Tip:
Highlight text to annotate it
X
(male narrator) So...it's really easy on a computer to do
all kinds of cool graphics, uh...graphical displays,
but that doesn't always mean that you should.
So we're gonna take a minute and talk
about some bad graphical representations of data.
So here is the car color data we were looking at earlier,
and-and here, it's done in a 3D bar graph.
Uh...the problem with this graph is that,
because of the perspective of it,
it's almost impossible to read now.
Uh...I mean, it's really hard to tell
which of these bars is actually larger
when it is on this slant like this,
and so this is really a bad graphical representation.
Now it's really hard to read.
Now here's another, uh...situation.
So suppose a labor union produces this graphic,
uh...designed to show the difference
between the average manager's salary
and the average worker's salary.
Now it could be...I mean, if I was to look at this,
I'd say, okay, if I was to imagine this
as, uh...you know, the worker's salary,
then, let's see, that, ooh...ooh, wow,
that...this one's like four of those.
And so if I was to look at this,
I might think that, wow, those manager's salaries
are four times bigger than the worker's salaries!
But...it's sometimes hard to tell,
'cause...'cause not everyone, you know, does things that way.
It could be that the manger's salaries
are only twice as big as the worker's salaries,
and the, uh...you know, the person creating
this representation said, oh, look, it's only twice as big.
Right?
And so that-that visual distortion
of area versus a length,
uh...really makes this hard to tell,
based on the graphic, uh...what's really going on.
So this...so this type of...of, uh...pictogram
is not a good representation.
Now there are some representations
where, say, you would use, like, some kind of picture,
uh...like a dollar sign to represent money,
and then you use a certain number of them
to represent relative sizes.
And that's a lot more acceptable,
because then it really is the length of the picture
that corresponds with the quantities.
Let's look at one more.
And this is a big problem with bar graphs.
Notice these two graphs here.
Um...these are two graphs from a poll
about same-sex marriage rights taken in December 2008.
Uh...and...they look quite different, right?
Which...which of these is misleading?
The one that's misleading, uh...is this one.
This is a bad graph.
It's a bad graph,
because this vertical axis is condensed.
Uh...it's condensed from 40 to 60,
and so it creates a misleadingly large disparity, right?
This bar now looks three times bigger
than-than-than this bar.
Whereas, if we actually look at it measured from 0,
we can see that the difference is much smaller,
uh...than-than this graphic would-would lead us to believe.