents were in the elderly category. In terms of competing demands, the odds of support decreased when there were more offspring (OR = 0.77, 0.59). On the other hand, whether or not respondents had living parents‐in‐law did not significantly affect support to their own parents. Women’s education level and age also did not significantly affect support to parents.

However, the effect of residence was conditional upon income and marital status. Through likelihood ratio tests, these two‐way interactions were significant: “Residence x Income” and “Residence x Marital status.” Thus, the inclusion of these interactions yielded a better fit in our final model (See Table 4). Women in rural areas who earned more than RM 1000 per month (OR = 2.47) were more likely to support parents compared to women in urban areas in the same income bracket, and compared to women in rural areas who were not working. Women in rural areas who were divorced/separated (OR = 4.52) were also more likely to support parents compared to women in urban areas with similar marital status, and compared to women in rural areas who were still married.

The effects of other predictors in the final model were rather similar to the main effects model. Ethnic, elderly parents and number of offspring were still significant. Malay women were more likely to support parents than were Chinese and Indian women, and women whose parents had more needs (OR = 1.33) were more likely to support. Those with more offspring (OR = 0.76, 0.59) were less likely to support parents, but the effect of having both sets of parents (living parents‐in‐law) was not significant. Women’s education level and age were also not significant in the final model.

Support to Parents‐In‐Law

We hypothesized that Chinese and Indian women would be more likely to support parents‐in‐law than Malay women. Our results from various crosstab analyses (See Table 1) showed that provision of support for parents‐in‐law was highest among Indian (71.3%) women, followed by the Chinese (63.6%). Although the percentage among Malay women (63.0%) was the lowest, it was not substantially different from the Chinese. With ethnicity as the sole predictor in logistic regression analysis (See Table 2), Indian women were 1.45 times more likely to support parents‐in‐law than were Malay women. After controlling for other variables in the main effects model (See Table 3), Indian women were still more likely to support parents‐in‐law than were Malay women by as much as 1.65 times. On the other hand, there were no significant differences in support between Malay and Chinese women in the ethnic only model and the main effects model. Then again, likelihood ratio tests indicated that there were significant interactions between ethnicity and other factors. Therefore, logistic regression estimates for ethnicity had to be interpreted as conditional upon these factors.

When we solely examined the main effects (See Table 3), ethnicity, education level, monthly income, age and marital status were significant. Variables which were not significant in support to parents‐in‐law were women’s place of residence, increased needs of parents‐in‐law, number of offspring and living parents. However, likelihood ratio tests revealed two significant interactions: “Ethnicity x Income” and “Ethnicity x Elderly in‐laws.” Thus, our interpretations differed considerably after we added these interactions in our final model.

In our final model (See Table 4), ethnicity was conditional upon monthly income and increased needs of parents‐in‐law. Chinese women who earned more than RM 1000 per month (OR = 0.50) were less likely to support parents‐in‐law compared to Malay women with similar income, and compared to Chinese women who were not working. On the other hand, there were no significant differences between Malay and Indian women with regard to income. When faced with increased needs of parents‐in‐law, Chinese women (OR = 0.49) were less likely to provide support compared to Malay women. Chinese women who had elderly parents‐in‐law were also less likely to provide support compared to Chinese women with younger parents‐in‐law (younger than 65 years). There were no significant differences between the Malays and Indians with regards to increased needs of parents‐in‐law.

Table 1.

Crosstab Analyses: Coresidence, Financial / Material help, and Support to Parents and Parents-In-Law by Malaysian Women of Different Ethnicity (%)

Table 2.

Logistic Regression Analyses: Step 1 - Ethnic only

Education level, age, and marital status were still significant in the final model. Odds ratios increased with higher levels of education. Respondents educated at the pre‐university/ college/tertiary level (OR = 3.13) were about 3 times more likely to support parents‐in‐law than respondents with no formal schooling. There were also differences in age groups, where respondents in their 30s and 40s were less likely to support parents‐in‐law, compared to their younger counterparts (OR = 0.77, 0.74). With regards to marital status, comparison between married and widowed women revealed a significant decrease in odds (OR = 0.24) to support parents‐in‐law. Women’s place of residence, number of offspring, and living parents were not significant in predicting support to parents‐in‐law.

Table 3.

Logistic Regression Analyses: Step2 - Main Effects

Table 4.

Logistic Regression Analyses: Step 3 - Main and Interaction Effects

Further Analysis

Further analyses was conducted in order to explain several puzzling results: (a) divorced/ separated women were more likely to support parents than were married ones, (b) in support to parents‐in‐law, Chinese women who were not working were more likely to support parents than were Chinese women who earned more than RM 1000 per month, and (c) in support to parents‐in‐law, Chinese women with younger parents‐in ‐law (younger than 65 years) were more likely to support compared to Chinese women who had elderly parents‐in‐law.

Our study defined support as either coresidence with parents or when the adult women provided financial or material help. Therefore, higher odds of support among divorced/separated women may be due to higher incidences of coresidence among them. The chi‐square test of independence between marital status and coresidence with parents confirmed this, and revealed an association between the two (χ² (2, N = 2912) = 273.38, p < 0.01). Close to 70% of divorced/ separated women were staying together with parents (not shown here). Further examination of this subgroup according to place of residence, revealed that divorced/ separated women in rural areas were more likely to stay together with parents, compared to those in urban areas (not shown here).

For the second poser, coresidence with parents‐in‐law was not a probable reason because there was no association between income and coresidence (χ² (2, N = 484) = 0.66, p > 0.50). A possible explanation could be that Chinese women who were not working had higher household income and benefited from support from their spouses. Our analyses did not include household income, as we were unable to obtain this information. This is a limitation in our study. For the third, Chinese women who had younger parents‐in‐law were probably younger as well. This result is similar to our findings that women in younger cohorts provided more support to parents‐in‐law. Younger cohorts also made up the highest percentage in coresiding with parents‐in‐law (not shown here).