I’ve shown before how misleading it can be to pick a very particular start or end date in a statistical time series for which more data are available than the data you choose to show. Here’s another example: in an effort to whip up some old fashioned hatred of asylum seekers, the French newspaper Le Figaro offered this graph:
A continuous and steep rise in the numbers! 70% increase in 5 years! Take it easy I would say. The increase would have looked a lot less ominous had they chosen to start the y-axis at 0 instead of 30.000. The most important bit of manipulation, however, occurs in the x-axis: go a bit further back in time, and you get this (look at the green line):
But of course they don’t want to take it easy. Angst is precisely the point, and choosing the low-point of 2007 as the cherry-picked starting date of the graph is intended to maximize angst. They don’t want people to realize that the current numbers are nothing more than a return to normal.
With thanks to @jacksometer for the pointer.