Received and perceived support do not predict the same thing

Received support is measured by asking what concrete help a person obtained over a given period: a hearing, a favour, a loan, information. Such questionnaires draw up an inventory, whose reliability depends on respondents' memory and on what they class under the word help.

Perceived support is measured by asking whether the person believes they could count on those around them if needed. That second measure concerns no actual event: it probes an expectation, formulated outside any particular situation.

The distinction is not a methodological nicety. The two measures go under the same word in everyday language as in many public reports, and what is established for one is frequently presented as holding for the other.

Studies regularly find that perceived support is associated with a better reported state, while received support is weakly so, or even negatively in some studies.

That result, described notably in the review by Sheldon Cohen and Thomas Wills published in 1985 and then in many replications, has been called the support paradox and is stable across populations. It is an association, not a demonstrated effect: nothing in it indicates which way the relationship runs.

The first is arithmetical: one receives help when one needs it, so received support is correlated with difficulties. The measure captures the situation before capturing the help. A person going through a hard period accumulates more help and a worse reported state, without the former explaining the latter.

The second concerns the cost of received support: explicit help may signal an incapacity or create a debt. Niall Bolger, Adam Zuckerman and Ronald Kessler described in 2000 an invisible support, more effective when not identified as such by the recipient, a result several experimental studies have since replicated.

It measures a belief about the availability of those around, which agrees only moderately with the help actually obtained when a need arises. The mismatch is not strange: a belief of this kind forms over a long history of relationships and is not revised at every episode.

It is also associated with stable traits of the person, which makes interpretation delicate: part of what the scale measures may stem from a general disposition rather than from the actual network.

Interventions that increase the help provided do not mechanically alter the feeling of being supported, and available evaluations find dissociated effects on the two measures.

That is one reason why evaluations of schemes against isolation produce apparently contradictory results depending on the indicator chosen.

The finding discredits neither the help provided nor measures of perception. It recalls that they do not answer the same question, and that an assessment based on only one of them describes only part of what has changed.