Pet Peeve: Misleading Charts and Graphs

[Carson]

This has always been an annoyance of mine. This is usually more of an ethical dilemma when used in advertising, however weather.com's temperature graphs are rather annoying as well; mostly because their scale makes them almost useless.

Care to guess the range of temperature before hitting the spoiler?

Spoiler

 

[SpeedSolve]

Oh jeez. :fp

 

[aronpm]

To be completely fair, for the scale to be totally accurate the minimum value would need to be -460F, and that would basically be more useless.

Here's yr.no's (as far as I am concerned, the best weather website) graph for Adelaide, and I think it's fine (as long as you know how to read it properly)

Spoiler

 

[Carrot]

Aron, I don't think you get it...

It's got a weird downslope before each rise in temperature, like they plotted a few temperatures and made excel make a tend graph.... why would the temperature fall so much so fast ? (Or am I stupid? )

 

[peterbone]

Odder, I think based on the fact that he removed the temperature scale he must mean that it doesn't start from zero (although I agree that the use of curve fitting here is misleading). I don't see a problem with not starting at zero at all. It allows you to see the variation in temperature much more clearly. Where you start the scale from is completely arbitrary. As long as you always put the scale on the axes then there should be no confusion. Zero Farenheit doesn't have any significance anyway. It's the use of farenheit that annoys me most.

 

[aronpm]

Aron, I don't think you get it...

It's got a weird downslope before each rise in temperature, like they plotted a few temperatures and made excel make a tend graph.... why would the temperature fall so much so fast ? (Or am I stupid? )

No, I get it perfectly well. The "problem" is that the y-axis doesn't start at zero which can be a problem if you're comparing something like 85% of people in Australia wear seatbelts and 82% of people in the United States wear seatbelts, because the graph's scale can make it look like twice as many people wear seat belts in Australia. (I made those numbers up obviously)

Since these weather graphs are just showing a trend, I don't think it's a problem, but that might just because unlike the general population I can math. However, I think the weather.com graph should have a larger range, like the yr.no graph.

 

[Achifaifa]

As long as you see the temperatures the axis, ¿Where is the problem? I also can't see your point with removing the reference. If you remove the numbers from a graph, the graph is just useless, so I fail to see what exactly annoys you in that useless graph.

 

[Kirjava]

Yeah I hate it when I remove information from a graph and make it hard for myself to understand.

 

[Godmil]

for the scale to be totally accurate the minimum value would need to be -460F,

No it wouldn't. That's a slippery slope argument. (yeah, I just wanted to use that pun).
I'd have it with a range that is reasonable for the temperature to be - so it would also give you a general idea of it being warm or cold.

 

[cmhardw]

I had a good math teacher in 7th grade that did a lesson on this very topic. I remember her explaining how most graphs you will see in the real world (advertising, etc.) are misleading. She had an interesting take on this. She gave us data comparing things, and asked us to make an honest graph as well as a dishonest graph to represent the data. I still remember that lesson to this day, and yes I think it sucks that things like this happen in the real world. Always make sure to read axes labels very carefully :\

 

[Dene]

This is a classic one which showed up several times throughout my schooling career

 

[Goosly]

Dene's one is indeed a very good example of how graphs can be misleading when not referred to zero
As for the temperature one, it would be better is it was referred to zero, but I think it's not necessary

 

[Carson]

To be completely fair, for the scale to be totally accurate the minimum value would need to be -460F, and that would basically be more useless.

Here's yr.no's (as far as I am concerned, the best weather website) graph for Adelaide, and I think it's fine (as long as you know how to read it properly)

Spoiler

I understand that this is a temperature graph, and that starting at zero (or absolute zero) would be a little extreme. I think the graph you linked is an excellent example of how to do things right! A slightly larger range is all that is really needed to make this graph useful.

Aron, I don't think you get it...

It's got a weird downslope before each rise in temperature, like they plotted a few temperatures and made excel make a tend graph.... why would the temperature fall so much so fast ? (Or am I stupid? )

This wasn't actually what I was getting at, but yet, there smoothing algorithm seems to be a little off.

As long as you see the temperatures the axis, ¿Where is the problem? I also can't see your point with removing the reference. If you remove the numbers from a graph, the graph is just useless, so I fail to see what exactly annoys you in that useless graph.

You shouldn't have to analyze every single number on a graph to understand it. To understand this graph's data, you have to rely more on the numbers than you do the actual graph. All of these numbers appear above the graph in the daily temp chart, which makes this graph pretty much useless. When the raw data can be understood more quickly than a graph, there is no reason to have a graph.


Just to make things clear, this isn't a big deal to me; I actually get a laugh out of the weather charts every time I see one. Graphs such as the one that Chris linked are the ones that really annoy me.

 

[andyfreeman]

Three degrees?

EDIT: Ooh, I'm right

I've seen worse............trust me. It's not that bad. But I do know what you mean.

 

[Achifaifa]

@Carson: I see where are you going with this, and you might be right about it. But if the shown data is correct, and the graphs are correctly labeled, there should be no problem at all. I tried to watch the graph you showed us again, and the only thing I see about the temperatures in the numberless one is that they are going to rise. To know how much, I go to the axis and see the values.

The graph is OK, if someone thinks the roads are going to melt because he just saw the line in the graph is not a graph failure, but a user failure to interpret the given information.

About Dene's ad... That might be worse because they want profits misleading the reader, but they give all the information in the ad, so it's not wrong at all.