Loneliness and isolation are not the same variable

Social isolation is objective: it is measured by contact frequency, network size, participation in collective activities. It presupposes no judgement by the person concerned, and it says nothing about how they experience their situation.

Loneliness is subjective: it denotes a felt gap between desired and actual relationships. A well-surrounded person may feel it; a very isolated person may not.

The distinction is not new. The British sociologist Peter Townsend was already drawing it in his 1950s surveys of older people, and Robert Weiss proposed in 1973 to separate, within loneliness itself, an emotional form and a social form.

Surveys measuring both find moderate correlations, regularly of the order of a few tenths. A substantial share of isolated people do not report being lonely, and vice versa.

That dissociation is stable across countries and instruments, which makes it one of the best established results in the field and one of the most ignored in public coverage. It stems from the fact that relational expectations differ greatly from one person to another, and the same network satisfies some and not others.

The two variables do not call for the same responses. Increasing an isolated person's contacts is a concrete operation; altering a perceived gap is not, in the same way.

Evaluations of interventions do find clearer effects on objective measures than on subjective ones, which is consistent with the nature of what is measured. The indicator chosen to evaluate a public policy thus shapes its assessment, depending on whether it counts contacts or records a feeling.

Syntheses by Julianne Holt-Lunstad and colleagues, including the 2015 one that treats isolation, loneliness and living alone separately, find associations between these variables and mortality, of comparable size, which is often read as proof of their equivalence.

It is the opposite: two weakly correlated variables each associated with the same outcome indicate two distinct routes, not one measured twice. These associations remain correlations observed in populations, whose causal direction is the subject of other debates.

French national surveys mainly measure relational isolation, through the frequency of contact with different circles, and publish series on that basis. This is the case of the annual studies by the Fondation de France, a major French philanthropic foundation, which distinguish family, friends, neighbours, work and associational life.

Measures of felt loneliness are more recent and less regular there, which limits comparison over time and explains why the same subject is documented by different indicators depending on the source cited.