CHIP50 Trust Tracker

How much Americans trust institutions and public figures, 2020 to today
Visit CHIP50.org →
Who
Loading the survey estimates…

Trust scoreboard

Sort
Click an institution to see its trend. The change column compares with a year earlier: the survey closest to a year before, or the previous year when a whole year is pooled. Sparklines show every survey since 2020, each on its own vertical scale (hover for the range); the COVID-19 wording is dashed.
Show

Bars show each group; thin lines are 95% margins of error; the vertical line is all . The change compares with a year earlier: the survey closest to a year before, or the previous year when a whole year is pooled.

Sorted by the distance between the highest and lowest group. Click a row to follow that gap over time.

The gap over time

Color by

States ranked

Number of clusters

The trust map

Institutions close together are trusted by the same people: someone who trusts one tends to trust the other. Circle size shows how widely each is trusted. Lines join pairs with a correlation of 0.6 or more. Hover over an institution to bring its links forward; drag one to move it, and it stays where you drop it (double-click it to let it go back). The map is turned to line up with the latest year, so positions can be compared across periods; it settles for a few seconds so that names do not overlap, and institutions with few or no links drift toward the edge.

The clusters

Who trusts what together

Each cell is the correlation between trust in two institutions across respondents (weighted). Blue: people who trust one tend to trust the other. Red: people who trust one tend to distrust the other. The tree on the left joins the most alike first; boxes mark the clusters.

How the clusters changed

Do the same people trust both?

The correlation between trust in two institutions in each survey, for everyone and within each party.

Download the estimates

Every estimate is a weighted percentage from one survey wave (or a pool of waves), with its margin of error and the number of respondents behind it. Estimates with an effective sample (after weighting) below are left out.

The columns