Introduction

Notwithstanding the infancy of research on age and social exclusion as a topic of research worldwide, the studies consistently evidenced the elderly persons as a cross section of the society where they are at risk for social exclusion alongside the advancing age (Levitas et al. 2007; Agulnik et al. 2002; Hoff 2008; Kneale 2012; Jose and Meena 2015; Mariyam and Jose 2015). Age is permeable than others social categories such as gender, caste and ethnicity since all people invariably age over time (Jose and Cherayi 2016; Jose et al. 2016; Howard 2000). Thus, advancing age is associated with deterioration in physical and cognitive functioning, shrinking of social roles, retirement from employment, and more importantly the threat to wellbeing (Hurd 1999; Lynch 2000; West et al. 2002; Jose et al. 2016). This is potential enough to dispose elderly persons to social exclusion (Jose and Cherayi 2016; Jose et al. 2016; Jose and Meena 2015) whereas the majority dimensions of social exclusion of older people are on poverty, health, life expectancy, fear of crime, poor housing and lack of independence that mirrors recent analyses of social exclusion in older people (Aldridge et al. 2011). The indicators used to operationalize and measure social exclusion in older persons often results from the loss of independence viz., lack of pension, public transport, housing and promoting the needs for state interventions.

Although, age in itself is not a dimension of social exclusion, the risk of reduced income increases with advancement in age (Agulnik et al. 2002). Age related characteristics such as disability, cognitive decline, low income and widowhood, labour market characteristics, economic decline and crime in local areas and age based discrimination dispose older persons at risk of social exclusion (Phillipson and Scharf 2004). Therefore, the advancing age was further associated with decreasing social relationships, restricted access to service provisions and material consumption, thus age is associated with greater risk for social exclusion. Besides, the elderly persons were less likely to live with their partners, more likely to be widowed, live alone and face exclusion from civic and cultural activities. They were likely to enjoy poor health that further restricted their independence whereas they spend more time at home and relay more on immediate environment (Popay et al. 2008; Burns et al. 2012).

The studies on social exclusion in India were emerging but such studies on older people are scant (Jose and Meena 2015; Mariyam and Jose 2014). As a result, we know little about the nature and severity of social exclusion of older people in India face. Little is known about the how socio-demographic variables influence social exclusion and its sub-dimensions. The present study is designed to address some of these identified concerns such as role of socio-demographic variables on social exclusion of older people.

Concise Review of Literature

Academic and policy literature frequently refer the term ‘social exclusion’ to designate the multi-dimensional disadvantages experienced by people at the social margins (Levitas et al. 2007). The dominant discourses conceptualize social exclusion as a multi-dimensional and dynamic process leading to negative outcomes in socio-economic, political and cultural aspects of certain aggregates of human lives (Popay et al. 2008). It is a process induced and mediated by poverty, lack of basic competencies and restricted lifelong learning opportunities due to discrimination. It sidelines individuals, households, groups, communities and countries at global levels (European Commission 2000). Its process covers the restriction and denial of individuals and social groups from fully participating in socio-economic and political life and asserting their due rights (Beall and Piron 2005).

In empirical literature, the indicators used to operationalize and measure social exclusion in older persons often results from the loss of independence viz., lack of pension, public transport, housing, and promoting the need for state intervention. It was difficult to introduce the centrality of labour market participation of elderly persons, which excluded them from dominant social exclusion discourse (Kneale 2012). The retirement from work was a normal process, which may not be treated as a social exclusion, though retirement has potential social exclusion effect. The retirement was predominantly shaped through experience of labour market participation, which makes social exclusion of elderly persons more complex (Kneale 2012).

Age in itself was not a dimension of social exclusion. The risk for reduced income increases with advancement in age (Agulnik et al. 2002). Age related characteristics such as disability, cognitive decline, low income and widowhood, labour market characteristics, economic decline and crime in local areas and age based discrimination dispose elderly persons at risk of social exclusion (Phillipson and Scharf 2004). Advancing age was associated with decreasing social relationships, service provisions and material consumption; therefore greater risk for social exclusion. For example, as the result of ageing, elderly persons were less likely to live with their partners, more likely to be widowed and live alone, and face exclusion from civic and cultural activities. They were more likely to expose to poor health, which further restricted their independence. Elderly persons were likely to spend more time at home and relay more on immediate environment (SEU 2006; Burns et al. 2012). In a logistic regression analysis Barnes et al. (2006) found that advanced old age, single person household, poor mental and physical health, lack of access to private transport, living in rented apartment and reliance on pensions as the main source of income as the significant predictors of social exclusion in older persons.

Social exclusion has a gender dimension too since older women were more likely to be socially excluded than older men. Kneale (2012) argued that gender bias that dispose older women more vulnerable to social exclusion since women outlive men. It was not because of gender rather women were more likely to live in 80s than men, which dispose them more vulnerable to social exclusion. Similarly, it was not the ethnicity in old age that leads to social exclusion rather their likelihood of living in rented apartment disposed them to social exclusion.

In United Kingdom, the lack of access to transport and poor accommodation dispose older persons vulnerable to social exclusion (SEU 2006; English Housing Conditions Study 2001). The older persons faced restricted access to information (Kneale 2012). In Australia, social exclusion of older persons focused on deprivation, poverty, crime and fear of crime, perceptions of neighbourhood and social integration (Scharf et al. 2002). The studies in Europe, especially in Briton, elderly persons’ need for health and physical mobility models such as transport, wellbeing needs and feeling of being valued (eg, Victor et al. 2003; Abbott and Sapsford 2005; Scharf et al. 2005). The low income, fear of crime, social integration, perception of neighbourhood, independence, access to healthcare and transport have repeatedly surfaced in qualitative analyses (Waterhouse and Angley 2005). The poor life satisfaction was usually an outcome of social exclusion but chronic poor life satisfaction further worsens social exclusion (Kneale 2012). Further, being women and widowhood during old age significantly associated with poverty (Ogg 2005; Gordon and Townsend 1990). The widowhood reduced subjective wellbeing and health during old age (Diener et al. 1999; Shields and Wooden 2003) whereas weak social and familial ties increased social isolation (Hoff 2008).

In Netherland, association between income and poverty have observed in very specific groups of older women aged 55–65 years and 66–75 years. They were women, who were unemployed during their middle age, widowed during their 40s, who relied on welfare benefits, therefore could not accumulate income for old age. Instead, they maintained their life during adulthood on minimum state welfare benefits. The women who belong to ethnic minority were another category who does not have completed their minimum residential period at the time of 65 years when they retire. As a result, these women had to dependent on state’s minimum pension. The needs for care as results of poor health, dementia and disability have reported as a major risk factor for income poverty in Ten European Countries (Hoff 2008).

Social exclusion has increasingly been used in academic and policy literature to designate the multi-dimensional disadvantage (Levitas et al. 2007) faced by people at the social margins (Jose 2014). Elderly persons were one of the cross section of especially at risk group for social exclusion. Nevertheless, there are little amount of studies on social exclusion of older persons are available in Indian literature. Available studies inadequately address how gender, social group affiliation, marital status and living status shape social exclusion in older persons (Jose and Meena 2015; Mariyam and Jose 2014).

Method

Using a cross sectional design, this study examined the influence of socio-demographic characteristics upon the social exclusion of older persons. The study universe constituted the older persons aged 65 years and more who resided in the southern state of Kerala in India. The sample was defined as ‘a person aged 65 years and more, presently living either with family of procreation or in an institutional care facility in the selected districts, namely Kannur, Ernakulam and Pathanamthitta’. The definition of elderly person was slightly deviated from official definition in India (National Policy on Older Persons 1999) which is marked by persons 60 years and above, which entitle them to avail social welfare benefits including old age pensions. Nonetheless, WHO (2010) defines the older persons aged 65 years and above, though this cut-off criterion has been questioned from many quarters, especially from the developing countries (Srivastava et al. 2013), where the threshold is argued to be lowered into 60 years so these elderly persons would be able to avail old age pensions and other welfare benefits (Srivastava et al. 2013; WHO 2010).

However, the present study has chosen 65 years as the minimum age to participate in the study primarily on two grounds. First, the older persons at their 60s are likely to be still economically active and engage in familial and social roles (Cornwell and Waite 2009); therefore such elderly are less likely to experience social exclusion as compared to more aged persons. Secondly, the state of Kerala rank top among other Indian states on human development index, which is comparable with many developed countries in the West (Human Development Report 2005). Hence, we treated the elderly persons of Kerala based on its performance on human development index and justifying the use of 65 years as the cut-off point to include older persons in to the study universe, consistent with WHO definition (WHO 2010). The eligibility criteria to recruit older persons in the study are: aged 65 years and more, who self-reported that they were not suffering from any severe forms of mental illness or cognitive deficits and provide informed oral consent.

Sample Size Estimation

To estimate the sample size for the survey section, we used the following formula (Jose and Cherayi 2016)

The formula is \( \mathrm{n}=\frac{\widehat{\mathrm{p}}\left(1\hbox{-} \widehat{\mathrm{p}}\right){\mathrm{z}}^2}{{\mathrm{ME}}^2} \)

Where, n is the sample size, \( \widehat{\mathrm{p}} \) is the prior judgment of the correct value of p which means the percentage of the population. ME is the desired margin of error, we need a margin of error less than 2.5 %. Therefore, in the proposed study we assumed a margin of error at 2.4 %. The z is the confidence interval; we have selected 90 % confidence interval, which means the z score is 1.645.

Where \( \begin{array}{cccc}\hfill \widehat{\mathrm{p}}=0.105\left(10.5\%\right);\hfill & \hfill 1\hbox{-} \widehat{\mathrm{p}}=0.895;\hfill & \hfill \mathrm{z}=1.645;\hfill & \hfill \mathrm{ME}=0.024\hfill \end{array} \)

Therefore \( \mathrm{n}=\frac{0.105\times 0.895\times {1.645}^2}{0.024^2}=507 \)

Thus, the calculated sample size is 507, which is round up into 600 for ensuring the equal proportion of population in the selected district. However, the final sample size for analysis was 573 as showed in the Tables 1 and 2. This was because; we have excluded the incomplete interviews, interviews with significant amount of missing responses (more than 5 %) and outlier cases.

Table 1 Shows the sampling procedure in detail
Full size table
Table 2 Shows the socio-demographic profile of the older persons
Full size table

Sample Selection Procedure

We used multi-stage cluster sampling procedure for this study (Bryman 2008; Bordens and Abbott 2011). This sampling procedure involves multiple stages of sampling since the primary unit (first stage of the sampling procedure) is not the unit of the population to be studied but grouping of those units that are known as clusters (Kish 1965). As Bryman (2008) recommends, we used probability sampling procedure (either simple random) at each stage of sample selection in cluster sampling.

In the first stage, we selected three districts each represents the south, the central and the northern divisions of the state of Kerala, namely Pathanamthitta, Ernakulam and Kannur. In the second stage, two to three administrative blocks from each selected district were randomly selected which were treated as clusters. This clusters/blocks1 consisted of different village panchayats2 (About 50 % of the village panchayats were randomly selected by using lottery technique.

We selected 50 % of the wards3 from each selected village panchayat (herein after VPs). The selected wards were treated as the primary sampling unit (PSUs) where our field researchers conducted a mapping exercise to identify the older persons with the help of local voluntary organization and local panchayat representatives. Through this exercise, we developed a sampling frame from where we randomly selected older persons who met the eligibility criteria.

Pathanamthitta district was selected from the south Kerala with 8 block panchayats in two blocks namely the Ranni (with 9 VPs) and Pandalam (with 6 VPs) constituted 14 VPs. Among 14 VPs, 8 were selected. Using the sampling frame developed at ward3-wise, we selected 22 participants each for structured interviews from five VPs, and 21 participants each from three VPs, thus 173 elderly persons.Footnote 1 , Footnote 2 , Footnote 3

Ernakulam district was selected from the central Kerala with 14 block panchayats and 84 VPs. We selected 3 block panchayats namely Vypin (with 5 VPs), Kothamangalam (with 10 VPs) and Parakadave (with 6VPs) constituting 21 VPs. Out of these, 10 VPs were randomly selected as final study units. Kannur district has 11 block panchayats and 81 VPs. Two block panchayats namely Edakkadu (with 8 VPs) and Peravoor (with 7 VPs) constituting 15 VPs. Out of these, 8 were selected as final study unit. We selected 25 samples from each selected village panchayat for the study, therefore sizing a sample of 200 elderly persons. Altogether, we completed 573 structured interviews in three districts namely Kannur (n = 200), Ernakulam (n = 200) and Pathanamthitta (n = 173) in Kerala.

Measurements

First, we developed a structured interview schedule based on the critical variables considered for the study. The instrument included socio-demographic profile and social exclusion scale. For measurement, we developed a socio-demographic schedule. It contained the variables such as age, gender, years of education, family income, occupation, social group and religion.

Secondly, we used the social exclusion scale developed by Jehoel-Gijsbers and Vrooman (2007). It is a 15-item instrument to measure social exclusion with four subscales viz., material deprivation, inadequate access to social rights, normative integration and inadequate social participation. The sample scale-items are as follows: ‘There are people with whom I can have a good conversation’, ‘I feel cut off from other people’, I give to good causes, I did not receive any medical care’, and I have enough money to afford basic needs of my home’ etc. The responses are rated on a five-point Likert type rating scale with never (1)…to always (5). The high scores on the scale indicate high level of social exclusion. Material deprivation subscale showed a reliability coefficient of 0.79 and access to social rights reported a reliability coefficient of 0.82. The social participation had a reliability coefficient of 0.77 and reliability coefficient of normative integration was 0.67. The reliability coefficient of the overall scale is 0.85.

The scale has reasonable reliability in Indian samples (Cherayi 2015; Cherayi and Jose 2016). In a translated version of the social exclusion scale in Indian samples. In the translated version of the subscale material deprivation yielded an Cronbach’s alpha of 0.77, social rights subscale yielded 0.69, social participation subscale yielded a reliability alpha coefficient of 0.71 and the normative integration has yielded an alpha coefficient of 0.69 with an overall reliability coefficient of 0.76 (Cherayi 2015). In the current sample, social exclusion yielded a global Cronbach’s alpha reliability of 0.79.

Two independent translators with excellent command over both English and Malayalam (to read, write and speak) were engaged in translation of the structured interview schedule. Both translators were postgraduate in social work with research experience and were currently working in research institutions, though they were not directly involved in the present study. The Principal Investigator and the Research Associate have independently reviewed the accuracy of the translation through back translation from Malayalam to English. We presented observations, feedback and findings of the reviewers in a meeting with the professional translators and research staff including principal investigator. This helped to revise and further refine the translated version (in Malayalam) of the structured interview schedule.

Data Analyses

First the data were edited, processed and transformed based on the factors of the standardized rating scales. Further, the data were submitted for examining normality of data on important variables studied using QQ Plots and found normal distribution. Finally, the descriptive statistics, Independent Sample t’ test, ANOVA and multiple regression analysis were performed in order to meet the set objectives.

Ethical Considerations

The study protocol received ethical clearance from the institutional review board specially constituted for this study, by the project implementing agency (ie, Jeevas Social Centre). As per the ethical guidelines, oral consent was obtained from each study participant with a view to protect their rights either to participate or to decline their participation, after informing about the study purpose and nature of their participation.

Results

Socio-Demographic Characteristics

Tables 1 and 2 shows the socio-demographic profile of the older persons. The elderly persons’ age ranged from 65 to 102 years with a mean age of 75.5 years and mode was 75 years (±7.6 years). Most elderly persons were women (61.2 %) and men were 38.3 %. The elderly persons years of education ranged from no formal education (Mode = 0 year) to 18 years of formal education with a mean of 4.2 years (±3.2 years). But 73.9 % of the older persons reported that they completed no formal education to 5 years of formal education at schools and 24.2 % completed 6–10 years of education. Evidently, the elderly persons were relatively less educated especially up to primary and secondary level of school education.

Most of the older persons are married (43.4 %) and unmarried older persons constituted merely 11 persons (1.9 %). About 8.3 % of the elderly men were widower and 46.3 % elderly women were widows. The religious composition reveals that 57.4 % of older persons were Hindus, the Muslims were 7.9 % and Christians were 34.8 %. Further, there is a significant under-reporting of family income under the pretext of perceived probable exclusion from welfare benefits. As a result, income showed skewed trends while most of the older persons reported that they do not have a sizable family income (Mode = 0). Income ranged from no income to Rs. 50000 per year with a mean of 3607.1 (±6294.6). The categorical analysis reveals that most of the older persons had no or less income (52 %; below Rs. 1000). About 32.2 % of them belong to the income category of Rs. 1001 to 5000. Only 8.9 % of the older persons reported that they have Rs. 10001 and above as monthly income.

Table 3 shows the descriptive scores on the social exclusion measured on four sub-dimensions. First sub-scale was on inadequate social participation. The scores obtained on this subscale ranged from 4 to 20 with a mean of 10.8 (±4.3). The result reveals that older persons experienced inadequate social participation. Second subscale was inadequate normative integration. The scores obtained on this subscale ranged from 4 to 20 with a mean of 12.3 (±3.3). The result reveals that older persons experienced moderate level of inadequate normative integration. Third subscale was on inadequate access to basic social rights. The scores obtained on this subscale ranged from 3 to 15 with a mean of 6.3 (±2.6). The result reveals that older persons suffered inadequate access to basic social rights. Fourth subscale was on material deprivation. The scores on this subscale ranged from 4 to 20 with a mean of 16.4 (±4.9). The result reveals that older persons experienced high level of material deprivation. The scores of overall social exclusion scale ranged from 15 to 73 with a mean of 45.9 (±11.1). The result reveals that older persons experienced high level of social exclusion.

Table 3 Shows the descriptive scores of the sub-dimension of social exclusion
Full size table

Socio-Demographic Characteristics and Social Exclusion

Table 4 shows the influence of gender on sub-dimensions of social exclusion. Inadequate social participation is significantly differed between gender (t(571) = −2.674; p < .01). The mean difference shows that older women experienced inadequate social participation (mean = 11.1) than older men (mean = 10.2). The material deprivation is significantly differed between gender (t(571) = −3.5; p < .001). The mean difference showed that older women experience high level of material deprivation (mean = 17) than older men (mean = 15.5). Finally, the social exclusion was significantly differed on gender (t(571) = −2.36; p < .05). The mean difference showed that the older women faced increased social exclusion (mean = 46.8) than the older men (mean = 44.5).

Table 4 Shows the influence of gender on social exclusion
Full size table

Table 5 shows the caste wise variance of the dimensions of social exclusion. The analysis of variance showed a significant variance of inadequate social participation on social groups (F(572) = 7.988; p < .001). The post hoc Tamhane test (equal variance not assumed) showed a significant subset difference between older persons in general and OBC categories (p < .001) SC and General (p < .001), SC and OBC (p < .001). There was no significant subsets difference between SC and ST older persons on inadequate social participation. Inadequate normative integration was significantly varied on social groups (F(572) = 4.191; p < .01). The post hoc Tukey test shows a significant subset difference on general and SC (p < .001) but there was no significant subset difference between SC and ST. The results indicated that older persons who belong to socially marginalized groups such as scheduled caste and tribes were especially vulnerable to inadequate normative integration. The older persons’ access to basic rights is significantly varied on social groups (F(572) = 2.75; p < .05). The post hoc Tamhane test (equal variance not assumed) showed a non-significant subset difference on social groups.

Table 5 Shows the caste wise variance of the dimensions of social exclusion
Full size table

The ANOVA revealed a significant variance of older persons’ material deprivation on social groups (F(572) = 5.667; p < .001). The post hoc Tamhane test (equal variance not assumed) showed a significant subset difference between older persons in general and not reported categories (p < .05) and General and SC (p < .001). The results indicated social group affiliation was a significant factor disposing older persons vulnerable to material deprivation. Social exclusion is significantly varied on social groups (F(572) = 8.708; p < .001). The post hoc Tukey HSD test showed a significant subset difference on general and SC (p < .001), OBC and SC (p < .001) but no such subset difference was observed between older persons who belong to SC and ST population.

Table 6 shows the marital status wise variance of the dimensions of social exclusion. ANOVA revealed a significant variance of inadequate social participation on marital status (F(572) = 4.587; p < .001). The post hoc Tukey HSD test shows a significant subset difference between married and widowed (p < .05) where widow experienced inadequate social participation than married older persons. The material deprivation was significantly varied on marital status of older persons (F(572) = 4.798; p < .001). The post hoc Tamhane test (equal variance not assumed) showed a significant subset difference between married and widowed older persons on material deprivation (p < .01) where widowed reported increased material deprivation. Further, the social exclusion is significantly varied on marital status of older persons (F(572) = 3.797; p < .05). The post hoc Tukey HSD test shows a significant subset difference between married and widowed older persons on social exclusion (p < .05) which disposed the widowed more vulnerable to social exclusion. Evidently, widowhood is likely to pose an increased risk for social exclusion in older persons.

Table 6 Shows the marital status wise variance of the dimensions of social exclusion
Full size table

Table 7 shows the current living status wise variance of the dimensions of social exclusion. Analysis of variance reveals that older persons’ inadequate social participation significantly varied on their current living statuses (F(572) = 6.585; p < .001). The post hoc Tamhane test (equal variance not assumed) shows a significant subset difference between older persons living with family and spouse and living alone (p < .05) and living in elder care homes (p < .01), where those who are living with family and spouse enjoyed better social participation. Access to basic social rights is significantly varied on older persons’ current living status (F(572) = 2.874; p < .05). The post hoc Tukey HSD test shows a significant subset difference between older persons living with family and spouse and living alone (p < .05). ANOVA reveals that social exclusion is significantly varied on older persons’ current living statuses (F(572) = 2.979; p < .05). The post hoc Tamhane test (equal variance not assumed) shows a significant subset difference between older persons living with family and spouse and living alone (p < .05) indicating severe social exclusion those who live alone.

Table 7 Shows the current living status wise variance of the dimensions of social exclusion
Full size table

Table 8 shows the results of the multiple linear regression, which examined socio-demographic predictors of social exclusion in older persons. We considered age, religion, castes, education, marital status, current earning status, and present living status between older men and women. All categorical variables were converted into binary responses with 0 and 1 (eg, male = 1; female = 0 for gender) before entering into regression model. First, we ruled out the possible role of multi-collinearity among the predictor variables through examining the tolerance and variance inflation factor (VIF). Multi-collinearity exists when tolerance is below 1 and VIF is greater than 10. In this analysis, tolerance level of each predictor variable was within the acceptance limit (ie, less than 1) and less than the VIF of 10 for exact values. Second, we examined the overall significance of the model using F statistic separately for gender groups (viz., older men and older women) and found the predictor models of social exclusion were significant at 0.001 level (F = 4.720; p < .001; and F = 8.281; p < .001) respectively for older men and older women.

Table 8 Shows the results of multiple linear regression analysis that examined the predictors of social exclusion
Full size table

Interestingly, the caste (being a member of more privileged caste viz., forward castes) was found to significantly reduce social exclusion in older women (β = −.216; p < .001) nonetheless this caste affiliation did not help older men’s to reduce the severity of social exclusion (β = −.084; p > .05). Evidently, caste alongside with gender significantly accumulates the social deficits in older women whereas to the older men, gender favours tend to buffer the accumulated effects caste affiliation. The years spent for formal education significantly reduce the severity of social exclusion for both older men and women (β = −.236; p < .001; β = −.189; p < .001) respectively. Nonetheless, more years spent for education helped men to reduce the severity of social exclusion to a great amount as compared to older women. Plausibly, this indicates the continuing influence of gender on education where women are likely to have fewer opportunities than men who are equally educated or relatively less educated. Similarly, older persons’ who are currently not earning significantly increased the severity of social exclusion (β = .155; p < .05; β = .150; p < .05) for older men and women respectively.

Nonetheless, the non-significant predictors were age (β = .002;p > .05 and β = −.008; p > .05), religion (β = .033; p > .05; β = .004; p > .05), marital status (β = −.095; p > .05; β = −.044; p > .05) and place of living (β = .028; p > .05; β = .077; p > .05) for older men and women respectively. For older men, overall model accounted 12.4 % of variance and for women; the model variance was 13.4 %. Evidently, caste, education, current earning status are the critical predictors of social exclusion of older persons.

Discussion

In Kerala, the state policy on ageing position the issues of the older persons as the social issues, thus reiterate the responsibility of the state government to provide care and protection while ensuring their needs for freedom, satisfaction with life, personhood and status. The document envisages empowering older persons economically, socially, psychologically, emotionally and in health (State Policy on Ageing 2006). Consistently, the Maintenance and Welfare of Parents and Senior Citizens Act, 2007 and National Policy for Senior Citizens 1999 and 2011 address need for mainstreaming senior citizens and promote ageing in place, whereas consider institutional care as last resort. It promotes inclusive, barrier free and age-friendly society, ensuing equal opportunities, protection of rights, enabling for full participation and social security to older persons living below poverty line. It also provides healthcare, shelter and welfare and income security; thereby improving their quality of life and achieve an identity for older persons.

In order to achieve the state and national policy prescriptions, the studies examining the how socio-demographic characteristics of older persons socially exclude them are of critical importance. In the present study, we found that the older persons reported less social participation, inadequate normative integration, limited access to basic social rights and increased material deprivation, implying high level of social exclusion. Further analysis revealed that being older women, widowed, living alone and of tribal and dalit origins reported severe social exclusion. The caste significantly reduced social exclusion in older women but it did not help older men. The education significantly reduced the severity of social exclusion for both older men and women and current earning status.

Gender and Ageing

Although more boys are born than girls are, men tend to have more mortality rate than women at all age groups. As a result, there is a sex ratio imbalance during old age. Hence, ageing in Kerala is disproportionately of women in ageing population. The sex ratio among 60+ 1304:1000 males in 1901 has decreased to 1289 in 1951 and 1229 in 2001. Highest sex ratio was among people aged 75 years and more. As per the projections, there are 1445 females for 1000 males among 75 years old and more groups. Hence, there is a considerably large number of women ageing in Kerala than men (Moli 2004; State Policy on Ageing 2006). This strengthens the discourses on the feminization of old age, invisibility of older women and a combination of ageism and sexism that dominate the research literature on elder abuse and neglect with significant focus on gender shaping the ageing experience in older adults (Jose and Meena 2015; Mariyam and Jose 2014; Sredhanya 2014; Jose and Cherayi 2015, 2016; Jose et al. 2016).

Older persons in general experience high level of social exclusion in terms of poor social participation, compromised access to social rights, poor social (normative) integration and increased level of material deprivation (Mariyam and Jose 2015; Jose and Meena 2015). However, social exclusion is found to significantly be influenced by gender where older women reported poor social participation, increased material deprivation and overall social exclusion. Evidently, older women are likely to experience severity of social exclusion on different life domains than elderly men do (Jose and Meena 2015; Hoff 2008; Ogg 2005; Kneale 2012; Shields and Wooden 2003). This study found that gender acts as a significant variable that qualitatively shape the experience of ageing differently to older women and men. Socio-cultural factors such as gender and patriarchal value systems significantly influence ageing experience even during old age (Jose and Meena 2015). For instance, this study revealed that social exclusion including its sub-dimensions significantly differed between older men and women whereas the older women are likely to experience high level of social exclusion, inadequate social participation and increased material deprivation. Mariyam and Jose (2014) found that gender significantly determine the nature and severity of social exclusion in older persons.

Social Groups and Social Exclusion

Older persons’ social group membership in marginalized groups itself is not social exclusion. Nevertheless, being a member of a marginalized group, the older persons are likely to experience increasing amount of deprivations from basic resources for sustenance such as material resources, access to occupational opportunities, social rights and social participation alongside with social integration with the mainstream society (Jose and Cherayi 2015; Jose and Meena 2015). Consistently, Kneale (2012) found that it is not the ethnicity in old age that leads to social exclusion rather older persons’ likelihood of living in rented apartment disposed them to social exclusion. Evidently, the effects of being a member of social group at the social margin dispose older persons increasingly vulnerable to social exclusion. This study found that the older persons from scheduled caste and scheduled tribes are especially vulnerable to inadequate social participation, normative integration and poor access to basic social rights. Further, social group affiliation is a significant factor disposing older persons vulnerable to material deprivation. Older persons from marginalized social groups reported severe level of social exclusion than those older adults from general and other backward social groups.

Marital Status and Social Exclusion

Older widows were at greater risk of poverty (Gordon and Townsend 1990), poor subjective wellbeing and poor health (Shields and Wooden 2003). Poor social ties including family ties have frequently reported as a factor for social isolation, whereas older women, especially widows are likely to experience severe amount of social isolation (Hoff 2008). Sredhanya (2014) found that older widows felt lonely, worthless, isolation and distancing within and outside families, poor support and excessive familial control on freedom, physical mobility outside homes, familial and social interactions that finally compromised social roles. The intense neglect and abuse were fund to relate with mental health problems in older widows such as anxiety and depression. The widows faced with problems in availing adequate diet, decent housing and access to timely and quality healthcare services.

In Kerala, statewide representative surveys (UNFPA 2013) reported that 89 % of the older men are currently living with spouses while only 33 % of the older women live with their spouses. The percentage of widowhood in older women was far higher (61 %) than in older men (ie, 10 %). This significantly higher proportion of widowed women during old age leaves many critical psychosocial issues behind. For instance, Sredhanya (2014) examined the factors significantly predicting abuse and neglect of older widowed women. The significant predictors were objective social integration, support from family and difficulties in current housing facilities. These three variables together accounted for about 50 % of variance on abuse and neglect. The social integration, family support and housing difficulties significantly disposed older widows vulnerable to abuse and neglect experience in familial and community life. We found that widowed and unmarried older persons experienced inadequate social participation than married older persons. Material deprivation significantly varied on marital status of older persons where widowed older persons reported increased material deprivation. The social exclusion significantly varied on marital status of older persons, which disposed widowed more vulnerable to social exclusion. Evidently, widowhood is likely to pose an increased risk for social exclusion in older persons.

Living Status and Social Exclusion

The decline in fertility, nuclearization of families and migrations significantly altered traditional co-residential patterns. However, 82 % of the older men and 88 % of the older women are co-residing with their families. About 15 % of the older persons live independently (live alone or with spouse) while 4 % live alone. Although, Kerala has an earlier demographic transition alongside with increased migration pattern, percentage of older persons living alone appears to be lower as compared to other states in India. For instance, in Tamilnadu, about 26.4 % of older women live alone while Punjab reported lowest rate of living alone older persons (.e., 3.2 %) with a national average of 10 % (UNFPA 2013).

Nevertheless, living alone is associated with social deficits accumulations, including poverty, which is especially the case of Kerala where about one fifth of the older persons living alone are from poorest wealth quintiles. Most of the older persons who live alone reported that either no children or children living away (due to migration etc.) (UNFPA 2013). Living alone increase risk for psychosocial problems during old age such as poor life satisfaction (Thomas and Nagaraju 2012), conflicts in social relations, social isolation, poor social integration and social exclusion (Jose and Cherayi 2015). Consistently, we found that the older persons’ inadequate social participation and access to social rights significantly varied on their current living status. Those who are live alone reported poor social participation and inadequate access to social rights. Further, older people who live alone reported highest level of social exclusion as compared to those who are living in elder care facilities as well as those who are living at homes.

Socio-Demographic Predictors of Social Exclusion

Nevertheless, the age in itself was not a dimension of social exclusion, the risk for reduced income increases with advancing age (Agulnik, Burchardt, and Evans 2002). Age related characteristics such as disability, cognitive decline, low income and widowhood, labour market characteristics, economic decline and crime in local areas and age based discrimination dispose the elderly persons at risk of social exclusion (Phillipson and Scharf 2004). Less formal schooling, restricted lifelong learning opportunities significantly reduce the capability of individuals and households including acquiring new set of skills, competing in the employment markets, and accessing institutions (Varghese 2011). Mariyam and Jose (2014) found that three significant predictors viz., self-perception of ageing, monthly family income and current earning status. In line with this, we found that the years of formal schooling, monthly family income and current earning status of the older persons have significant inverse effect and explained 10.3 % of the variance on social exclusion. However, studies on the predictors of social exclusion in ageing research do not cover adequately unless we examine concerns such as old age, single person households, poor mental and physical health, inadequate private transport, living in rented homes, and completely dependent on pensions (Barnes et al. 2006). Hence, future studies needs consider to take up variable selections which are conceptually grounded in order to have a comprehensive picture of the social exclusion of older persons.

Conclusion

The gender significantly influences social exclusion where the older women are likely to experience severe level of social exclusion than the older men. The older persons from dalit and tribal groups reported high level of social exclusion than the older persons of other social groups. The unmarried and widowed older persons reported high level of social exclusion than the older persons who are married and currently living with their spouses. The older persons living alone and in elder care facilities reported high level of social exclusion than those who are living with spouse and children. Older persons completed more number of years in formal schooling; family monthly income and current earning status have significantly predicted social exclusion. Hence, the study concludes that socio-demographic characteristics significantly shape social exclusion in older persons.