African populations, 1987
The family of power transformations is a flexible way to remove skewness of a data set. Drag the vertical red line on the axis to apply a power transformation. Explain that this expands one end of the axis and 'squashes up' the values at the other end. Select Raw values for a much smoother (and easier to understand) axis.
The transformation used is actually a Box-Cox transformation (but you are probably best not to say much about this to the students). A log transformation therefore corresponds to a power of zero.
The left and right arrow keys can be used to fine-tune the power (after you have first adjusted the power by dragging).
Although any power is valid, it is usually easier to interpret the analysis if 'neat' powers are used, even if a nearby transformation would result in a slightly more symmetric distribution. For example, a log transformation would probably be used for this data set.
This data set contains the populations (in millions) of all African countries in 1987.
Note that Nigeria seems an outlier in an ordinary plot of populations but appears less unusual on a log scale so it is just the most extreme value in a very skew distribution.