What the data actually shows
The long-run American picture comes mainly from the General Social Survey, which has asked since the 1970s whether "most people can be trusted" or whether "you can't be too careful." The share choosing "most people can be trusted" has fallen over the decades — from somewhere around half of adults in the early surveys to roughly a third in recent years. The exact figures move year to year, but the downward direction over the long run is one of the more robust findings in the literature.
The cross-national picture is the important corrective. Drawing on the World Values Survey and the OECD's wellbeing data, researchers find that high-trust countries — several in northern Europe, for instance — have stayed high or risen, while the US and some others have slid. Trust is not falling on a single global schedule; it tracks national context, which is itself evidence that the decline is not simply "what modernity does to people."
It is also worth separating two things the surveys measure. Generalized trust (in strangers, in "most people") and institutional trust (in government, media, business) do not move together perfectly. Confidence in major institutions has fallen sharply in the US on many measures, and that institutional erosion is often clearer and faster than the slower drift in person-to-person trust.
The same approximate GSS figures as bars: from roughly half in the early decades to about a third in recent years. A sizable long-run drop, but the direction is the robust part, not the exact percentage.
Why this feels different from how it actually is
Trust feels like it is in freefall partly because the most visible signals — political conflict, viral outrage, scam warnings, alarming headlines — are exactly the inputs that push the "you can't be too careful" answer. The information environment selects for threat, so the felt level of distrust can run well ahead of the slower-moving survey reality.
There is also a gap between what people say in the abstract and how they behave in practice. Most everyday cooperation — leaving packages, trusting strangers with directions, lending small things — still happens at high rates, even among people who tell a survey that "most people" cannot be trusted. The abstract question captures a mood as much as a measured behavior.
And the decline is uneven across groups and places, so your personal experience may not match the aggregate. Trust tends to be lower where inequality and insecurity are higher, which means two people in the same country can be reading very different local realities and both be broadly right about what they see.
The data on the trend is much firmer than the data on the explanation.
What the research says to do about it
If the question is what is associated with higher trust, the research points less to individual habits and more to conditions: lower inequality, economic security, and well-functioning, even-handed institutions tend to travel with higher generalized trust. This is correlational and the direction of causation is debated, but the association is consistent across the cross-national data.
At the personal level, the closest thing to a robust finding is that repeated, low-stakes contact with a range of people is associated with more generous estimates of how trustworthy others are. People whose lives are structured around varied face-to-face interaction tend to give the more trusting answer; people who are isolated tend toward the wary one.
Calibrating your own estimate against the data is itself worthwhile. Because the threat-heavy information environment inflates perceived danger, many people sharply overestimate how untrustworthy strangers actually are. Checking the abstract feeling against your own lived rate of being treated decently is a reasonable corrective.
What the research says does not help
Treating the decline as total, universal, and irreversible does not match the evidence and is not helpful. Several countries have maintained or rebuilt high trust, which means the trend is a result of conditions rather than an iron law — and overstating it can become self-confirming, since expecting the worst of strangers tends to make people behave more guardedly.
Reasoning about "how trusting people are" from social media, comment sections, or the news will mislead you, because those environments amplify the most distrust-provoking material. The aggregate behavior of people in ordinary settings looks far more cooperative than the online signal suggests.
Looking for a single villain — one technology, one group, one decade — does not survive contact with the research. The leading explanations (rising inequality, shifting institutions, media change, generational turnover, and others) are debated and probably partial; no one factor has been shown to account for the trend on its own.
Several countries have maintained or rebuilt high trust — which means the trend is a result of conditions, not an iron law.
What this looks like in real life
Trust that held, or rose
Several high-trust societies — several in northern Europe, for instance — have stayed high or climbed over the same decades the US slid. That single fact does a lot of work: if trust were simply eroded by modern life, it would fall everywhere. It doesn't, which points to national conditions — inequality, security, how even-handed institutions are — rather than an iron law.
What people say vs. what they do
Someone tells a survey that 'you can't be too careful,' then spends the day leaving packages on the porch, giving a stranger directions, and lending a neighbour a tool. Most everyday cooperation still runs at high rates even among the wary answerers. The abstract question captures a mood shaped by a threat-heavy information environment as much as a measured behaviour.
Real numbers in context
In the General Social Survey, the share of US adults agreeing that "most people can be trusted" has fallen over the long run from roughly half in the early decades to about a third in recent years — a sizable drop, though the figures wobble survey to survey and should be read as a trend, not a precise number.
The cross-national data complicates the story usefully: in the World Values Survey and OECD measures, several high-trust societies have stayed high or climbed while the US declined. Generalized trust and institutional trust also diverge — confidence in major US institutions has fallen faster and further than person-to-person trust, so "trust is declining" means different things depending on which measure you mean.
The share of US adults choosing 'most people can be trusted' over 'you can't be too careful' in the General Social Survey. Figures are approximate and wobble survey to survey — read as a direction, not a precise number.
| Period | Say 'most people can be trusted' |
|---|---|
| Early GSS surveys (from the 1970s) | ≈ half of adults |
| Recent years | ≈ a third of adults |