Comments on: Consumer Credit Graph [Chart Review] http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/ Sat, 30 Apr 2011 08:44:13 -0400 hourly 1 http://wordpress.org/?v=3.1.3 By: DSA Insights » Blog Archive » Seasonality in Data http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/comment-page-1/#comment-2384 DSA Insights » Blog Archive » Seasonality in Data Thu, 04 Dec 2008 12:28:06 +0000 http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/#comment-2384 [...] Links « Consumer Credit Graph [Chart Review] [...] [...] Links « Consumer Credit Graph [Chart Review] [...]

]]> By: derek http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/comment-page-1/#comment-2381 derek Wed, 03 Dec 2008 18:40:40 +0000 http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/#comment-2381 I agree that not starting at zero is important for this line graph, but I don't agree that that makes the zigzag a good idea. Instead, the zero symbol should be simply omitted altogether, instead of being introduced only as something that then needs to be counteracted by a lightning bolt. It's something I see a lot from people who don't get that the horizontal scale is a measure of the distance from the left, and the vertical scale is a measure of distance from the bottom, not vice versa. The horizontal scale is not a "zero line". Cartographers understand this, as they put their scales anywhere on the map that's convenient, and know they don't need to explain that it's not the equator. I agree with Jon that this is a poor data set for a line graph anyway, because it's mainly a straight and monotonic ascent, with a small descent on the end. It needs text to explain either that the small downturn is unusual in the context of the recent past, and is therefore evidence of some sort of turnaround, or that the negative change is small, contradicting perhaps some claim that credit outstanding has massively dropped. This graph is failing to opine, and, like people who similarly fail, risks its audience going away not having learned what it thinks it knows. A longer baseline might have told a stronger story. Perhaps the line has been monotonically increasing for so long that this small fluctuation is shocking. Or perhaps fluctuations have occurred frequently in the past, so that this one is not shocking in context. Or maybe it needed to be normalized to the growing US population or the growing US economy, to show a more horizontal line with a more shockingly steep drop. I agree that not starting at zero is important for this line graph, but I don’t agree that that makes the zigzag a good idea. Instead, the zero symbol should be simply omitted altogether, instead of being introduced only as something that then needs to be counteracted by a lightning bolt.

It’s something I see a lot from people who don’t get that the horizontal scale is a measure of the distance from the left, and the vertical scale is a measure of distance from the bottom, not vice versa. The horizontal scale is not a “zero line”. Cartographers understand this, as they put their scales anywhere on the map that’s convenient, and know they don’t need to explain that it’s not the equator.

I agree with Jon that this is a poor data set for a line graph anyway, because it’s mainly a straight and monotonic ascent, with a small descent on the end. It needs text to explain either that the small downturn is unusual in the context of the recent past, and is therefore evidence of some sort of turnaround, or that the negative change is small, contradicting perhaps some claim that credit outstanding has massively dropped. This graph is failing to opine, and, like people who similarly fail, risks its audience going away not having learned what it thinks it knows.

A longer baseline might have told a stronger story. Perhaps the line has been monotonically increasing for so long that this small fluctuation is shocking. Or perhaps fluctuations have occurred frequently in the past, so that this one is not shocking in context. Or maybe it needed to be normalized to the growing US population or the growing US economy, to show a more horizontal line with a more shockingly steep drop.

]]> By: Jon Peltier http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/comment-page-1/#comment-2380 Jon Peltier Wed, 03 Dec 2008 18:23:42 +0000 http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/#comment-2380 Tony - I saw that caption. It was not so much a description of the particular chart (it said nothing about the chart) as it was a general statement that didn't change as you scroll through the five charts. Tony -
I saw that caption. It was not so much a description of the particular chart (it said nothing about the chart) as it was a general statement that didn’t change as you scroll through the five charts.

]]> By: Tony http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/comment-page-1/#comment-2379 Tony Wed, 03 Dec 2008 17:32:06 +0000 http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/#comment-2379 Excellent point Jon! The fact that the y-axis is only 3.5% of the total range, magnifies the trend exponentially. I would really have to see the rest of the data to say it the scale is warranted. Yes, the decline starting in July 2008, does signify something, which is probably what the designer had in mind. Here is the text that accompanied this chart in BusinessWeek to give more context. Consumers Crunched: "The Standard & Poor’s 500-stock index has skidded 23.2% since Sept. 1 amid a continuing housing bust and a consumer credit contraction. And despite emergency actions from global central banks, including the coordinated Oct. 8 rate cuts, panicked investors have driven world markets into a nosedive". (David Foster) Excellent point Jon! The fact that the y-axis is only 3.5% of the total range, magnifies the trend exponentially. I would really have to see the rest of the data to say it the scale is warranted.

Yes, the decline starting in July 2008, does signify something, which is probably what the designer had in mind.

Here is the text that accompanied this chart in BusinessWeek to give more context.

Consumers Crunched:
“The Standard & Poor’s 500-stock index has skidded 23.2% since Sept. 1 amid a continuing housing bust and a consumer credit contraction. And despite emergency actions from global central banks, including the coordinated Oct. 8 rate cuts, panicked investors have driven world markets into a nosedive”. (David Foster)

]]> By: Jon Peltier http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/comment-page-1/#comment-2378 Jon Peltier Wed, 03 Dec 2008 17:13:19 +0000 http://supportanalytics.com/blog/2008/12/consumer-credit-graph-chart-review/#comment-2378 The caption of this chart in Business Week gives no context. The entire span of data in the chart is about 3.5% of the initial amount. The dip from July to August is 0.3% of the peak. Are these significant enough to warrant a scale like this? Whatever the magnitudes, the change in the last month signifies something. Why did credit go down? Did enough people start defaulting to reverse the rise of credit card accounts? The caption of this chart in Business Week gives no context.

The entire span of data in the chart is about 3.5% of the initial amount. The dip from July to August is 0.3% of the peak. Are these significant enough to warrant a scale like this?

Whatever the magnitudes, the change in the last month signifies something. Why did credit go down? Did enough people start defaulting to reverse the rise of credit card accounts?

]]>