« The missing earmark caveat | Main | Two-candidate dynamics in the GOP race »

January 29, 2008

Comments

It's an inevitable consequence of the use of any average that it provides only a single number and will therefore be misleading to some extent. Brendan characterizes the $1800 average over all quintiles cited by Bush as misleading but cites without disapproval the average of $814 for the middle quintile. But that's an average too! And therefore misleading. For some in that middle quintile, the amount will be less, for others more.

Should politicians and policy makers avoid the use of any averages? Should every speech describe not an overall average but five averages by quintiles (no wait, that's still too coarse and misleading), or ten averages by deciles (still somewhat misleading). And which way shall we cut the country into quintiles and deciles? By income? By wealth? Should we subdivide the overall average by other ways like gender or race or region?

Sure, politicians could spend ten minutes reciting tables of figures, but that simply isn't feasible. They could ignore the use of any statistics whatsover, but that makes the public discourse less fact-based, not more. They could use only anecdotal examples (this poor child couldn't get insurance coverage), but that's even more misleading than the use of averages. Or they could use statistics, and the people could understand that those statistics are by necessity simplifications whose implications need to be considered by those who pore over tables of numbers.

In the meantime I look forward to Brendan being as outraged by the use of averages when they come from Democratic politicians, from global warming alarmists, from people conmplaining about income disparities among races or genders. When Brendan takes on all these frequent citers of averages, we'll know his beef is with the inherent coarseness of the use of an average as a statistical measure, rather than simply another arrow in his quill to use against the Administration.

As it is, one could say that Brendan's average outrage over the use of averages is low, but rises to high levels only when averages are used by Republicsns to justify their preferred tax policy.

Per Rob's comment, the suggestion that I want elaborate presentations of statistical details is absurd; I'm objecting to unrepresentative statistics. Politicians could easily use medians instead of means to discuss the benefits of tax cuts, but they won't because the median benefit (which is what most people think of when they hear "average" in this context) is typically much less than the mean. The statistics don't have to be complicated; they just have to not be misleading.

That said, I welcome examples of misleading averages in any policy context. This is a non-partisan issue.

...from global warming alarmists...

Always with global warming these days. Who are these global warming alarmists and where are their abuses of statistics?

We can start with Wikipedia: "The global average air temperature near the Earth's surface rose 0.74 ± 0.18 °C (1.33 ± 0.32 °F) during the 100 year period ending in 2005." See that? Average! Do we really have to go on? The global warming literature is replete with average increases in temperature.

My complaint isn't necessarily with the use of averages, it's with the selective outrage over the use of averages. Sauce for the goose . . . .

Politicians are going to use statistics to emphasize their side of an argument.

Journalists (and other politicians) should point out the short-comings of the use of those statistics, when it occurs.

Pointing out the actual facts doesn't indicate a sense of "outrage", nor does it represent "an attack".

The comments to this entry are closed.