The closing narrative of the McCain campaign is apparently going to be “Obama’s a pinko commie socialist who wants to raise your taxes,” which means it’s time for all good liberals to bust out the graphs to show why this is false. Well, graph, singular. You know the one:
I don’t remember who first posted it (I got it here), but it’s been everywhere this campaign. It shows a head-to-head comparison of the consequences of the McCain and Obama tax plans for various income groups.
I hate this graphic.
Not because of the information it contains, mind– that’s fine. I hate this graph because it does such a terrible job of presenting the information it contains.
I’m not just talking about the fact that it blindly uses all the idiotic default settings of Microsoft Excel (well, ok, it doesn’t have the dark-grey background, but it’s got all the rest). This is a graph that only a policy wonk could love.
First of all, what’s being plotted? The axis labels don’t really help– “Percent” is only slightly more illuminating than “Y Data.” You need to look up at the title to find that it’s “Average Percentage Change in After-Tax Income, 2009.” Not only is that information that needs to be on the graph itself, not just in the title, it’s a confusing quantity to plot.
I understand why they did it that way– you’d need a log scale to plot the absolute dollar values and keep everything visible, but you can’t do that since the quantity changes sign for Obama. Still, this is a terrible variable for making your point. “Average Percentage Change in After-Tax Income” is not an intuitively obvious thing– the average person does not have a terribly good grasp of percentages. It also tends to understate the hugeness of the difference at the high end, because you’re taking percentages of larger values.
The horizontal axis is even worse. The first five bars represent the five income quintiles. Two quick questions, and no fair Googling:
- What’s a quintile?
- What incomes fall in the third quintile for the United States?
Actuaries and policy wonks may think in income quintiles, but this is not a common quantity. The average voter is not going to grasp this concept– something like 20% of people famously think their income is in the top 1% (though some people challenge that statement). Adding the bars for the top 1% and top 0.1% don’t really help, either.
So, basically, the graph-maker has plotted confusion as a function of obfuscation. The graph shows a percentage change in income, which most people won’t understand very well, plotted as a function of income quintile, which most people won’t understand at all.
All this graph really gets you is “The bars get bigger at the high end, and Obama’s bars go negative.” This works wonderfully with people who have a realistic idea of where to locate themselves on the horizontal axis, but it’s a dismal failure otherwise. For a lot of under-informed voters, it probably plays right into the McCain narrative, as people will say “Well, I’m middle-class, so I must be in the middle, maybe a bit above, and hey! Obama’s going to raise my taxes!”
There is a vastly better version of the same graph floating around (I found it here):
This is much more like it, and not just because it avoid Excel’s dreadful aesthetic. It’s the same information, but it turns the graph on end, and gives actual numbers for all the important quantities. The income ranges are defined by top and bottom incomes within each range, so there’s no room for confusion about where a given voter should fall. There are bars showing the percentage change for each group, but also boxes giving the absolute dollar value of the average tax change.
This graph lays out the difference even more starkly. The difference between plans for millionaires is almost $1,000,000 (a big increase from Obama, a big cut from McCain), while the difference between tax plans for someone right in the middle of the middle class is about $700 in Obama’s favor.
This is how you present the relevant information properly. The dreadful Excel chart up top is fine for preaching to the liberal choir, but won’t do anything but confuse an undecided voter.
In addition, since three of the seven entries are devoted to the top quintile (with no context given to let people know what that is), the first graph could be misinterpreted as Obama raising taxes on almost half of the population. Even though the second graph devotes a larger fraction of its real estate to the top earners, at least we know what the ranges are, and there is little danger of anybody misplacing themselves on the chart unless they were already close to a bin boundary.
Are you sure chart #1 was made by an Obama supporter and not a McCain supporter?
An even better representation is this graph, which I found here. It takes the same information as in the last graph you posted, but scales the categories in relation to each other.
Your first graph is modified from here:http://www.taxpolicycenter.org/UploadedPDF/411693_CandidateTaxPlans.pdf.
The original does a better job of putting the last two columns apart from the quintiles. But I agree, the second is better for talking to the general public.
Oh, and Chad, with all the free time you have now, being a stay at home dad for the semester you couldn’t follow three links back to the original?
An even better representation is this graph, which I found here. It takes the same information as in the last graph you posted, but scales the categories in relation to each other.
That is nice.
The only thing it’s missing that I would like to see is the absolute values of the tax increases/ decreases. You need the percentages to make the graph, but the absolute numbers tell the story better.
I should also note that I think Obama has done a pretty good job with this, always making sure to say not only that he cuts taxes for 98% of the population, but also that the cut-off is an income of $250,000 per year. The absolute dollar amount is critical, given how bad most people’s grasp of percentages is.
Oh, and Chad, with all the free time you have now, being a stay at home dad for the semester you couldn’t follow three links back to the original?
No.
That’s what commenters are for.
Stay at home parent . . . free time.
Stay at home parent . . . free time.
Stay at home parent . . . free time.
Sorry, does not compute.
The graph Evan linked to has exactly what I want to see: the relative change in taxation related to the number of people in the tax bracket. Put that way, it’s blindingly clear that McCain is proposing yet another big tax break for the wealthy with crumbs for the majority of Americans.
I agree with Chad, I’d like to see absolute numbers in addition to the percentages and I’d like to see the net difference in overall tax revenue between what we have now and what the candidates propose. At the end of the day, I want to know how much more or less I’ll be paying along with what the country is getting to pay for the bank bailouts, both wars, etc.
Yes, but I don’t like the absolute numbers on the top tax brackets. You’ve got people earning $3million, who are going to pay $300k extra, mixed in with people earning $300million, we are going to pay an extra $3million. The average they show, about $1million, is going to be skewed enormously by the tiny numbers of people at the ultra high end of the income scale.
Some come close but in case direct explanation was missed: one problem with the top graph, it has overlapping categories shown as separate graph bars. IOW, we have a “top quintile” which includes everyone in the top 20% of earners (and, I think most people can understand that idea with a brief and simple explanation.) However, the next bar shows “Top 1 Percent” which is a subcategory of the top Q’ile, and the bar after that is “Top 0.1 Percent” which is a further subcategory of the top one percent. So indeed, that makes the total impact seem more because it looks like three separate groups are shown with big drops in their remaining income.
BTW, the complaint by McC/Palin that a slightly higher graduated income tax (not just the having, versus not) is “socialist” is misleading and irks me. Hey, when are those who say they don’t like “redistribution” by tax rates, going to prove your honesty by opposing the current cap gains rate already being lower than the tax on people who, ironically, actually did work hard to produce new value and earn the money directly? (I mean the base CG tax rate before any inflationary “time” adjustments – I support indexing.) McCain wants to make the CG rate even lower! I don’t want my earned income to be redistributed to mostly worthless traders/traitors!
I don’t get it: There are 6 tax brackets, not 5, 7, or 9. (as depicted by the above graphs) Taxes change (typicaly) for the whole bracket, not just parts of them. You /can/ use deductions as an argument for additional breaks for one or two brackets, but assuming that those would apply to only one or two tax brackets I believe would be folly. But even regardless of that, how is it decided that the division of tax breaks is divided where they are. I could understand if at least the lower part of the chart followed the 2009 or 2008 Income tax brackets, but the chart made above doesn’t even do that.
WhiteTiger, read the PDF – the changes here reflect lots of different data: bracket percentages, alternative minimum tax, credits, exemptions. Credits all have different thresholds of course. Then they made some assumptions about who would get which credits. Combined these create as complex a system as you can imagine for different income amounts. They have then averaged the tax amounts within each quintile to attempt to reflect a realistic end result. (This is my lay-person’s interpretation.)
There is plenty of room for bias to have been introduced in this document. Whatever your views, read the source yourself. Don’t decide based on a graph.
But even regardless of that, how is it decided that the division of tax breaks is divided where they are. I could understand if at least the lower part of the chart followed the 2009 or 2008 Income tax brackets, but the chart made above doesn’t even do that.
Wesele – Because the graph is a composite of several different kinds of data, including (but not limited to) the effects of the actual marginal-tax cutoffs, in practice both are smooth curves. But people are terrible at understanding smooth curves, so we use brackets, with the average within a bracket chosen to represent the members of the bracket as a whole. The quintiles are chosen because (a) they correlate with some of the sources of data, so using them minimizes errors in converting that data to a curve, and (b) they are more easily grasped by most people.
You could argue that setting the cutoffs at round numbers ($50K, $100K, etc) would be more intuitive. But the nice thing about quintiles (or any N-tiles) is that they scale things by percentages of the population. If we used round numbers then we’d have non-round numbers for what percentage of the US population falls into each slot – so it’s non-round numbers one way or the other, either in dollars or percentages. Pick one.
To the wonks, the use of quntiles is equivalent to the much more useful visual display Evan linked to. So they tend to think on those numbers.
Besides, come back to this graph next year and all of your potential round numbers of dollars turn into inflation-adjusted screwy figures anyway. Quintiles stay quintiles.