Table 1.

Four kinds of reliability (Cranford et al., 2006), with descriptions and design examples

	Description	Design Decision Question	Empirical Example
I. Between subjects at the same occasion of measurement	Cross-sectional, single occasion	How many days of data are required to construct a reliable measure of differences between people measured at the same occasion?	Finding: Education correlated negatively with diurnal cortisol slope (Cohen et al., 2006). Sampling: 5 samples per day, 1 day, 781 participants
II. Between subjects at different occasions of measurement	Cross-sectional, different occasions	How many days of data are required to construct a reliable measure of between people measured at different occasions?	Finding: Psychological and sleep variables measured once were unrelated to diurnal cortisol mean, slope, or AUC measured once in breast cancer patients at varied points post-diagnosis (Carlson et al., 2007) Sampling: 4 samples per day, 1 day, 33 participants
III. Between subjects across multiple occasions	Longitudinal or repeated measures, multiple occasions	How many days of data over how many time points are required construct a reliable measure of stable differences between people?	Finding: A personality subscale of the SCL-90 measured at one time point significantly correlated with mean basal cortisol levels averaged across 3–6 years (Lupien et al., 1996) Sampling: 24 samples per day, 1 day, 3–6 occasions, 19 participants
IV. Within subjects, change across occasions	Longitudinal or repeated measures, multiple occasions	How many days of data at each time point are required construct a reliable measure of differences between occasions within the same person?	Finding: Foster children receiving an intervention had steeper diurnal cortisol slopes over time; children in regular foster care had flatter slopes over time (Fisher et al., 2007). Sampling: 2 samples per day, 2 days, 12 occasions, 117 participants Note: A latent variable was used to isolate occasion from day variance

Note: See Kudielka and colleagues (2012) for design considerations regarding within-day research questions such as those associated with ecological momentary assessment