There are times when it is reasonable to show a truncated y-axis. This is not one of them; at least, not if you plan to use the word “rapidly” to describe a 4 pp. change. - ThreadSky

benstanley.eu • 70 days ago

There are times when it is reasonable to show a truncated y-axis. This is not one of them; at least, not if you plan to use the word “rapidly” to describe a 4 pp. change.

Reposted from Jolyon Maugham

Younger people are rapidly losing faith in democracy.

Comments

glenagalt.bsky.social•70 days ago

In the words of my old maths teacher "Somebody stole the Orange*"

*Orange= his preferred term for "origin" when he wanted to wake us up with a laugh if we were looking bored.

johnspringford.bsky.social•70 days ago

Also, this is a vague question - democracy can easily mean "the outcomes of my country's political process" here

rowyourbot.bsky.social•69 days ago

Or it could mean—democracy isn’t able to protect itself from antidemocratic forces. There are so many interpretations of this!

bimblinghill.bsky.social•70 days ago

Yes. A rare graph crime from the FT they're normally very good.

redkeith.bsky.social•69 days ago

You have created a very erudite thread, Ben, most of which is sailing over my head.
But you've got a great bio, made me laugh.
Clever, too.

nibblerrat.bsky.social•70 days ago

It's a very odd graph in many ways.

teaanddata.bsky.social•70 days ago

Similarly, if showing multi graphs of the same thing across different countries, use the the same scale.

sajuria.com•70 days ago

And let's not get into the debate on whether satisfaction with democracy is a valid measure of faith (?) in democracy

benstanley.eu•70 days ago

I think there are often cases when using a truncated y-axis is better than using the full range. In this case - leaving aside concerns about whether “satisfaction with” means “faith in” - a more limited range is useful to compare the extent of change between groups in a variable whose range

benstanley.eu•70 days ago

is unlikely to come anywhere near zero. Yet you have to be careful with language. Suggesting something is changing “rapidly” on a plot with a truncated range can serve to exaggerate greatly what is going on.

benstanley.eu•70 days ago

I’d also like to see some confidence intervals here, but that’s another category of gripe.

rikkidean.bsky.social•70 days ago

It might also be nice to know where these people are from too. I presume UK. Weird just to omit this information, as if all cohorts everywhere will have the same trend.

rafalhalik.bsky.social•70 days ago

Erich Fromm and his Escape from Freedom
We can observe decrasing trend of civil society activity People realized that they preffer their predefined buble and do not want cross bondries.Activity in public space is useles. Matters only me and now and nothing else.Democcracy is useles in such context

kamilaceran.bsky.social•70 days ago

I don't quite get this graph. Is it change of their opinion in time, while they age? Then why GenX age ends at 50, as they can be almost 60 now? Or is it their today's opinion, but then why the generations overlap so much?

rikkidean.bsky.social•70 days ago

I was also totally confused by this, but guess it is the former. Also not clear whether it is longitudinal data actually showing change for the same sample or aggregation of time series for different samples. It is just a really terrible graph all round.

jwvbsky.bsky.social•70 days ago

Agreed. 4pp is snail pace.

Comments

Posting Rules

Reply