[../../banner7.htm]
 


Boxplot

A boxplot is a graphical way of summarizing the distribution of the scores of a group of patients.


Boxplots (often called "box and whisker" plots) are a way of summarizing a distribution of scores of a group of patients. By portraying the scores for more than one group next to each other, one can frequently see the major trends in the dataset.

The "box" in a boxplot shows the median score as a line and the first (25th percentile) and third quartile (75th percentile) of the score distribution as the lower and upper parts of the box.

The median is the score at the 50% percentile: half of all patients get a score higher than the median, and 50% get a score lower. It is the middle point in the distribution of scores.

The 25th percentile is the point at which 25% of the patients score lower (and 75% score higher). The 75th percentile is the point at which 75% of the patients score lower (and 25% score higher). Thus, the area in the "box" represents the middle 50% of the patients.

The "whiskers" shown above and below the boxes technically represent the largest and smallest observed scores that are less than 1.5 box lengths from the end of the box. In practice, these scores are about the lowest and highest values one is likely to observe.

On very rare occasions, scores are shown as open circles "o" or stars. These scores are ones that are, respectively "very rare" and "exceedingly rare." Such scores may represent very unusual patients or data processing errors.

In comparing the boxplots across groups, a simple summary is to say that the "box" area for one group is higher or lower than that for another group. This comparison is analogous to saying that one group tends to have higher scores than another. To the extent that the boxes do not overlap, the groups are quite different from one another.

Formal statistical tests, of course, are used to compare the groups to one another. As a general rule-of-thumb for visual displays, if the degree of overlap is less than 80% for moderately-sized samples, the differences will usually prove to be statistically significant when a formal test is conducted.


TheMeasurementGroup.com Glossary Index

 

 

 

 


© Copyright 1999-2005 by The Measurement Group LLC. All rights reserved.