> The bar chart with independents, republicans and democrats makes it seem like every group is equally large.

The chart was about relative take on the subject in each constituency. But again bar charts prove their superiority: it's perfectly possible to scale the bar charts according to the size of the charted group (although you're going to lose the comparability of relative constituencies, which again was the point), almost impossible to do so with pie charts.

I was making no claim about showing the relative sizes of those groups. And I don't believe the three pie chart version was either.

This does point out the burden of bar charts: are can be unintentionally more informative than you wanted them to be.

I believe that even if this was the case then sausages of varying lengths would be clearer than pies of varying areas.