The data are troubling, no need for embellishment. In 1987, 61.1 percent of Americans were satisfied with their jobs. Over a 22 year period, that percentage has dropped to 45.3 percent. But the AP chart that accompanied the story in today’s (printed) Seattle Times implied that there was a four-fold difference between 1987 and 2009!
Without Numeric Scale
The Miami Herald included the AP graphic produced for the web. Let’s look at that AP graphic stripped of x- and y-axis data.
It’s clear that the column on the far right is approximately one-quarter the size of the column on the far left. If the far left column is 61.1 percent job satisfaction, then based on this visual representation, the far right number should be about 15 percent!
Thus this graphic visually misrepresents the drop in worker satisfaction. This is not a trivial or academic argument. Our brains process the relationships between the lines more quickly – and persuasively – than the numbers. This (mis)representation of the data is what our brains grasp.
AP Graphic As Presented
Notice that the graphic artist has perverted the chart by changing the “0,0” point to be “40,0”.
The web version of the graphic has a teeny visual break between the columns and the x-axis. That break is not part of the print-version of the graphic. (Eventually I’ll scan the Seattle Times image.) Nevertheless, we can also see that the slight fade between the columns and the x-axis is not a visual clue that there is anything out of the ordinary with the y-axis, since the base number is 40, not 0.
There is a standard method for showing when a y-axis starts at any other number than zero. You “break” the y-axis line – literally.
The graphic also misrepresents the time series. There should also be a break in the x-axis as well because 13 years are represented by the first three bars, and then each bar represents one year. Thus it looks like our job satisfaction has had a four-fold drop in less than 10 years!
I place this graphic in the category of “chart junk” because the chart misrepresents the data in two ways and, in so doing, it triggers an “ohmygod” or fear-based response. I believe it is designed to cause an emotional reaction.
This isn’t the first time I’ve opined about a chart that I thought was intentionally misleading (and I’m certain it won’t be the last). This soapbox is related to my complaints about time-series economic data that have not been inflation adjusted.
What do you think? Is the AP chart an example of deliberate deception or laziness or a clueless art director/editor?
2 replies on “Chart Junk: AP Deceives With Job Satisfaction Graphic”
I would assume that the bias is NOT related to the subject itself in any way.
Instead, the bias is much more likely to be the tendency to dramatize a story — to add a wow factor which in the mistaken calculation of some editors make it more likely to sell – or at least hold readers’ attention.
Andy Perkins
The Satisfaction Questionnaire Blog
I remember taking a political science class from a hardcore conservative professor—we spent a good week, at one point, looking at “liberally” skewed news graphs and while I never really bought into the whole liberal bias thing, it was hard to ignore how stupid some of these graphs were. This AP one’s no different.
As to whether the graph is purposely deceptive, lazy or clueless (or some combination of the three)…I have no idea. I know I’m really not surprised by it. If I’m giving the AP the benefit of the doubt, they sized this graph to fit anywhere on a printed page and the skewed illustration is the result. But even then, why use the same graphic online? Your online story exists in a boundless environment—use it.