Likely based on surveys. This is pretty common for these sort of things. Depression questionnaires can help track severity of depression over time for example. Often in treatment people might not even notice how much progress they've actually made, it's great to take out past questionnaires and show them the actual numbers and tends to give a real boost of positivity afterwards.

Most likely based on survey results on perception of mental health issues. So results can be quantified based on the outcomes of those surveys. Example: a person takes a survey and results show they responded favorably to 50% of proposed questions. They watch the video and take another survey with different but similar questions and now respond favorably to 60% of the asked questions, which would then represent a 20% increase in questions that responded to favorably.

At 20 minutes, all racism ends