Failing to maintain graph axes at comparable scale (thus ignoring absolute variances between countries and regions)
If I plotted the Euro graph to the “most violent countries” scale, it would be unreadable.
Failing to compare regions/country clusters with significantly different correlations (again, willfully ignoring variances between politically, geographically and/or culturally different regions)
Were I to do this, the USA would be the only point on its own graph. I go into this in more depth here:
Failing to provide context for varying correlations (e.g. legislation or cause of gun ownership)
This is actually something I see crop up quite a bit in some left of center studies which claim “more gun laws = less homicides.” In fact, “more gun laws” is a result of fewer gun owners.
Willfully misrepresenting graphs by failing to correctly represent axis descriptors (e.g. “gun deaths” does — by definition — include both homicide and suicide. This is only misleading if you do not understand words).
The context within those articles is clear. They never once mention suicide in suicide dominated graphs, never define gun deaths, and intentionally mislead the reader.
In fact, Vox went back and stealth-edited their intentionally misleading article, which sat in its original state, for years, to correct it. I don’t know if my article had anything to do with this or not. They did so in the fall of this year.
