Sometimes, starting the Y-axis at zero is the BEST way to lie with statistics

Excellent video. That is all.

Here, have a Chart With Upwards Trend and Yen Sign Emoji. :chart:

1 Like

Like I said–we have to agree to disagree. I’m just saying I wouldn’t present that as a graph to a decision maker. Some would misinterpret the ratios and some would call it out for what it is–poor graphical presentation of information.

There are no other graphs which put the trend in its proper perspective in that Vox article. Some folks need text, some need graphs, and some need tables. When you present this sort of information to decision makers, all of it needs to be presented in all ways.

From your snarky comments, it sounds like you’re trolling me, so I’m going to end this here. Best–OO

My favorite is the inappropriate semilog plot.

Another great one is the indisputable correlation between any two increasing trends:

9 Likes

Wouldn’t the logical thing to do to emphasize the drop in participation be, to plot the inverse, ie. the percentage of nonparticipants? With a y axis of 0?

I’m totally snarky, but that’s not driving trollies. You’ve got to admit that the incomplete percent axis bit is an interesting point, true?

You just happened to go a bit strong on something that I’ve had to battle with more than once myself.

I still think you’re missing a couple of key points, like why the appropriate bottom axis isn’t zero but at best the point at which the issue becomes catastrophic…I agree there are always better things that can be done, but I’m disagreeing that the zero axis version is even vaguely useful when compared to every other visualization I’ve identified.

That’s what you’re arguing for, and I see it as a purists argument that misses the point of delivering information well, which was also the point of the video, which is what you snarked on.

Totally agreed, though I’ve got to admit sometimes spurious correlations can be fun. :smile:

5 Likes

Relevant for Canadian values of global temperature, I guess. 0 degrees C and 2/3 of 270K (surface melting point) are important because that’s a nice basic offset for PCR in electrolyte and its natural variants. We should let National Review tell us about how it makes most of its subscriptions to space- and vacuum-hardened nanolife anyhow.

1 Like

I agree as well, that is a good one… :smile:

We all agree that correlation isn’t causation, it gets really tricky when correlation is influenced by a unidentified third thing and seemingly unrelated things actually are tied together in an unexpected way.

A really bad example but I’m making one up so whatever…forest fires don’t drive almond butter prices, but they are both affected by drought, which might drive a correlation in graphs of almond butter prices and forest fires. That graph is meaningless without the precipitation information though, which is the key factor.

Those mystery relations are fun to try and suss out, but they just as often don’t exist, so one can go mad trying to find relationships that don’t exist outside of chance.

Fun, all right, and this one highlights the fact that what’s missing is a hypothesis. In this case the graph supports a Nick Cage-drownings connection, meaning there’s a need for more research.

1 Like

I suppose they could be suicides? Too bad there’s not a data point on ‘last movie watched before drowning’.

This topic was automatically closed after 5 days. New replies are no longer allowed.