### Fixing my Graphs: Redo of Life Expectancy Graphs from This Morning's Post

by meep

If you saw this morning’s post when I put it up originally, and then looked at it now, you will see that two of the graphs have changed. I want to discuss those changes here for those who want to think about proper data visualization.

Someone pointed out to me that it really wasn’t appropriate to make my life expectancy comparison graphs like this: [one of my original graphs on the post]

The reason I connected up the data points was to make it easier to make comparisons between consecutive points, but it is misleading because there is no connection, really, between the points. If I do a line graph of stock prices, for example, I have the horizontal axis as time and it makes sense to connect the dots. There really is some sort of connection.

But here the axis is categorical — there is nothing continuous, before-and-after, to connect the dots.

**COLUMN GRAPHS INSTEAD OF LINE GRAPHS**

So why not do a clustered column graph? There are a couple ways I could do it. There’s this, so that you can see how life expectancies increase over each subslice (male, female, sex-blended):

But that’s kind of difficult to “read” in terms of which column is which.

**COMBINATION GRAPHS**

Yes, I could do three separate column graphs, but let’s try something different — with a dashed line set at the level of the sex-blended public safety officers so we can more easily eyeball what’s higher and what’s lower:

Ah, I like that. I made this using the “combination graph” option in Excel. Excel has really made it easier for users to manipulate the elements in charts so that one can get the results one wants.

**CHOICE OF VERTICAL AXIS**

What about the vertical scale? It runs from 76 to 94, but the youngest of the ages is just scant of 82 and the highest scant of 92. Why not do that?

Using this “short” axis makes it much easier to make eyeball comparisons due to the compactness of the scale (yowza on the teachers, again). This distorts the importance of the differences, though. A difference of only one year in life expectancies looks huge here.

But why not use a longer axis, starting at age 25 (after all, this life expectancy is assuming the person is currently age 25):

That shows you that, even though you can see some differences, from a percentage standpoint the differences are large only in a few cases — essentially, the differences in life expectancies for private pensions, public safety officers, and general government employees are small. The teachers are appreciably higher, and the general U.S. population appreciably lower.

**FINAL CHOICE**

That said, I do want people to see the differences, but not to exaggerate the differences. I have no issue with keeping the 76 to 94 axis, and I used it also for the life expectancy from age 65 graph.

So yes, I replaced the graphs in the original post, and you can play with graph choices by downloading my spreadsheet, and having fun with it yourself.

Related Posts

Mortality Monday: How Young is "So Young to Die"?

Geography of Mortality: State Ranking by Increase in Total Mortality and COVID Mortality, 2020-2021, Provisional

Merry Christmas! Have Some Finalized 2021 U.S. Mortality Stats!