More on Representation

I just applied for a Science Communications course in Banff, so I'm all about looking at science in the media these last days.

The results of a survey of 2,164 American adults conducted by the Yale Project on Climate Change and the George Mason University Center for Climate Change Communication have just been released <pdf>. The answers to "How much do you think global warming will harm:" are summarized as such:

Never mind there is no y-axis label, % is clearly implied. What's nice is you can see the decrease of harm as you move from left to right, i.e. closer to the level of the individual. It's very straight forward and the result is discussed in context in the essay. In a recent post at fivethirtyeight.com, the data are redrawn and presented to the public as such:

In no way whatsoever would such an inverted pyramid be acceptable in scientific publications to represent this form of data. There are two major concerns here. First, the red arrows point from the above value, not from the root. This indicates that 61% of people that said yes to "Plants and Animals" also said yes to "Future Generations" (i.e. 819 respondents), however what is actually meant is 61% of the total n of 2,164 replies (i.e. 1320 respondents). If the nested percentage is followed through to the last stage, we are to believe that 5 people successively answered yes to all possible options, including "You" (i.e. 0.25%). But respondents were allowed to pick and choose, they are allowed to choose "You" and not choose "Plants and Animals". There is no basis for nesting the arrows as such and the data become falsely represented. Second, the size of each pyramid level is not proportional to the percentage they are expected to represent. If each block was proportionally representational, then the lower "You" would be roughly half the size of the upper "Plants and Animals". The inverted pyramid dramatizes the trend to absurd levels. Although it is certainly a pretty picture to look at. Both of these faults mean that the figure is not easily understandable and does not convey it's information accurately. The pity is that they original figure is just fine and shows exactly what the authors discuss in the text.

The last point I want to make clear is that Plants and Animals are great and I'm sorely hurt to hear their fate is so poorly predicted. Canadian musicians need our support! :)