Using the 10’s digit for the stem and the units digit for the leaf made good sense with this data set;
other choices make sense depending on the type of data. For example, suppose we had a set of gas
mileage tests on a particular car (e.g., 28.3, 27.5, 28.1, ...). In this case, it might make sense to make the
stems the integer part of the number and the leaf the decimal part. As another example, consider
measurements on a microscopic computer part (0.0018, 0.0023, 0.0021, ...). Here you’d probably want
to ignore the 0.00 (since that doesn’t help distinguish between the values) and use the first nonzero digit
as the stem and the second nonzero digit as the leaf.
Some data lend themselves to breaking the stem into two or more parts. For these data, the stem “4”
could be shown with leaves broken up 0–4 and 5– 9. Done this way, the stemplot for the scores data
would look like this (there is a single “1” because there are no leaves with the values 0–4 for a stem of 1;
similarly, there is only one “5” since there are no values in the 55–59 range.):
The visual image is of data that are slightly skewed to the right (that is, toward the higher scores). We
do notice a cluster of scores in the high 20s that was not obvious when we used an increment of 10 rather
than 5. There is no hard-and-fast rule about how to break up the stems—it’s easy to try different
arrangements on most computer packages.
Sometimes plotting more than one stemplot, side-by-side or back-to-back, can provide us with
comparative information. The following stemplot shows the results of two quizzes given for this class
(one of them the one discussed above):