Evaluating 100 dashboards can be quite of an endeavor. Interesting findings though.

For the dashboards to be random they shouldn’t come from the same source. This by itself can imply bias.

I suppose in (1) you mean visual cues or elements instead of visuals, which in Power BI can be charts or other graphical representation means, while for (8) it would be useful to mention how many dashboards involved pie charts (12% from 100 is different from 12% from 50).

Explaining acronyms and concepts is usually done outside of the dashboard for organizations setup. Depending on the chart types, some from the points you made in the end make sense but might not apply to all.

Of the chosen dashboards only the first on Pandemic has something appealing. I wonder though if the numbers are reported to the population.

Good idea!