12
intercorrelated, whereas in the case of an index, items that intercorrelate to the small-
est extent possible are usually favored in the interest of parsimony. Similarities and
differences between these approaches to PRO measurement models are illustrated in
more detail in Fig. 1.2.
Measurement Theory and Models
PROs typically focus on subjective and nonobservable or “latent” phenomena, such
as the amount of pain a patient experiences or the amount of difficulty faced by the
patient in participating in their normal social roles. Measurement theories provide a
general framework for linking observable variables, such as test or item scores, to
latent variables, such as pain and physical disability [ 47 ].
Classical Test Theory
Classical test theory (CTT) is a theory that can be applied to PRO (total) scores
revolving around the concepts: observed score, true score, and error score. Classical
test theory is concerned with the relations between these concepts in order to evalu-
ate to what extent the total score of a(n) (PRO) instrument is affected by the random
measurement error. It may also provide instrument developers with a framework
Fig. 1.2 Graphical representation of latent variables with effect indicators and latent variables
with causal indicators
M.A.H.O. Voshaar and M.A.F.J. van de Laar