Any graph that doesn't start at zero is intended to incite emotion rather than reason.
That's a bit of a stretch.
Graphs are generally bad at presenting anything. This graph is zoomed on the Y axis to show the meaningful difference in the data, which would otherwise be imperceptible to the human eye.
That being said... it's also a 3D column chart, which is always wrong to use to present anything for a lot of reasons not worth going into, but related to the human eye's inability to project lines into 3D space accurately.
Intentional or not, such a graph will always create emotion outside of the scope of the data being presented (see exception below). That isn't really controversial. What is potentially controversial is the intention. I assert the usage of such axis shifting is intentional more often than not.
Graphs are generally bad at presenting anything.
That is completely untrue. Graphs are excellent at presenting numerical data of two variables. All numerical datasets can be broken down into two variables, though multivariate datasets can lose some context when they are broken down in that manner.
This graph is zoomed on the Y axis to show the meaningful difference in the data, which would otherwise be imperceptible to the human eye.
Here you have shown evidence to support my first assertion. This is intention to increase a reaction. If there is no meaningful difference in the data, then intentionally showing a difference in the data shows a purpose to increase the reaction of the audience to be what you want it to be. If the raw data would not, under any honest exposition create such a reaction, then showing the data in this manner creates a dishonest reaction.
Now, there are exceptions to that; such as in the case where you are looking for a molehill in a mountain (noise). For example, looking for the Higgs Boson. It requires digging through massive amounts of data to get a signal, thus showing the signal requires reducing the noise. Though in that case you can reasonable set the baseline to zero wherever you want, since the energy signatures are relative to begin with. This arbitrary baseline is not the important part of this exception though, it really extends to anything that has noise.
However, I assert in any case where there is no need to reduce noise (reduce the impact of external, not relevant data), the creation of a graph with an axis shift is fuckery, most often intentional (even if someone doesn't realize it is fuckery because they have been trained to do so).
That's a bit of a stretch.
Graphs are generally bad at presenting anything. This graph is zoomed on the Y axis to show the meaningful difference in the data, which would otherwise be imperceptible to the human eye.
That being said... it's also a 3D column chart, which is always wrong to use to present anything for a lot of reasons not worth going into, but related to the human eye's inability to project lines into 3D space accurately.
Intentional or not, such a graph will always create emotion outside of the scope of the data being presented (see exception below). That isn't really controversial. What is potentially controversial is the intention. I assert the usage of such axis shifting is intentional more often than not.
That is completely untrue. Graphs are excellent at presenting numerical data of two variables. All numerical datasets can be broken down into two variables, though multivariate datasets can lose some context when they are broken down in that manner.
Here you have shown evidence to support my first assertion. This is intention to increase a reaction. If there is no meaningful difference in the data, then intentionally showing a difference in the data shows a purpose to increase the reaction of the audience to be what you want it to be. If the raw data would not, under any honest exposition create such a reaction, then showing the data in this manner creates a dishonest reaction.
Now, there are exceptions to that; such as in the case where you are looking for a molehill in a mountain (noise). For example, looking for the Higgs Boson. It requires digging through massive amounts of data to get a signal, thus showing the signal requires reducing the noise. Though in that case you can reasonable set the baseline to zero wherever you want, since the energy signatures are relative to begin with. This arbitrary baseline is not the important part of this exception though, it really extends to anything that has noise.
However, I assert in any case where there is no need to reduce noise (reduce the impact of external, not relevant data), the creation of a graph with an axis shift is fuckery, most often intentional (even if someone doesn't realize it is fuckery because they have been trained to do so).
I agree with you about showing 2D data in 3D.