0:00:07.808,0:00:10.839
A toothpaste brand claims[br]their product will destroy more plaque

0:00:10.839,0:00:12.910
than any product ever made.

0:00:12.910,0:00:16.411
A politician tells you their plan[br]will create the most jobs.

0:00:16.411,0:00:18.951
We're so used to hearing these[br]kinds of exaggerations

0:00:18.951,0:00:20.850
in advertising and politics

0:00:20.850,0:00:23.131
that we might not even bat an eye.

0:00:23.131,0:00:26.111
But what about when the claim[br]is accompanied by a graph?

0:00:26.111,0:00:28.471
Afterall, a graph isn't an opinion.

0:00:28.471,0:00:32.611
It represents cold, hard numbers,[br]and who can argue with those?

0:00:32.611,0:00:36.403
Yet, as it turns out, there are plenty[br]of ways graphs can mislead

0:00:36.403,0:00:38.192
and outright manipulate.

0:00:38.192,0:00:40.745
Here are some things to look out for.

0:00:40.745,0:00:45.760
In this 1992 ad, Chevy claimed to make[br]the most reliable trucks in America

0:00:45.760,0:00:47.510
using this graph.

0:00:47.510,0:00:51.963
Not only does it show that 98% of all[br]Chevy trucks sold in the last ten years

0:00:51.963,0:00:53.592
are still on the road,

0:00:53.592,0:00:57.338
but it looks like they're twice[br]as dependable as Toyota trucks.

0:00:57.338,0:01:00.634
That is, until you take a closer look[br]at the numbers on the left

0:01:00.634,0:01:05.472
and see that the figure for Toyota[br]is about 96.5%.

0:01:05.472,0:01:09.313
The scale only goes between 95 and 100%.

0:01:09.313,0:01:12.963
If it went from 0 to 100,[br]it would look like this.

0:01:12.963,0:01:16.243
This is one of the most common[br]ways graphs misrepresent data,

0:01:16.243,0:01:18.333
by distorting the scale.

0:01:18.333,0:01:20.804
Zooming in on a small portion[br]of the y-axis

0:01:20.804,0:01:25.703
exaggerates a barely detectable difference[br]between the things being compared.

0:01:25.703,0:01:27.974
And it's especially misleading [br]with bar graphs

0:01:27.974,0:01:31.023
since we assume the difference[br]in the size of the bars

0:01:31.023,0:01:33.233
is proportional to the values.

0:01:33.233,0:01:36.125
But the scale can also be distorted[br]along the x-axis,

0:01:36.125,0:01:40.414
usually in line graphs[br]showing something changing over time.

0:01:40.414,0:01:44.747
This chart showing the rise[br]in American unemployment from 2008 to 2010

0:01:44.747,0:01:47.996
manipulates the x-axis in two ways.

0:01:47.996,0:01:50.395
First of all, the scale is inconsistent,

0:01:50.395,0:01:53.416
compressing the 15-month span[br]after March 2009

0:01:53.416,0:01:56.755
to look shorter than [br]the preceding six months.

0:01:56.755,0:02:00.106
Using more consistent data points[br]gives a different picture

0:02:00.106,0:02:03.705
with job losses tapering off[br]by the end of 2009.

0:02:03.705,0:02:06.675
And if you wonder why[br]they were increasing in the first place,

0:02:06.675,0:02:10.615
the timeline starts immediately after[br]the U.S.'s biggest financial collapse

0:02:10.615,0:02:12.626
since the Great Depression.

0:02:12.626,0:02:15.219
These techniques are known as[br]cherry picking.

0:02:15.219,0:02:18.869
A time range can be carefully chosen[br]to exclude the impact of a major event

0:02:18.869,0:02:20.648
right outside it.

0:02:20.648,0:02:24.762
And picking specific data points[br]can hide important changes in between.

0:02:24.762,0:02:27.356
Even when there's nothing wrong[br]with the graph itself,

0:02:27.356,0:02:30.937
leaving out relevant data can give[br]a misleading impression.

0:02:30.937,0:02:33.997
This chart of how many people watch[br]the Super Bowl each year

0:02:33.997,0:02:37.626
makes it look like the event's[br]popularity is exploding.

0:02:37.626,0:02:40.198
But it's not accounting [br]for population growth.

0:02:40.198,0:02:41.967
The ratings have actually held steady

0:02:41.967,0:02:45.109
because while the number [br]of football fans has increased,

0:02:45.109,0:02:47.959
their share of overall viewership has not.

0:02:47.959,0:02:49.888
Finally, a graph can't tell you much

0:02:49.888,0:02:53.318
if you don't know the full significance [br]of what's being presented.

0:02:53.318,0:02:56.457
Both of the following graphs[br]use the same ocean temperature data

0:02:56.457,0:02:59.719
from the National Centers [br]for Environmental Information.

0:02:59.719,0:03:02.490
So why do they seem to give[br]opposite impressions?

0:03:02.490,0:03:05.279
The first graph plots the average[br]annual ocean temperature

0:03:05.279,0:03:07.987
from 1880 to 2016,

0:03:07.987,0:03:10.149
making the change look insignificant.

0:03:10.149,0:03:12.878
But in fact, a rise of even[br]half a degree Celsius

0:03:12.878,0:03:15.799
can cause massive ecological disruption.

0:03:15.799,0:03:17.219
This is why the second graph,

0:03:17.219,0:03:19.858
which show the average temperature[br]variation each year,

0:03:19.858,0:03:22.390
is far more significant.

0:03:22.390,0:03:27.379
When they're used well, graphs can[br]help us intuitively grasp complex data.

0:03:27.379,0:03:31.180
But as visual software has enabled[br]more usage of graphs throughout all media,

0:03:31.180,0:03:35.900
it's also made them easier to use[br]in a careless or dishonest way.

0:03:35.900,0:03:39.560
So the next time you see a graph,[br]don't be swayed by the lines and curves.

0:03:39.560,0:03:40.882
Look at the labels,

0:03:40.882,0:03:42.130
the numbers,

0:03:42.130,0:03:43.048
the scale,

0:03:43.048,0:03:44.360
and the context,

0:03:44.360,0:03:46.780
and ask what story the picture[br]is trying to tell.