They're not, though? E.g., https://www.youtube.com/watch?v=fKv1Mixv0Hk&t=278s — note that the final bar is also shorter, so really you need to elongate it a bit in your mind (and compress the bar above it): the proportion of the "many adverse experiences" group is definite greater than the other two. (I wish they'd've just labelled the %'age on the screen, made the bar lengths equal — I have a lot of issues with the data visualization here, but none severe enough that they defeat the core point of the video.)

Edit: okay, I've counted the miniature people on this chart. For this specific example, they are: no adverse exp.: 7 aff, 109 total; some adversity: 16 aff, 239 total; many adverse exp.: 24 aff, 152 total. In percentages, that's "No adverse experiences" → 6.4% victims of crime, "Some adverse experiences" → 6.7% victims of crime, "Many adverse experiences" → 15.8%. The last group is more than double the other two. (The first two, in this example are equal; but the visualization also roughly shows that.)