n=...), and opacity is proportional to the response rate (respondents ÷ enrolled). The three rating items use different Likert anchors and aren’t directly comparable across panels. The x-axis is treated as discrete so the 2018–2022 gap is collapsed.
