Separating Reliability and Stability in Test-Retest Correlation

Stability of self-esteem across the life span.

“…Participants rated the item ("I see myself as someone who has high self-esteem") on a 5-point Likert scale ranging from 1 (disagree strongly) to 5 (agree strongly). Using longitudinal data, Robins, Hendin, et al (2001) estimated the reliability of the SISE to be .75 (Heise, 1969). Robins, Hendin, et al (2001) provided extensive validation data for the SISE.…”

Oral health-related quality of life of children in relation to dental appearance and educational transition

“…34 Future research could subject measures of OHRQoL to path analysis techniques, aimed at distinguishing true reliability from temporal instability. 35 It should also be recognised that the examination of before and after scores alone does not provide information about the number of children who experienced an improvement in their OHRQoL or answer questions related to the magnitude or meaningfulness of these changes. 31,36 If it is, however, to be accepted that the measure has reliably assessed a true change in children's OHRQoL, then the examination of the CPQ subscales allows exploration of other possible explanations for this apparent improvement in OHRQoL.…”

I Like It... I Like It Not: Evaluating User Ratings Noise in Recommender Systems

“…However, it is not sufficient to compute the correlation between two different trials of the same test. As Heise explains [7], the correlation is aggregating two effects: the instrument's reliability and the stability of the user's judgements. That is, if we measure how much a user likes an item at two different times (separated by a month, for instance) and find a different rating, this could be due to either the reliability of the measure and the user's response or to the fact that the user's opinion has changed during that period.…”