Declining trust in institutions is one of the most cited findings in social research. The underlying measurement is considerably less stable than the headline treatment suggests.
The question determines the answer
Asking about confidence in an institution as an abstraction produces different results from asking about a respondent's most recent direct experience with it.
People frequently report low confidence in a category while rating their own encounters positively, which suggests the abstract question is measuring something closer to general sentiment.
Small wording changes matter too. Naming a body, describing its function, or offering a middle option each shift the distribution enough to change the headline.
Response rates have fallen substantially
Survey participation has declined sharply over recent decades, and the people who still answer differ from those who do not in ways related to civic engagement.
Weighting corrects for observable characteristics such as age and region, but it cannot correct for an unobserved willingness to talk to strangers about institutions.
If trusting people are more likely to respond, measured trust is biased upward, and if the reverse holds it is biased downward. The direction is genuinely uncertain.
Comparisons over time carry hidden changes
Long trend series often span shifts from door-to-door interviewing to telephone and then to online panels, each of which produces slightly different answers to identical questions.
Analysts adjust for these transitions, but the adjustments are estimates, and a trend spanning several decades accumulates more of them than most summaries acknowledge.
Careful series run old and new methods in parallel for a period so the difference can be measured directly. Where that overlap was skipped, the correction is an assumption rather than a finding.
Trust and trustworthiness are not the same
A fall in measured trust may reflect institutions performing worse, or better public information about performance that was always poor, or changes in what the public expects.
These have opposite implications. The first argues for reform, the second for continued disclosure, and the third for a conversation about expectations rather than performance.
A single confidence figure cannot separate them, and most reporting does not attempt to. Distinguishing the three requires asking respondents what specifically they believe an institution is doing wrong.
What the data can still support
Comparisons between institutions asked about in the same survey are more reliable than levels, because the wording and sample are held constant across them.
Sharp movements following a specific event are also informative, since the methodology did not change between the two waves that bracket it.
Treated that way the research remains useful. Treated as a thermometer reading of national mood, it carries more precision than the instrument can deliver.