Yeah these graphs are frustrating. The knife one also has a Y axis starting at 600 along with the numbers being a “rolling 12 month average” of stabbings, whatever the hell that’s supposed to mean.
Rolling averages are used to smooth-out graphs with high-frequency noise. Since measuring the stabbings per month only gives you some 50 stabbings on average, maybe one month you’d get 30 and the next month 70 due to stochastic reasons, and so to make the graph smoother and more readable and long-term trends more visible, you can do a rolling average.
As for the “starting in 600”, that’s common practice and it’s good as long as the axes are properly labelled, which they are.
What’s mildly infuriating is that the viral content graph doesn’t show the years where knife crime actually peaked.
Yeah these graphs are frustrating. The knife one also has a Y axis starting at 600 along with the numbers being a “rolling 12 month average” of stabbings, whatever the hell that’s supposed to mean.
Rolling averages are used to smooth-out graphs with high-frequency noise. Since measuring the stabbings per month only gives you some 50 stabbings on average, maybe one month you’d get 30 and the next month 70 due to stochastic reasons, and so to make the graph smoother and more readable and long-term trends more visible, you can do a rolling average.
As for the “starting in 600”, that’s common practice and it’s good as long as the axes are properly labelled, which they are.