Different types of graphs, with different scales, can portray the same set of data in many different ways. These three images show how one can take the same information but use it in different ways in order to convince people there has been significant change, or almost no change at all.
Source: Dougherty, Jack, Jesse Wanzer, and Christina Ramsay. “Sheff V. O’Neill: Weak Desegregation Remedies and Strong Disincentives in Connecticut, 1996-2008.” Papers and Publications (January 1, 2009). http://digitalrepository.trincoll.edu/cssp_papers/3.
The graphs that I made below both use the above bar graph as their data, but appear convey very different messages. One shows very minor changes, which would be used if one wanted to show how little progress has been made. The second one has a very steep curve, used if one wanted to display an extreme change in percentage of minority students in reduced isolation settings.
See how this graph has a scale on the Y axis that goes up to 100. This leads to a very horizontal line–one that looks as if almost no change has happend. Using a bigger scale is one of the ways people can “lie with statistics” in order to prove the point that they want.
This graph on the other hand starts at 8 instead of zero, and ends at just 18 on the Y axis. This leads to a very vertical line, so it looks like there has been a very high increase in percentage of students in reduced isolation settings. One would use a small scale on the Y axis if they want to falsely prove a point that there has been a lot of change.
I was surprised how drastic the difference in graphs was just by changing the axis. So, in conclusion, when looking at graphs, one should always check the numbers on the axis and think about the scale the author is using before coming to conclusions about graphs.