John Burn-Murdoch+ Your Authors @jburnmurdoch Stories, stats & scatterplots for @FinancialTimes | Daily updates of the coronavirus trajectory tracker | [email protected] | #dataviz May. 19, 2020 1 min read + Your Authors

Great to see research into log vs linear, but this study has a fairly major issue:

If your test is "which one elicits more accurate reading of absolute numbers", obv linear wins, coz whole point of log charts is to relegate [less useful] absolute nums & promote *rate of change*

The claim that linear performs better than log rests solely on 3 Qs about absolute nums. In Qs 1 and 2, someone who identified the period with biggest relative difference (what log scales are for) rather than biggest absolute difference (what linear is for) was marked as "wrong".

Indeed, as the Qs were worded — number of X — they *were* wrong.

But both charts refer to contexts where the correct answer to the "number of" question is misleading.

In both cases, biggest relative rates — where people "wrongly" looked on log charts — are what actually matter.

In my thread back in early March, I explained that the idea behind log y scale was:
• highlight relative rate of change
• de-emphasise absolute numbers
• facilitate comparisons between countries where absolute nums differ by orders of magnitude

I’d love to see a study that tests for success or failure on those metrics, and it may well show linear is still better! That would also be amazingly useful knowledge :-)

To be clear, I’m really glad this work exists! It is a very interesting paper, with loads of interesting and valid findings, and I’m criticising with the benefit of hindsight and not having done this research.

Bring on the next study :-)

Ultimately, an equally valid interpretation of the paper’s results is:

• When it comes to interpreting rates of change, people are more accurate with log scales than linear scales

• Trade-of is they struggle to read absolute numbers from the log scale

Useful findings!


You can follow @jburnmurdoch.



Bookmark

____
Tip: mention @threader_app on a Twitter thread with the keyword “compile” to get a link to it.

Enjoy Threader? Sign up.

Since you’re here...

... we’re asking visitors like you to make a contribution to support this independent project. In these uncertain times, access to information is vital. Threader gets 1,000,000+ visits a month and our iOS Twitter client was featured as an App of the Day by Apple. Your financial support will help two developers to keep working on this app. Everyone’s contribution, big or small, is so valuable. Support Threader by becoming premium or by donating on PayPal. Thank you.


Follow Threader