Bad Chart Thursday: Magic Eye CAPTCHA Chart Explains Trump’s Popularity
I am going to admit straight out: I was never able to see the image hidden in those magic eye pictures (a.k.a. stereograms) from the nineties. Today, I tried to create my own magic eye picture, and even knowing what image was hidden, I still couldn’t see it.
I’m pretty sure the reason I never see the Magic Eye image is that I do not have a lot of patience for things that annoy me–which brings me to the topic of today’s bad chart: Donald Trump.
Political writers have been trying to puzzle out why Trump is so popular in the polls and what this says about the Republican Party and U.S. voters. Perhaps if they study him until their brains unfocus, they’ll see what these Trump fans see.
Vox has had several articles on Donald, two of them featuring one of the most bizarrely terrible charts I’ve seen in a long while. It is a heady mix of annoyances: tiny, difficult-to-read labels; unclear numbers; unexplained colors; and a Magic Eye CAPTCHA combo effect, because each on its own is apparently not maddening enough.
The chart was created by Lee Drutman and first appeared in his article “What Donald Trump gets about the electorate.” I ran across it shared in another, more recent Vox article by Matthew Yglesias, “The conservative establishment is in deep denial about Donald Trump’s appeal,” where it was particularly confusing because it did not have Lee’s explanatory text.
See for yourself:
I’m going to guess that no one reading this has high-magnification bionic focal zoom vision (because the CAPTCHA should have weeded out the bots), so here’s a more accessible version of the labels:
Vertical, from top: 5. IMMIGRATION should be Decreased a lot; 4. Decreased a little; 3. Left the same as it is now; 2. Increased a little; 1. IMMIGRATION should be increased a lot.
Horizontal, from left: 1. SOCIAL SECURITY should be increased; 2. Kept the Same; 3. SOCIAL SECURITY should be Decreased
The chart is intended to show selected data from the American National Election Studies 2012 Time Series Study to illustrate the percentages of U.S. voters who hold similar positions to Trump on social security (increase it) and immigration (decrease it), positions that are in opposition to those of the Republican party. Drutman refers to Trump’s views on these issues as populist.
He explains the colors as representing his breakdown of voters into 5 categories (not their self-identification): orange = populists, blue = liberals, purple = moderates, light red = political conservatives, and dark red = business Republicans. These colors are unexplained in Yglesias’s article, which is one example of why a chart needs to be clear on its own and not rely on explanation in the text. Even terrible charts get shared, often out of context. As an image on Twitter, for example, with the unexplained colors and tiny labels, we’re left with only the title to explain what we’re looking at. Without that title, this chart is indistinguishable from a chart depicting “Which Skittle are you?” or perhaps a scratch-off lottery ticket. (We were so close to winning 20% of social security.)
The dots are intended to visually represent the percentages, so basically they are completely unnecessary additions that obscure the numbers they are intended to underscore.
Further on in the article, Drutman presents a chart intended to show how his categories break down within self-reported party affiliations:
The mini bar charts (which incidentally do not include overpriced tiny liquor bottles, except those drained by the pixies who apparently chose the type size) are placed side-by-side for comparison, specifically to see how the populist view dominates among Republican voters. These comparisons are misleading, however, because the y-axis numbers and intervals differ from chart to chart. Within a chart, the comparison of bar heights is effective, but between charts, the bar heights do not accurately compare to each other. You wouldn’t know it at a glance, but the upper left chart shows the highest number of populists (with “populists” defined solely by positions on immigration and social security).
The article itself is interesting, despite these bad charts, although any article trying to boil down the reasons for a candidate’s popularity is prone to oversimplifying the complexity of views and motivations among so many people. Odds are, Trump’s positions on these issues, and the fact that they seem more in line with the Republican electorate than the official party line, are factors for some people who favored Trump in the polls. Others no doubt have different motivations.
I tend to think that the appeal of Trump for some people is the same appeal of Fox News or of comedians who rely on being deliberately offensive to make up for a lack of talent and originality–in other words, like Fox News. The audience can live vicariously through the outright, often over-the-top racism and misogyny while still being able to claim that they aren’t racist misogynists–they just agree with his other views and admire his straight talking as a principle, ignoring how the racism and misogyny are often inherent in the views themselves, not just in how he expresses them.
Trump gives voice to their own bigotry in a way that gives them permission to feel that this bigotry is justified and shared by others while patting themselves on the back for not being as racist or as misogynistic as he is. He’s both folk hero and scapegoat all in one.
Trump is himself like a human magic eye. The select few are proud to see the substance beyond his noise, while the rest of us wish that irritating shit would go back to the nineties for good.
Without having run an analysis on these numbers, what I take away from all those plots is that opinion on social security, opinion on immigration and party affiliation are only slightly correlated with each other. In the first chart, the ratios of numbers in the first column are pretty similar to those in the second and third columns (e.g. first row number slightly bigger than second row number, whichever column you look at.) Similarly for the rows (e.g. first column number approx equal to second column number, whichever row you look at.) There is a little more in the second chart: red and brown increase and blue decreases with increasing Republicanship, but purple is roughly half of orange throughout.
Looks like Drutman just discovered ggplot in R and decided to try to cram as many of its features into one plot as possible.
You must log in to post a comment.