Hardship Among Older Adults in the HRS: Exploring Measurement Differences Across Socio-Demographic Characteristics

Abstract

Measures of hardship have been proffered as better indicators of economic well-being than traditional measures of socioeconomic status (SES). However, there is a dearth of research on latent factor structures and measurement bias in items assessing hardship across socio-demographic characteristics, especially among older adults. As such, the purpose of this study was to determine the factor structure of items measuring hardship in the Health and Retirement Study (HRS) and to determine measurement bias across socio-demographic groups (i.e., race/ethnicity, gender, and age). The participants were HRS subjects who completed an additional psychosocial survey (N = 3074). The results revealed a single latent factor for hardship (comparative fit index = 0.99, root mean square error of approximation = 0.02) using confirmatory factor analysis on eight items in the HRS. The multiple indicator, multiple causes (MIMIC) model was used to determine measurement bias in the items due to socio-demographic characteristics. Compared to white respondents, black respondents were more likely to endorse items of financial dissatisfaction (Odds Ratio (OR) = 2.19, 95 % Confidence Interval (CI) = 1.43, 3.35), while Latino respondents were more likely to endorse food insecurity (OR = 2.78, 95 % CI = 1.60, 4.83); and older individuals (age 65 and older) were less likely to endorse having moved to a worse residence/neighborhood (OR = 0.32, CI = 0.18, 0.57) and being unemployed (OR = 0.28, CI = 0.20, 0.38). These results indicate that there is differential item functioning for specific measures of hardship suggesting that there are differences observed for the measurement of hardship for these items across racial/ethnic and age groups.