Interpreting the box plot

This section will show how to interpret a box plot through a few examples.

First, if you look at figure 26.84, you can see a box plot for an experiment with 5 groups and 27 samples.

Image box_plot_with_data_requiring_normalization_web
Figure 26.84: Box plot for an experiment with 5 groups and 27 samples.

None of the samples stand out as having distributions that are atypical: the boxes and whiskers ranges are about equally sized. The locations of the distributions however, differ some, and indicate that normalization may be required. Figure 26.85 shows a box plot for the same experiment after quantile normalization: the distributions have been brought into par.

Image box_plot_after_quantile_normalization_web
Figure 26.85: Box plot after quantile normalization.

In figure 26.86 a box plot for a two group experiment with 5 samples in each group is shown.

Image box_plot_with_bad_quality_array_web
Figure 26.86: Box plot for a two-group experiment with 5 samples.

The distribution of values in the second sample from the left is quite different from those of other samples, and could indicate that the sample should not be used.