Graphs that don't start at 0

[OuterGazer]

I just wanted to add to what Ben mentioned on graphs not starting at 0. I agree that most of the time could be used to mislead an statistic. However, it is something that comes across very often and it is not always used in a misleading way. Cutting our graph origin is useful when we want to show small variations in data that are relevant to the context in which we are working.

For example, imagine we have a bank account that never goes under 1000€ and never above 1010€ throughout a whole year. If we just plotted the line with a graph starting at 0 the line would be almost horizontal and we wouldn’t be able to see the fluctuation clearly. In this case it would make much more sense to start the graph at 1000 in the origin.

Other than that, great video!

[garypettie]

You’re absolutely right. If things aren’t moving much and you need a better look at what’s going on, it can be useful to zoom in on your graphs.
There are also times where you may want to represent data using different tick spacing, like logarithmic scales or plotting additional data on a secondary axis.

I think Ben’s main point is that these technique are often used disingenuously to mislead and to make the data look better (or worse) than it really is, so always be on the look out for that kind of behaviour.

When you’re presenting your findings to others, be sure to consider the implications of manipulating you graphs and always label them accurately so that the reader is aware of the changes you made. In many cases, it can also be useful to show both the scaled and unscaled graphs for additional context.