A question of percentages - need input | WallStats.com The Art of Information

A question of percentages - need input

Posted on 25. May, 2009 by Jess in death and taxes

So one of the first issues to arise in creating the Death and Taxes poster for 2010 is that of percentages.

Normally, for all the data points I list the Agency/Program name, followed by the dollar amount, then the percentage change from last year. However this year it’s a bit tricky because all of the stimulous funding that was approved late in ‘09 is reflected in the ‘09 budget totals. Which mean that many of the ‘09 budget figures are grossly inflated and so comparing ‘09 to ‘10 leads to misconceptions.

Let me show you. This is how the Department of Energy would look if I labeled the same way I did in previous years.

Now on face value it looks as if President Obama has gutted the Energy budget by slashing it by 66%. The truth is that the stimulus money inflated the ‘DoE’s 09 budget to $70 Billion, and for ‘10 it’s budget returned to normal. This is the case with hundreds of programs and departments.

So what to do about it? Well here are a few examples of what I could do. I could use two percentage figures per stat (see below). The first one is the 09′ to ‘10 comparison, the second is the ‘08 to ‘10 comparison which is a more more accurate reflection of the direction of the budget level. This way, programs that receive stimulus money should have a negative for the first percentage, though that seems a bit counter intuitive.

The next example uses the first percentage as an ‘08 to ‘09 comparison and the second percentage as an ‘08 to ‘10 comparison. The good thing about this example is that the stimulus funding is reflected in a intuitive manner, with a positive percentage change, however this creates the problem of having no direct ‘09 to ‘10 comparison.

The third example bypasses ‘09 and it’s stimulus influx all together by comparing ‘08 to ‘10. This is probably the simplest method, but of course, there is no ‘09 comparison and the percentage figures are higher than they normally would be because its comparing over a 2 year span, rather and a single year.

So look over these and let me know which you think would be best suited to use on the 2010 poster. Of if you think there is a better way to do to, please let me know. Of course, all of this will have to be labeled and explain on the poster itself.Also note that there is really no way to separate the stimulus money from the ‘09 totals.

The figures (in billions) for the Dept Energy are:

2008	22.739
2009	70.092
2010	24.047

Thanks for your input!

Tags: death and taxes, questions, statistics

1Trackbacks/Pingbacks

Pingback: Death and Taxes 2010 - Data only. | WallStats.com The Art of Information on June 16, 2009

17 Comments

Robert Kosara, May 25, 2009:

Interesting problem. I would suggest picking the more important (and sensible) figure, and make that the focus, plus show the stimulus-inflated number in a way that makes it clear that it’s secondary (smaller, less saturated/darker color, etc.).

In this case, I’d say make it “Dept. of Energy 24.047bn +6% (+208%)”, with the number in parentheses being the de-emphasized one.
Kenton Jacobsen, May 25, 2009:

Let me just say first that I’m a big fan of your work, I found your stuff on Mint, then came here.

I know that for me looking at your presentation of the data, the thing I want to see is the single year change, since the data is skewed for this particular set, I think I would most want to see the 08,10 number. If there were another way to note that this particular department saw a significant spike the previous year due to the stimulus, I wouldn’t mind that, but another percent in there would just confuse the issue I think.

Great work, keep it up!

-Kenton Jacobsen
Tim, May 25, 2009:

Jess, is it possible to allow for the stimulus spending?

For example if the stimulus spending for Dept OF Energy = “X” then deduct X from the prev year total and work out the % change based on what is left. You could show that as the emphasised figure and the total incl stimulus as the de-emphasised figure. More work of course.
Sam, May 26, 2009:

If you can get the data, I think the most usefull way would be to do the ‘09 to ‘10 comparison with the stimulus money taken out and either call out the stimulus money separately on the line (last item) or make a whole other poster just for it. Otherwise I think the ‘08-’10 comparison is the more usefull and easy to understand with raw doller stimulus info, not percentages.

I love the poster and other infographics. Keep it up!

-Sam
Jess, May 26, 2009:

Well as i mentioned previously, there is no way to split out the stimulus funds from the 2009 totals, otherwise it would be a no brainer.

So I think the best solution is to use the suggested that Robert suggested, keeping the 08 v 09 figure but de-emphasizing it. It doesn’t add a extra layer of complexity to the chart, but it also adds so much data on the stimulus, and inline too.
Nathan, May 26, 2009:

It’s too bad the stimulus money can’t be sifted out. That would be incredibly interesting to have pre- and post-stimulus compared against each other.

Since that’s not the case though…I’d go with Robert’s suggestion too.
Jason, May 26, 2009:

I think Robert’s solution is the right one, too. You could even go with a smaller font size or a slightly more transparent font to really de-emphasize things. I’m surprised that with all the focus on transparency in government at recovery.gov there isn’t more info available about how the stimulus funds affected the various programs.
Juhan Sonin, May 26, 2009:

This should not be an exercise in using “sensible” figures.
It’s an exercise in Truth.

Use the actual budget line number from 2009 and projection for 2010 for the diff: $70 billion in 2009 and $24 billion in 2010. Those are the real numbers, regardless of what’s “sensible”.

You’re running into a problem with your current design. It maybe time to evolve into a completely different treatment… and one that allows people to easily see (and even hack and reuse) the diffs over time (with sparklines, bar charts, etc).

Keep pushing Jess!

-Juhan
Jess, May 27, 2009:

Juhan,

I have long given up the search for the ‘truth’ in data visualization. I’d rather show mediated meaning than the raw truth (as if there is truth to begin with in federal budget data), because the unmediated truth is ripe for distortion.

Presenting the ‘09 to ‘10 numbers alone would make it seem as if the Obama administration gutted the entire government. When the opposite is true. It’s a matter of scope and scale. If there is a spike in a data set, should I portray the end of the the spike and quick fall to normalcy AS the normalcy.

The truth doesn’t exist, but at least I can equip the reader to make a better and more informed decision.
Juhan Sonin, May 27, 2009:

There are several truths especially in regards to the US budget. Show the multiples and diffs (OMB data vs NYTimes assessment vs …).

I agree that a trend is important. How has each line item/budget changed over the past 5 years? Since the 1st budget?

I want to see and sort the biggest to the smallest budgets. That alone is a hugely useful chart. Add in total % of the budget allocation over time and average allocation per line item (from the 1st recorded budget or the past n years) and there’s a powerful story.
KM, June 2, 2009:

Jess…

I’m only 1-day new here but I really Love your stuff. My 2-cents is this:

Compare 2010 to 09 (post-stimulus) like the first example shows (i.e. -66%). That’s as factual as can be.

Use an asterisk on line items where stimulus money was used and create an addendum poster strictly showing how stimulus appropriations were allocated (which would be a cool poster in itself…!)

I’m assuming that you’ve already created a 2008 poster and 2009 (original budget and post-Stimulus) posters for those hardcore fans who want true side-by-side comparisons.

Good Luck…!
KM, June 2, 2009:

Hey…

Just (re)read the thread and, in particular, how its not possible to separate the 09 stimulus money from the original budget… That’s a shame… but still, I’d recommend comparing 09 (post-stimulus) to 2010. ten years from now (and multiple stimulus packages later) you’ll thank yourself for sticking to simple year-over-year analysis.

The primary purpose of your poster is, first and foremost, to show the money trail for a particular year. Unless you come up with a brilliant way to graphically compare 3-5 years at a glance (…and I don’t put it past you…) then I would stick with “current year vs. previous year” analysis…
Doug Hall, June 9, 2009:

If you want to use one of the 3 you have proposed, I would go with the 3rd. As you note, that results in a 2 year change, something many people will not catch.

I believe a better alternative is to calculate the average annual compounded rate from the 2 year rate. For a small increase like 6%, that would be 3% per year. For a two year change of +28% however, the compounding effect means that the average annual rate would be 13%.

I suggest this from experience. The New Hampshire state budget is a two year budget (not one year like most states). Thus, when the press says “the proposed budget is up 8% from the current budget” most people don’t realize that is an annual rate of increase of 4%. I was Executive Director of the NH center for Public Policy Studies for 10 years and did a lot of state budget analysis. I always had to convert the 2 year changes into compund annual increases for people to understand.

Your overall material could describe the situation. As a suggestion, “The 2009 budget was unusual and contained one-time distortions because of stimulus money. For this chart, the percentage change shown is calculated using 2008 as the base year and showing the average annual increase to reach the 2010 level” or something like that.
Jess, June 9, 2009:

Doug, I do like that suggestion. It does slightly unnerve me that the ‘08 to ‘10 figure is twice what it should be. I will consider your suggestion.
sushu, June 15, 2009:

Something like
Dept. of Energy 24.047bn +3% (s+208%)

But what are you going to do with the ones that didn’t receive stimulus money?
Erik, July 10, 2009:

I think the stimulus money should be in the budget poster. It is part of the money “WE” will be required to cover as taxpayers. Since you have notes, I think the second method you proposed melded with Doug’s suggestion of taking the 08-10 increase and dividing by 2 for the second number will provide the most useful and intuitive explanation. If you leave the stimulus money out of the equation, you are, in my opinion, letting the agencies that receive that money do so without proper accountability.
Rosjanki, August 29, 2009:

Very interesting…