ABSTRACT
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
In many biostatistical applications, there is a need to consider temporal variation. The most common examples are often found in clinical or behavioral intervention trials where a state can be reached by a patient. The time at which the state is reached could be of primary interest. The end point could be a vital outcome (death) or could disease remission, cure, or cessation of a behavior. In all cases, the time of the event is the important random variable. This is the typical scenario where survival analysis is employed (see Chapter 17). On the other hand, in some clinical or intervention studies, the variation of response over time is monitored. For example, cholesterol concentrations in blood might be monitored under different treatments, and the effect of these treatments over time is to be examined. In an intervention trial for diet change, food intake might be repeatedly measured via self-report
T&F Cat #K23899 — K23899 C018 — page 315 — 2/17/2016 — 20:27
T&F Cat #K23899 — K23899 C018 — page 316 — 2/17/2016 — 20:27
of
questionnaire. In these cases, the time of measurement is usually fixed and the measurement itself is the random variable.
