Network News

X My Profile
View More Activity

Lies, damn lies, and the 'Y' axis

Alexander Hart catches the GOP's "Pledge" in some serious chart-based trickery. Here's their graph showing that President Obama has exploded the size of the government:


Looking at the chart, Obama seems to have slightly more than doubled the size of the government. But notice the numbers on the left: The "Y" axis begins at 17 percent. Here's what it looks like if it begins, as it should, at 0:


Three pages later, the document complains that "Americans have lost trust with their government." I wonder why?

Update: Commenters noted that Alex's graph brought the Y axis to 100, when the data only went up to the low-20s. That also isn't a very honest way of presenting the data, they said, and they're right. I should've looked at it more carefully. Here's how I'd make the graph:


You're dealing with about a 3.5 percentage point increase in the size of the government as a percentage of the economy -- at least on the graph. More precisely, government was 20.8 percent of the economy in Bush's final year and is projected to settle at about 22.8 percent in 2013, 2014, and 2015. So there's been an increase, but not a doubling.

By Ezra Klein  | September 23, 2010; 10:40 AM ET
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: The budget in 2060
Next: Lunch Break


Both graphs are pretty dishonest.

Posted by: amr1776 | September 23, 2010 11:13 AM | Report abuse

In other news, a dark pot was overheard making remarks regarding the darkness of a kettle.

Wouldn't the graph be honest if it began at zero and stopped at 20%? This would make the Obama/Pelosi government expansion "off the charts." There are some health care cost/benefit graphs that might serve as examples...

Posted by: rmgregory | September 23, 2010 11:17 AM | Report abuse

Oh god. Neither of these graphs do a good job showing the importance of the figures presented. Very "honest" to plot this in 20% point increments.

Posted by: wswest | September 23, 2010 11:32 AM | Report abuse

Most of us are capable of reading the y axis. The second graph is actually more dishonest because the numbers are indecipherable. At least the first graph shows that spending has increased from 19.5% to 23.5% of GDP (as my eyes are calibrated) from Bush to Obama.

Posted by: Beagle1 | September 23, 2010 11:33 AM | Report abuse

A chart which has 0 and 100 on the Y axis when all of data are between 19.5 and 23.5 is also dishonest. On top of that, the very wide bars obscure the differences further, as the eye notices the area of each bar and notices they are relatively similar.

Show a chart with the Y axis from 0 to 25 and the same bars as the Republicans used. That will be an honest chart, and it will be pretty obvious that the federal government will consume about 20% more of GDP (or 4 percentage points for of GDP if you prefer) under the Democratic roadmap than it did under Dubya.

While the Republican chart is indeed presented dishonestly, a casual observer is more likely to draw the correct conclusions with it than the so called "honest" chart.

Posted by: justin84 | September 23, 2010 11:36 AM | Report abuse

I think there's a difference between "honest" and "fair". The Republican graph is not honest because there is no reasonable reason to bound the graph at 17% and 24% other than to exaggerate the differnces. Ezra's right that the original graph makes it look like the last bar is twice as big as the others. And you know what? It *is* twice as big on the graph, but the last bar's numbers aren't nearly twice as large.

Ezra's graph is not dishonest. In a graph about percentages it's not dishonest to plot the graphs on a Y-axis with 0% as the minimum and 100% as the maximum. And the size difference between the bars is roughly proportional to the difference in the numbers. Is Ezra's chart fair? Well, I guess that's a matter of perspective. I certainly think so, but I think there's a reasonable argument that bounding the graph at 0% and 25% would be a reasonable way to make their argument and a couple percent at the scale we're talking about is a serious difference that should be clear to the viewer. But there's a difference between bounding the graph and the seemingly arbitrary 17% and 24% and bounding it at 0% and 25%. The latter may be more fair to the argument the Republicans are making than what Ezra's doing, but the former is just dishonest manipulation.

Posted by: MosBen | September 23, 2010 11:56 AM | Report abuse

I agree that the second chart isn't very useful either. The y axis should start at 0 but should only go up to maybe 30% or so... Possibly more misleading is the use of "ave" spending during the Clinton and Bush terms, which hides the fact that federal spending as % of gdp fell under Clinton and increased under Bush.

Posted by: tcjutras | September 23, 2010 12:00 PM | Report abuse

The obvious dishonesty:

The eight years of Clinton and eight years of Bush are history and can be plotted. The "Democratic Budget Blueprint" charted runs ELEVEN years (??) and of course is a projection not historical data.


Posted by: LucasLazor | September 23, 2010 12:05 PM | Report abuse

You tell 'em Ezra! Way to speak up.

Posted by: ania8 | September 23, 2010 12:06 PM | Report abuse

the second graph is honest.
the possible range is 0 to 100. the graph covers the whole range.
starting at 17?
too funny in a fratboy little penis kind of way.

Posted by: newagent99 | September 23, 2010 12:06 PM | Report abuse

Thanks for that childish interlude, Ezra.

Posted by: theo2709 | September 23, 2010 12:09 PM | Report abuse

Frankly, even the updated graph is still not a good representation. Make the Y axis up to 30% and use 1% as Hart did; from 0 to 30%.

Posted by: trumeau | September 23, 2010 12:23 PM | Report abuse

It is also relevant that the economy shrank after Bush left office. If Obama had done nothing new, government spending as a share of the economy would have risen.

Posted by: hunterda98 | September 23, 2010 12:30 PM | Report abuse

There's a little extra dishonesty in the presentation. The graph is (finally) a decent representation of the quantity. The quantity itself, however, is a fundamentally dishonest representation of the claim, that Obama has dramatically increased the government. The "government" divided by "economy" number gets bigger when "economy" decreases. It's a bad comparison among administrations with dramatically different economies.

Posted by: bill_who | September 23, 2010 12:34 PM | Report abuse

bill_who - I agree. I also think that taking an avg of the bush years and comparing them to Obama isn't very honest. Why not compare just the last year of Bush when the economy started crashing and government had to respond?

On the other hand, it's not like a little chart manipulation is the only thing misleading in the Republican plan...

Posted by: Levijohn | September 23, 2010 12:40 PM | Report abuse

Honesty does not mean the perfect version of the graph to maximally make the speaker's point without, well, cheating. The original graph cheats, and is dishonest. Ezra is making the point that we're talking about a few percentage points increase, which isn't a huge increase. The graph bounded at 0 and 100 makes this point without any manipulation. It is not dishonest at all. A graph bounded at 0 and 30 would be useful to someone arguing that the "size of government" has increased by hundreds of billions of dollars, which is a lot even if it's only a few percentage points. That hypothetical graph would not be dishonest. Ezra's updated graph splits the difference. It is not dishonest. Arguably I think the updated graph is the most "fair", but again, that's a judgment call and something we can argue about, but neither of the two latter graphs are dishonest.

Posted by: MosBen | September 23, 2010 12:50 PM | Report abuse

Your correct about the baseline problem of the GOP histogram. I would have more respect for this post if it weren't for the fact that your blog and the Post use this same technique to distort data all the time, see:

Posted by: BottyGuy | September 23, 2010 1:02 PM | Report abuse

If you decide to make a graph showing the differences in three "things" that are valued at 52%, 60% and 45% would you honestly automatically run the graph to 100% and present it without any breaks? Why even graphically present the information at that point? Breaking an axis happens all the time and certainly isn't "dirty pool." It occurs all the time in all types of research. If you get "tricked" by a simple bar graph, you are in for it when political advertisements begin to plague our TVs.

Posted by: RisingTideLiftsAllBoats | September 23, 2010 1:15 PM | Report abuse

Who cares about the chart-based trickery when the whole Pledge is nothing more than a farcical attempt to create smoke and erect mirrors. There's no there, there!

Posted by: AuthorEditor | September 23, 2010 1:16 PM | Report abuse

So basically the Republicans were being dishonest about something the President did or didn't do?
And we are surprised by this....

Posted by: vintagejulie | September 23, 2010 1:19 PM | Report abuse

Since Ezra is a fan of 1) criticizing tax cuts for their 'lost revenue' contribution to the deficit, and 2) bar charts, I have an idea for him:
Create a bar chart with a Y-axis that goes 0% to 100%. Then, plot one bar that shows growth in federal REVENUES from 2000-2008, and another bar beside it that shows growth in SPENDING 2000-2008.

Then, paste the bar chart in another article explaining again how Bush tax cuts caused the deficit. The bar chart will help readers finally understand the fallacy of Ezra's continued tortured-economic logic that the giant deficit was caused by tax cuts, and will be fixed by letting the cuts expire.

Posted by: dbw1 | September 23, 2010 1:24 PM | Report abuse

Can we take this as a pledge that charts with dodgy X and Y axis scaling will never again darken the door of Ezra's blog?

One of my particular favorites in the category of questionable Y axis scales are the global warming charts.

Posted by: bgmma50 | September 23, 2010 1:37 PM | Report abuse

If we're going to discuss optics, once has to mention the width of the bars. The bars are so wide (interestingly, this is true on all the charts) that it distorts the height relationship that a bar graph is supposed to illustrated. Having the bars be a little less zaftig would give a more reasonable impression of the actual difference.

Seriously, do the same chart with the bars half-as-wide as they are now, and tell me that doesn't magnify the difference between the bars.

That being said, here's the best chart representing recent government growth:

I think that makes what's going on a little more clear. The Republicans should have used that chart. ;)

Posted by: Kevin_Willis | September 23, 2010 1:43 PM | Report abuse


Thanks for the updated charts, though one nitpick - the title should be federal government size as a percentage of GDP, not just government size. Total government size will probably be about 40% of GDP (give or take) during that time period.

"Honesty does not mean the perfect version of the graph to maximally make the speaker's point without, well, cheating. The original graph cheats, and is dishonest. Ezra is making the point that we're talking about a few percentage points increase, which isn't a huge increase. The graph bounded at 0 and 100 makes this point without any manipulation."


The dishonesty of Ezra's second chart has bamboozled you.

The increase in the federal government's share of GDP *is* a huge increase. The difference between the two averages is 4% of GDP by my eye. Using OMB GDP numbers, that's about $9 trillion in extra spending from 2009-2020 above what it would otherwise be if spending were to be capped at Bush era levels as a percentage of GDP. Note that the total 2009-2020 deficit is expected to be only $8.5 trillion per OMB.

Now you can say that a lot of that spending is automatic and that spending would have risen somewhat anyway under a Bush budget. You can also note that Bush's last year was a lot higher than his average. But it is not a small increase by any stretch of the imagination. The increase accounts for the entire deficit of the period and then some.

"the second graph is honest.
the possible range is 0 to 100. the graph covers the whole range."


Would it be honest for insurance companies to chart the percentage of policyholders whose policies they rescind each year in a chart from 0 to 100? What about a chart showing the percent of the American deaths each year, with the PPACA and without it (using Ezra's numbers)? In theory both of these could hit 100%. That being said, it would be dishonest/misleading to use the whole 0-100 scale. In these two cases, it would be dishonest even to use 0-10 as the scale given the data.

Posted by: justin84 | September 23, 2010 1:51 PM | Report abuse

I wonder if Ezra is a fan of the chart posted over on the Fix today, showing that the more people found out about what was in the Obama health care reform plan, the more they turned against it....and why Democrats running for office are staying as far away from it as possible in their campaigns.

You would think Democrats would be proud to brag about passing such a well thought out and historic piece of legislation...

Posted by: dbw1 | September 23, 2010 1:57 PM | Report abuse

In addition to the above comments... is there any reason to plot this as averages in a bar graph? Every chart I've seen of spending over time is a line chart with a data point each year. I'm guessing this would make the comparison look even less favorable for Republicans...

Posted by: blackwagonwrxsti | September 23, 2010 3:01 PM | Report abuse

The GOP's graph is fine because its bounds are within the range that government spending has remained since 1946. No need to go to 100% because that's communism and no need to go down to zero either. The 17% to 24% range is the bounds of a reasonable argument about the size of our government as it stands today. Obama's budget sets us on a course to spend 23.5% of GDP by 2020 according to his OMB. Visualizing that rise in a way that makes it stand out is appropriate considering it is a departure from our post-WWII average of 19.6%. And that change is significant. There is nothing unethical about designing a graph so that it would seem significant to the viewer.

Posted by: hjindc03 | September 23, 2010 3:30 PM | Report abuse

Ezra is such a lying hack here.

Here is one graph where you didn't start the Y axis at 0. There are many others.

Posted by: krazen1211 | September 23, 2010 4:54 PM | Report abuse

"bill_who - I agree. I also think that taking an avg of the bush years and comparing them to Obama isn't very honest. Why not compare just the last year of Bush when the economy started crashing and government had to respond?"

Even the last year of the Bush Presidency had less spending than the Obama Presidency. And that's with much of the 2008 tarp being repaid in 2009.

Makes more sense to read a whole book, though, and not the ending.

Posted by: krazen1211 | September 23, 2010 4:56 PM | Report abuse

Justin84, I wasn't bamboozled by anything, but the chart is not, or shouldn't be, the beginning or the ending of any debate. The Republicans want to argue that there's this huge increase in government spending as a percent of GDP during the last two years, and that that's a big problem. It's fine to have an honest chart to support their position, and the people here suggesting charts bounded by 0 and 30 are proposing, I think, honest charts to support that argument. Someone else, maybe Ezra, will make the argument that we're talking about a few percents, which amount to a lot of money, but which aren't exactly some kind of government takeover of whatever and look, here's a chart bounded by 0 and 100 and it shows that there's just a small increase. That's an honest chart too. Which chart is fair is just a matter of reasonable debate and is as influenced by how convinced you are by the accompanying arguments as anything else. And yeah, the line between honest and dishonest is might get a little tough, but I think the original chart is an unfair and clear attempt to mislead the viewer rather than a reasonable support for any arguments.

Posted by: MosBen | September 23, 2010 5:45 PM | Report abuse

Can we just stipulate that Republicans are lying snakes and go from there? Further discussion would at least proceed from a fact-based premise.

Posted by: aliancia | September 23, 2010 6:19 PM | Report abuse


We can agree to disagree, but given that the increase in government spending from 2001-2008 to 2009-2020 accounts for more than the cummulative 2009-2020 deficits, an extra $9 trillion (roughly twice as much as the Bush tax cuts) I don't think there is much support for saying the spending difference is small.

I do think you can attack the Republican's overall point from different angles (look at Bush's last year, if Bush were President now spending wouldn't be projected at 19.5% of GDP for 2009-2020 because of the recession or Social Security or Medicare, the Republicans are all talk and will spend just as much if in power, etc.), but the jump in spending is significant given the experience of the past few decades.

Moreover, while you can look at 19.5% or 23.5% of GDP and say the difference is small and both are reasonable, there are other levels of government to consider. The government share of the economy is now in the low 40% range (just shy of 44% in 2010). That's within spitting distance of many of the European welfare states which are around 50%. A couple of percentage points in either direction matters quite a bit.

Posted by: justin84 | September 23, 2010 7:05 PM | Report abuse

It is nice to see that someone called out the GOP for dishonest representation of the numbers, but also note the numbers provided are also dishonest.

The graph shows a Bush administration average of about 19.6% of GDP. This number is reached by averaging deficits for FY2001-2008, however FY01 was proposed and signed by Clinton and FY09 would be the eight year of Bush budgets. If you take the FY01 out and add FY09, the Bush average tops 20% (even if you shift a full 2% of the BushFY09 deficit to ObamaFY10 to account for the ARRA and other Obama proposals).

A more accurate graph would use the following averages:
Clinton FY94-01 = 19.4%
Bush FY02-09 = 20.2%
Obama FY10-17 = 23.4%

An honest graph would be one that would show the Obama bar as 20% larger than the Clinton bar because the intent is to reflect their relative size to each other.

Posted by: MassachusettsLiberalinDC | September 24, 2010 12:44 AM | Report abuse

Post a Comment

We encourage users to analyze, comment on and even challenge's articles, blogs, reviews and multimedia features.

User reviews and comments that include profanity or personal attacks or other inappropriate comments or material will be removed from the site. Additionally, entries that are unsigned or contain "signatures" by someone other than the actual author will be removed. Finally, we will take steps to block users who violate any of our posting standards, terms of use or privacy policies or any other policies governing this site. Please review the full rules governing commentaries and discussions.

characters remaining

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company