Data
Data for the present study were drawn from the first wave of Longitudinal Study of Ageing in India (LASI wave-1) conducted during 2017–18. The survey was conducted by the International Institute for Population Sciences (IIPS), Mumbai, in collaboration with Harvard T. H. Chan School of Public Health (HSPH) and the University of Southern California (USC) under the stewardship of the Ministry of Health and Family Welfare (MoHFW), Government of India. LASI is a nationally representative longitudinal study of ageing and health that also covers the economic and social aspects of population ageing in India. The multistage stratified area probability cluster sampling method was used to select the sample. Within each state, a three-stage sampling design in rural areas and a four-stage sampling design in urban areas was adopted in the LASI wave-1 [27]. The study covered a total sample of 72,250 individuals aged 45 years and above and their spouses, irrespective of their age. Of which, around 31,464 were older adults aged 60 years and above [27]. The data is collected from 35 states and union territories of India (excluding Sikkim). LASI is envisioned to be conducted every two years for the next 25 years. The number of targeted primary sampling units (PSUs) in a state was given proportionally to each sub-state area in the first step, the selection of PSUs (sub-districts or Tehsils/Talukas) (level 1 stratification). The PSUs were chosen using Probability Proportional to Size (PPS) sampling in each area, with the number of households in each PSU serving as the size measure. The second stage entailed selecting a predetermined number of secondary sampling units (SSUs) from the selected PSUs, which are villages in rural regions and wards in urban areas. The third step in rural regions entailed selecting a number of households (HHs) (i.e. 32) from each designated village or village segment (for villages with more than 500 HHs). In metropolitan regions, the fourth round of selection entailed selecting a number of HHs (35 in this case) from each Census enumeration block (CEB). The interviews were conducted using computer-assisted personal interview (CAPI). The sample included for the present study was 31,464 older adults aged 60 and above.
Outcome variable
The outcome variable used in the study was the perceived age discrimination based on a set of questions asked to the respondents. First, respondents were asked how often the below-listed things have happened to them in their day-to-day life: 1. you were treated with less courtesy or respect than other people; 2. received poorer service than other people at restaurants or stores; 3. people act as if they think you are not smart; 4. people act as if they are afraid of you; 5. threatened or harassed; 6. receive poorer service or treatment than other people from doctors or hospitals. The possible response options were recorded on 1 (almost everyday) to 6 (never) scale. Further, they were asked about the perceived reasons for such discrimination that included: age, gender, religion, caste, weight, physical disability, physical appearance, financial status and other reasons. Respondents who reported any experience of discrimination related to their age were used as sample of perceived age discrimination. Others, including those who did not experience any discrimination and experienced discrimination on other reasons excluding age, were considered as sample of not perceived as age discrimination. Perceived age discrimination takes the value ‘1’ if the respondent reported ‘yes’ otherwise, it takes the value ‘0’ representing no.
Explanatory variables
Respondents' age was recoded into three categories: 60–69, 70–79, 80 years and above. Gender was categorized as male and female. Marital status was recoded as currently in marital union, widowed and, currently not in marital union (divorced/separated/deserted/live in relationship/never married) [28]. Educational attainment was classified as no education, 1–5 years, 5–10 years and, more than 10 years of education. Living arrangement was categorized as living with spouse and children, living with children and others, living with spouse and others and living alone. Social participation (member of any social organizations, religious groups, clubs, or societies) was coded as no and yes. Working status was categorized as never worked, earlier worked but currently not working and currently working. Residence was coded as rural and urban. Monthly per capita consumption expenditure (MPCE) quintile was classified as poorest, poor, middle, rich and richest. Caste was categorized as Other Backward Classes (OBC), Schedule castes and Schedule Tribes (SC/STs) and General (other than OBC/SC/ST). Religion was classified as Hindu, Muslim and others.
This study included four health measures i.e., self-rated general health (SRH), ability to do activity of daily life (ADL), ability to do instrumental activities of daily living (IADL) and chronic condition. SRH had a scale of 1 to 5 from “very good” to “very poor” and was categorized as 0 as good (representing very good, good and fair) and 1 as poor (representing poor and very poor) [29]. To quantify ADLs, respondents were asked, “Have you any difficulties in dressing, walking, bathing, eating, mobility and toilet?” A composite index was constructed from the questions mentioned above. The response variable “difficulty in ADL” was described as 0 as “no” and 1 as “yes” [30, 31]. The Cronbach’s alpha value for ADL scale was 0.869. To quantify IADLs, respondents were asked, “Have you any difficulties in preparing meal, shopping, making telephone, medication, doing work in garden or home, money handling and getting around?” A composite index was constructed from the questions mentioned above. The response variable “difficulty in IADL” was described as 0 as “no” and 1 as “yes” [31]. The Cronbach’s alpha value for IADL scale was 0.879. Respondents were asked about nine chronic conditions and one composite index was calculated to measure chronic conditions. Further, response variable 'chronic condition' was categorized into three categories: 0 as 'no chronic condition', 1 as 'having one condition', 2 as 'having 2 or more chronic conditions'.
Statistical analysis
Univariate, bivariate and multivariable analyses have been conducted to examine the prevalence and factors associated with perceived age discrimination among older adults in India. Initially, descriptive statistics were performed to describe the variables of interest. Next, bivariate analysis with a chi-square test was employed to investigate the association of various socio-demographic and health-related factors with the perceived age discrimination. Further, a multivariable logistic regression model was used to determine the significant predictors of perceived age discrimination. The odds ratios of experiencing perceived age discrimination are reported by adjusting for various socio-economic covariates. Variance inflation factor was estimated to measure the multicollinearity among the variables used [32]. All the statistical analysis was performed using STATA-14.2. Additionally, the weights were applied which make the results nationally representative.