In 1985, the average American reported having three people with whom they discussed important matters. By 2004, that number had fallen to two. By the early 2020s, a significant portion of Americans reported having one confidant or none. The confidant count — the number of people you would actually talk to about something that matters — is one of the most precisely tracked metrics in the sociology of social connection, and its trajectory across four decades of American data is one of the clearest signals of structural social deterioration in any developed society.
The confidant is not the same as the friend. The friendship count, as the preceding article on cross-cultural census data established, is inflated by definitional looseness and cultural variation. The confidant is operationally defined by the survey instrument: the person to whom you actually disclose, the person with whom you discuss important personal matters. It is therefore a measure of something more specific than friendship — a measure of the inner sanctum, the part of a person's network that actually bears the weight of their inner life.
Confidant data from the General Social Survey, which has tracked this variable since 1985, tells a story of loss that is not well-represented in public discourse. The 2004 McPherson, Smith-Lovin, and Brashears analysis of the GSS data documented a 33 percent reduction in mean network size and a near doubling of the percentage of Americans reporting zero confidants. The finding was published in the flagship journal of American sociology and received sustained academic attention. It has not produced equivalent cultural or policy response.
Global data on confidants is patchier than the American data, partly because the GSS's specific operationalization — naming the people with whom you have discussed "important matters" in the past six months — has not been consistently replicated in other national surveys. But the available data from Western Europe, Australia, and parts of Asia suggests that the decline is not uniquely American. It is a feature of late modernity in developed societies, produced by the same structural conditions: residential mobility, labor market churn, declining institutional participation, and the substitution of digital contact for the kind of repeated in-person presence that builds confidant-level trust.
The confidant count matters not merely as a social welfare statistic but as a predictor of health outcomes. The research connecting confidant availability to mortality, disease progression, and mental health outcomes is among the most robustly replicated in epidemiology. A person with zero confidants faces a measurably higher risk of premature death than a person with three. This is not a soft social finding. It is hard health data, as reliable in its effect size as the relationship between smoking and lung cancer.