Statistical significance refers to the likelihood that the relationship between two variables is caused by chance. When the difference between a minority rate and the white rate are statistically significant, and therefore unlikely to occur by mere chance, the minority bar or circle will have a darker shade in the graphs.
For the metrics associated with each minority group, we calculated whether the values for that group were statistically significantlyFor the purposes of this analysis, statistically significant means a z-score (from now on referred to as significance score) of 4 or higher in the z-test for two population proportions. different than the corresponding values for white drivers. We use the lightness or darkness of the shade of the circle or bar representing the data to illustrate this significance.
If a minority-representing bar or circle is fully shaded, that shows that the minority values represented in that plot are significantly different than the corresponding white values for that agency. If, however, the bar or circle is transparent, this shows that the rate value for the minority group cannot be said to be statistically significantly different from the value for white drivers.
If the rates are not considered to be statistically significantly different, it means one of two things: 1) the rates themselves are similar; or 2) the number of occurrences of an event is so small that even stark differences are not meaningful.
Anywhere "significant" is used throughout this analysis, we are referring to statistical significance.