Chart redraw: Troops Vs. Cost (Time Magazine)

Time Magazine published a very boring combo dual-axes chart with a broken scale. Most of the time these charts beg for a connected scatterplot, so I made the one above. The original chart was something like this: time-budget-vs-costs I’m sorry, Time mag,  but my chart tells hands down a much more interesting story.

[UPDATE] Made a new, annotated version.

15 thoughts on “Chart redraw: Troops Vs. Cost (Time Magazine)”

  1. I have to admit, I find the dual axes version easier to read. It takes more thought to work out what the story is in the scatterplot, at least for me.

  2. Thanks Naomi. I think dual-axes charts are even more misleading, deceptive and dangerous than a simple bar chart with no zero.

  3. Serg: I’d like to know that too. There is no note/reference in the original chart and I had no time to read the full article to check.

  4. Álvaro: I wasn’t “forced” to use the time series. I wanted to add a chart to show variation for each series. They were added not because people can’t understand the main chart but because they add detail to the analysis. To help reading it I added arrows that explain what each turn means.

    The purpose of this chart is to show variation of one series against the other. You can’t do that with a dual-axes charts, no matter what you do.

    Your example and the original chart can’t show all the twists and turns in the data, mainly the changes every 10 years.

    Now it takes more time to understand the scatter plot, but that’s just because you have read more of them. Just wait and you’ll see.

  5. No matter the choice of graph, the implicit message in presenting this data is that there is something to be gleaned from looking at the correlation (or lack thereof) between these two variables. Your redesign makes it very clear that headcount and budget can vary widely independently of each other, perhaps more than one might expect.

    I wonder whether two parallel time series–number of troops and budget in dollars/troop–might support the same conclusions while making it easier for the viewer to overlay his or her knowledge of recent administrations (figuring out which party to blame seems to be the implied next step in any governmental spending graph).

  6. Good discussions, I don’t like dual-axis charts, but aside from the axis not starting at zero I think Álvaros’ chart is easy to read and it has a definite chronological sequence I think transposing the axis makes it clearer with budget along the x-axis and Number of troops up the y-axis

  7. Nigel: a traditional line chart or something a little less traditional like Alvaro’s is one of the best chart types we can use and I wouldn’t replace it with a connected scatterplot to see how a series changes over time. That said, there are (or aren’t) relationships between two variables that you can’t see with a line chart and only when you plot one against the other some things start to make sense or even you discover things you didn’t suspect.I think this chart makes changes to the American strategy more explicit than in the original chart

    Yes, it’s easy for us to read a line chart, and a connected scattterplot requires some effort if you are not accustomed to read it. But I would argue that only when you are comparing the design of the same chart types you can say one is easier to read than the other. If you have two chart types you have to add the quantity and quality of your insights to the equation.

    In a different post I used a connected scatterplot to show the impact of parameters when calculating population projections that remain hidden in a regular line chart.

  8. Fair points Jorge. I think by making the narration/story telling and highlighting more obvious, would be a great help

  9. Very well done. I like your redesign and agree that it clarifies the data. Maybe the original _graph_ is easier to read, but the _data_ is easier to understand in the connected scatterplot.

Comments are closed.