Objective and Subjective Measures Differ in Rating Older Adults’ Well-Being

When policymakers try to identify the most disadvantaged older adults, the answer they get depends surprisingly heavily on how they choose to measure disadvantage. That is the central conclusion of a new study published in Social Indicators Research by Veerle Van Loon and Koen Decancq of the University of Antwerp and KU Leuven, who compared three fundamentally different methods for assessing well-being in later life—objective, subjective, and preference-based—and found that they frequently point to completely different people as being the worst-off. In a sample of 813 adults aged 50 and older in Flanders, Belgium, fewer than seven percent of respondents landed in the bottom fifth of the well-being distribution under all three approaches at once.

The research tackles a problem that has grown more urgent as populations age. Measuring the well-being of older adults has long outgrown the narrow focus on disease and disability that defined early gerontology, and contemporary researchers now agree that well-being is a multidimensional concept encompassing health, social support, financial security, leisure, meaningful activity, and more. But agreeing that well-being has many dimensions is not the same as agreeing how to collapse those dimensions into a single score that can guide policy. Without such an aggregation, comparing two people—one thriving socially but struggling financially, the other the reverse—becomes impossible. And without comparison, there can be no monitoring of inequality, no evaluation of policies, and no principled allocation of scarce resources across populations.

The three aggregation strategies the researchers examined each rest on a different answer to a deceptively simple question: whose judgment should decide how much each dimension of life matters? The objective approach lets experts decide, often by simply giving every dimension equal weight. Classic examples in gerontology include the Rowe and Kahn model of successful aging and the CASP-19 quality-of-life scale, both of which impose a top-down definition of what a good later life looks like. Critics have long objected that this is paternalistic—the weights may not reflect what older people themselves actually care about. The subjective approach goes to the opposite extreme, simply asking people how satisfied they are with their lives and treating that self-report as the ultimate arbiter. It respects autonomy, but it carries well-known vulnerabilities: people use rating scales differently, some have “expensive tastes,” and many adapt their expectations downward in response to prolonged hardship, reporting contentment despite objectively poor circumstances. The preference-based approach, rooted in economic theory, tries to split the difference by empirically estimating how much weight older people themselves place on each life domain, in a way that is insulated from the distortions of adaptation and scale-use differences.

The study’s methodological innovation was to compare all three approaches on exactly the same dimensions for the same individuals, removing a major weakness of earlier comparative research. The six dimensions were health, social relations, income, leisure, engagement in meaningful activities, and religion, each coded into four levels from worst to best. Combining the levels across six dimensions produced 4,096 possible “well-being states,” described by six-letter codes—a structure reminiscent of the EQ-5D health-state classifications used in health economics. Under the objective approach, each level received a score from 1 to 4, summed across dimensions, so a state of all-worst levels scored 6 and all-best levels scored 24. Under the subjective approach, respondents read a personalized vignette describing their own life across the six dimensions, generated from their earlier survey answers, and rated their satisfaction with that life on an 11-point scale. For the preference-based approach, the researchers ran a factorial survey experiment: each respondent evaluated seven randomly composed vignettes depicting different combinations of the six dimensions, and a multilevel regression model converted those evaluations into relative importance weights for every dimension level.

The experiment revealed what older Flemish adults actually prioritize. Health commanded the largest weight, followed by social relations and income; leisure and engagement mattered less, and religion—despite being included because qualitative research flags it as salient for some populations—carried essentially no weight, or even a slightly negative one, for most respondents. The weights were also nonlinear: moving from moderately severe to mild health problems mattered more than moving from mild to no problems, and an income rise from 1,500 to 2,700 euros counted for more than subsequent gains up to 5,000 euros. Because the dimension levels in the vignettes varied randomly, these weights could be interpreted causally, a significant advantage over observational estimates based on regressing life satisfaction on circumstances.

When the researchers ranked the same 813 respondents using each approach, the results diverged dramatically. The Spearman rank correlation between the objective and subjective rankings was a mere 0.27, indicating that knowing where someone stands on the expert-weighted index tells you almost nothing about where they place themselves in life satisfaction. The preference-based approach aligned far more closely with the objective one, with a correlation of 0.85, yet even here 44 respondents crossed the critical 20th-percentile threshold in opposite directions under the two methods. The subjective and preference-based rankings correlated at just 0.34. Defining the worst-off as the bottom 20 percent under each method, fully 35.4 percent of the sample fell into that zone under at least one approach, but only 6.7 percent were identified consistently by all three. The subjective approach was the biggest outlier, sharing less than nine percent of its worst-off group with either other method and singling out a distinct 10 percent of respondents whom neither of the other approaches flagged.

More revealing still were the sociodemographic profiles of the different worst-off groups. On basics like age, gender, and household size, the three methods agreed broadly. But on education, employment, and health they split in a patterned way. The objective and preference-based approaches both identified worst-off groups dominated by people with lower educational attainment and greater physical disadvantage: tertiary education accounted for only about 30 percent of their worst-off groups, versus 52.4 percent in the full sample, and disability prevalence ran at 47 to 57 percent. The subjective approach told the opposite story. Its worst-off group was half tertiary-educated—matching the sample average, with essentially no educational gradient—contained a higher share of people still employed, and showed the highest depression scores of the three groups, averaging 4.2 on the CES-D scale compared with roughly 3.3 for the other two.

The researchers interpret this asymmetry through the lens of well-established theories of aging. Processes of adaptation and selective optimization, described by Paul and Margret Baltes and elaborated in socioemotional selectivity theory, mean that people recalibrate their aspirations and evaluative standards as they age. Older adults with declining health or limited resources may nonetheless report high life satisfaction—the well-documented “well-being paradox”—while those still grappling with psychological distress, often in midlife rather than old age, rate their lives poorly. The subjective approach, in other words, drifted away from structural conditions and toward mental health, identifying a group whose disadvantage lies in depression rather than material deprivation. From a social-justice standpoint, this is a serious concern: a policy that allocated resources purely on self-reported satisfaction might systematically bypass people facing the most adverse objective circumstances simply because they had lowered their expectations.

The authors are careful about the limits of their findings. The data were collected online in May 2020, at the tail end of the first COVID-19 wave, a period that reshaped priorities—research elsewhere found that health and social relationships gained salience during the pandemic—and it is unclear whether those shifts are temporary. The sample came from a non-probability online panel, underrepresenting the very old, migrants, and lower-educated adults, so the results are descriptive rather than population-representative. The preference weights were estimated as a single homogeneous set, which cannot capture the fact that religion may matter enormously to some individuals even if the average respondent ignores it. And because the subjective analysis lacked anchoring vignettes to correct for scale-use heterogeneity, some of the divergence attributed to adaptation may partly reflect differences in how people use rating scales.

Even with those caveats, the message for policy is stark. Choosing an aggregation method is not a technical detail to be settled by convenience; it is a normative decision about whose perspective on the good life should prevail—experts’, the individual’s own felt experience, or the individual’s stated priorities. The study shows that these choices are not innocuous: they produce different lists of who needs help most, with different educational, occupational, and health profiles, and therefore different implications for who receives support. The authors argue that no single approach is definitively correct, and that the real demand is transparency—making explicit the assumptions behind whatever measure a government or agency adopts when it claims to identify the most vulnerable older adults.