I applaud Sam Jaffe's efforts to improve scientific data presentation.1 However, his statement that "bar graphs are a universal no-no" is not warranted. Worse yet, his suggested remedy is flawed. He suggests using line graphs to illustrate categorical data instead of bar graphs. The problem is that line graphs are based on the assumption that the data are continuously variable, unless there are breaks in the X axis. Also, points on a line graph do not solve his basic objection to bar graphs, which was that bar graphs emphasize "the size of the columns, whereas the only point that's important is the top of the bar." A line graph of the same data has the same problem: the only point that is important is the height of the data points, not the lines. The lines actually misrepresent reality.
I vote for histograms. Pie charts are an alternative, but then he ...
