4 ms·
What is the color and size scale here? It could be deeply misleading, for example if it’s a linear scale counting number of anomalies in a bucket (like number o
by mlthoughts2018 8y ago
What is the color and size scale here? It could be deeply misleading, for example if it’s a linear scale counting number of anomalies in a bucket (like number of +2.0 anomaly measurements in a given country).
From the upper right, it seems like it’s I guess the average per year per country, placed in bins for -2, -1, 0, 1, 2, ... except how were the bins constructed? In reference to global temperature in some specific year? Country-average in some specific year? Very hard to read or know the significance of it.
I hate visualizations like this for exactly this reason. The video hits me in the gut and makes me want to believe the final two decades of data are alarming, but it’s not clear at all what statistical claim is being measured, and how unlikely that specific claim would be under some counterfactual distribution.
I’m not denying (or confirming!) the implication of the data. Just pointing out that this is sort of like “ooh shiny” shallow attention grabbing, which IMO does more harm than good.
Statistical visualizations should not feel like they have an agenda in the idiosyncratic choice of axes, scale, colorbar, etc. It’s just not clear to me here that it’s not got one.
- lucb1e 8y agoThe scale is visible in the right top corner (literally in every frame of the video). Did you just happen to overlook it or are you asking something else?
- mlthoughts2018 8y agoI meant that the scale in the upper right is uninformative. If the circles’ sizes and colors were based on country-specific average anomaly divided by country-specific standard deviation of anomaly, based on a post-industrial reference year, say in the 50s-80s, then the colorization and size changes would possibly be useful. Basically bucketing Wald statistics instead of temperature anomaly, and ensuring it’s all always country-specific.
- mikeash 8y agoThe intent of this visualization is to hit you in the gut. If you want something more concrete, there’s plenty of it out there. We can have both kinds of things.
- mlthoughts2018 8y agoI disagree in the strongest possible terms. Manipulating presentation of data to create an emotional reaction so that policy decisions are made emotionally is not ok. If a presentation of the data that conforms to standard techniques for an unbiased representation would not result in “hitting you in the gut” then that has to be accepted, and worked with, not sacrificing intellectual integrity to elide the data and change things for an agenda.
- DFHippie 8y agoThis is a fine sentiment (double entendre intended), but some people, enough to sway public policy, cannot be reached by a simple presentation of facts. If these people control policy, and one party to the debate has no qualms about controlling them through appeals to their gut, what should one do? If in 100 years humanity is reduced to an impoverished remnant living in the ruins of its greatness, will they be grateful that we kept our rhetoric pure? (Note, I'm not talking about lying, only about presenting information in a way designed to appeal to the audience's emotions.)
- TangoTrotFox 8y agoI'm trying to get you to invest in my fund. I give you a set of cherry picked graphs that indicate I'm earning an unbelievable amount of money with minimal to no variance. Then somebody else shows you a complete graph showing my longterm profits (which are rather different than what I lead you to believe) and other data indicating that I was being somewhat less than completely honest with you. What's going to be your opinion of the value of my fund? Of me? Did I do anything wrong, or can I just claim I was only trying to "appeal to the audience's emotion"? Myopic thinking and Machiavellian logic are a great combination to undermine the very ends you aim to achieve.
- DFHippie 8y agoIn your example you are attempting to deceive. That is the immorality. Not the appeal to emotions.