Back

Here’s the sentence:

❝ While TMTG has not disclosed the size of its user base, the research firm Similarweb estimated that in March it had 7.7m visits – while X, formerly Twitter, had 6.1bn. ❞

At a quick glance, if you’re not paying close attention, it almost sounds like TMTG and X have about the same amount of visits. Did it trip you up for a moment? In fact, X has 792 times more.

2/

Here is the data journalism rule I wish, wish, wish news sources would follow:

️ When comparing two quantities, they must be compared in like terms: ️
1. Same units
2. Same multiplier (thousand / million / billion)
3. Same denominator

3/

The sentence above fails the second criterion. It should read:

❝ While TMTG has not disclosed the size of its user base, the research firm Similarweb estimated that in March it had 7.7m visits – while X, formerly Twitter, had 6,100m. ❞

4/

Same units (this on a hypothetical example):

“The old truck weighs over 3 tons, whereas the new delivery cart weighs a mere 700 pounds.”

“The old truck weighs 6000 pounds, whereas the new delivery cart weighs a mere 700.”

5/

Same denominator is a subtle one (also a hypothetical example):

“The House bill would cost $3.2 trillion over the next decade, whereas the Senate bill would cost $1.5 trillion in this budget cycle, to be adjusted after.”

Getting better:

“The House bill would cost $320 billion / year over the next decade, whereas the Senate bill would cost $750 billion / year over the next two years.”

But wait…

6/

…do they •end• after 10 years vs 2 years? Or are we estimating an ongoing cost out to different time ranges?

“The House bill would cost $320 billion / year, with the price fixed over the next decade. However, the Senate bill would cost $750 billion / year for the next two years, at which point Congress would negotiate a new cost.”

That’s better reporting (of this completely hypothetical thing).

7/

Edward Tufte rightly says the fundamental act of data analysis is to make comparisons. Comparisons in unlike terms are hard to make.

This seems like a rule that news editors could readily learn and apply. Whose door can I beat down with my obviously Brilliant and Very Important idea? Anybody have an inside track with the AP Style Guide?

(Will add examples to this thread as I find them.)

8/

@donw
Related pet peeve: reporting differences when ratios are appropriate. If a stock is up 10 points, does that mean it doubled? Or is that a slight fluctuation? Depends on the stock!

@inthehands Yeah that's similarly maddening. It's not that hard to add “, an x% increase” unless you don't know how to do the division.

MNSHO has long been that news is failing us on the last of the WWWWW: the why. They have a rabid obsession with never doing even simple analysis in favor of just quoting people, but it seems like basic math they can do without leaving their View From Nowhere.

@donw @jwz
It’s “(Year-end projection)” that really clinches it

@argonaut
A classic. And yeah, at a certain point no amount of writing can compensate for innumeracy, but journalists can at least not throw obstacles in the path of readers trying to make quantitative judgments.