Chapter 6 Longitudinal Data

Although the term longitudinal naturally suggests that data are collected over time, the models and methods we will discuss broadly apply to any kind of repeated measurement data. That is, although repeated measurements most often take place over time, this is not the only way that measures may be taken repeatedly on the same unit. For example,

• The units may be human subjects. For each subject, reduction in diastolic blood pressure is measured on several occasions, each involving the administration of a different dose of anti-hypertensive medication. Thus, the subject is measured repeatedly with varying doses.

• The units may be trees in a forest. For each tree, measurements of the tree’s diameter are made at several different points along the tree’s trunk. Thus, the tree is measured repeatedly over positions along the trunk.

• The units may be pregnant female rats. Each rat gives birth to a litter of pups, and the birth weight of each pup is recorded. Thus, the rat is measured repeatedly over each of her pups. The third example is slightly different from the other two in that there is no natural order to the repeated measurements.

Thus, the methods will apply more broadly than the strict definition of the term longitudinal data indicates – the term will mean, to us, data in the form of repeated measurements that may well be over time but may also be over some other set of conditions. Because time is most often the measurement condition, however, many of our examples will involve repeated measurement over time. We will use the term outcome to denote the measurement of interest. Because units are often human or animal subjects, we use the terms unit, individual, and subject interchangeably.