Research Information System
Frontiers in Psychology, vol.17, pp.1-21, 2026 (SSCI)
Introduction: Bedtime procrastination is defined as deliberately delaying sleep
without any external conditions preventing sleep. One of the most frequently
used scales in this field is the Bedtime Procrastination Scale (BPS). The original
form of the scale consists of nine items rated on a 5-point Likert scale. The
BPS is a measurement tool that has been applied to many cultures, both in the
language in which it was developed and in adaptations to different languages.
This study aims to examine the reliability coefficients obtained from different
studies for the BPS using meta-analysis methods and to determine the average
effect size for the scale.
Method: For this purpose, studies were searched in the Scopus, Proquest, Web
of Science, ScienceDirect, EBSCO, and Google Scholar databases between 2014
and 2025 using the keyword “Bedtime Procrastination Scale,” and analyses were
performed on 128 reliability coefficients (127 for α and 11 studies for ω). The
Bonnet transformation was used to obtain the average reliability coefficient.
Results: Cronbach’s alpha (α) was estimated at 0.855 [95% CI (0.843, 0.865)], and
McDonald’s omega (ω) was estimated at 0.867 [95% CI (0.834, 0.894)]. There was
no publication and reporting bias found for either reliability coefficient analysis;
however, the magnitude of heterogeneity suggests that moderator analyses
are warranted to explain systematic variability across studies. The moderator
analysis found that the variables mean age, SD age, region, and sample group
were significant for the Cronbach alpha coefficient, while only the sample group
variable was significant for the McDonald’s omega coefficient.
Discussion: Overall, the findings indicate that the Bedtime Procrastination Scale
demonstrates high and acceptable reliability across studies for both Cronbach’s
alpha and McDonald’s omega. While age, region, and sample type emerged
as significant moderators (for Cronbach’s alpha), a substantial proportion of
heterogeneity remained unexplained, indicating that reliability variability cannot
be attributed to a single set of study characteristics. Although reliability was
generally adequate, the observed heterogeneity and wide prediction intervals
suggest that caution is warranted when the scale is used in high-stakes or
critical decision-making contexts. Moreover, recommendations were made for
both researchers and practitioners.