The cult of raw unadjusted data
A nice plot of densities, but what's behind the colors?

The art of making simple things harder

It's no longer a shock when a TV network such as MSNBC plays loose with the scaling of the column heights, as in this recent example:

Rachelbitecofer_markp_2024candidatescashonhand

Hat tip to Mark P. for forwarding the image, and Rachel for the original tweet.

***

What's shocking is that the designer appears to believe that the column heights of a column chart can be determined without reference to the data.

There is not a single relationship that has been retained on this chart. The designer just picks whatever size column is desired.

One obvious distortion is between the Biden and Trump columns. Trump's number is about 1/3 of Biden's (120 vs 40), and yet the red column's height is 70% of the blue's.

Furthermore, amongst the red columns, the heights are also haphazard. Trump's number is almost 3 times larger than Haley's; the ratio of column heights is almost 4 times. Haley's number is just a tad higher than DeSantis and yet Haley's column is twice the height of DeSantis.

Junkcharts_msnbc_candidatecash_analysis

***

There is a further, subtle distortion of the column's widths. By curving the chart canvas, certain columns are widened more than others. The diagram above retains the distorted widths and you can see that the Desantis column is wider than that of Haley's.

Here is what the undistorted column chart looks like:

Junkcharts_redo_msnbc_candidatecash

It's easy to make such a chart in Excel or any charting software, so it's mystery why this type of distortion happens. Did the designer open up an empty canvas and start putting up columns of any size?

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

The comments to this entry are closed.