Four attitudes, one discontinuity

Weekly weighted means, January 2020–February 2021. The dashed line marks George Floyd's murder (May 25); the faint line marks Election Day. Flat horizontal segments show the nine-week pre- and post-Floyd averages.

Takeaway: each series is roughly flat through the spring, then breaks sharply at the week of May 25 — discrimination perceived against Black Americans and racial sympathy up, police favorability and racial resentment down — before drifting back toward where it started.

Nationscape weekly waves, weighted; thin partial weeks (n < 250) dropped. Per week n ≈ 3,400–12,600. Source: Democracy Fund + UCLA Nationscape.

The shift reached almost every group

Estimated jump at Floyd in a standardized racial-sympathy index (z-scored composite of the four items, higher = more sympathetic), by demographic group, with 95% confidence intervals. The vertical line is the national average jump.

Takeaway: Democrats and Black Americans moved most, but Republicans, white Americans, the old, and the least-educated all moved too — the dots sit to the right of zero across the board.

Each estimate is the level shift from a separate weighted interrupted-time-series regression within the group (Feb–Aug 2020), standard errors clustered by week. Index combines perceived discrimination, the two resentment items, and Black favorability. Source: Nationscape.

The shift was nearly nationwide

Change in weighted mean perceived discrimination against Black Americans (five-point scale), nine weeks after Floyd minus nine weeks before, by state. States with fewer than 120 interviews in either window are left unshaded.

Weighted pre/post difference; small-sample states (and those Nationscape does not separately identify) unshaded. The largest measured increases are in smaller states where weekly samples are thin, so individual state magnitudes carry wide error; the national breadth is the robust signal. Source: Nationscape.