The survey was conducted from December 14–December 31, 2022, for Sample 1 and November 29–December 13, 2022, for Sample 2. Participants were recruited via an online crowd-sourcing platform (CrowdWorks) with 4.8 million enrollments. Participants were selected on a first-come, first-served basis. Informed consent was obtained from all participants. This study used two samples (Samples 1 and 2). The exclusion criteria of both samples were (a) a self-reported current history of a sleep disorder and (b) failing the attention check while answering the questionnaire [23]. The demographic characteristics of the two samples are shown in Table 1. Sample 1 was used for exploratory factor analysis, internal consistency, and to confirm the test–retest reliability of the BPS. Among those who responded to the initial survey (n = 252; Time 1), 189 participants completed a follow-up survey (Time 2) within approximately 14 days (Mean = 13.9 ± 0.5 days). No participants in Sample 1 met the exclusion criteria, and all 252 participants were included. The COSMIN checklist was used to determine the sample size for the study [24]. The COSMIN checklist establishes the number of people needed to validate health-related measures; for factor structure validity, internal consistency, and test–retest reliability, a minimum of 100 participants is desirable [24]. Therefore, the sample sizes for Samples 1 and 2 were set to at least 100 participants, considering the expected number of missing data.

For Sample 2, participants from five age groups were recruited. The age groups were 18–20 years (n = 97), 21–30 years (n = 134), 31–40 years (n = 138), 41–50 years (n = 139), and ≥ 51 years (n = 132). The breakdown of employment status by age group of Sample 2 can be found in Table 2. Sample 2 was used for confirmatory factor analysis, criterion-related validity, construct validity, and for determining the internal consistency of the BPS. Additionally, we explored the relationship between BP and demographic variables in Sample 2. Among those who initially participated in the Sample 2 survey, 13 participants who failed the attention check were removed, leaving 640 participants for the analysis.

The BPS used in this study [7] is a 5-point scale (1 [never] to 5 [always]) with nine items. The BPS was translated from the original English into Japanese after obtaining permission from the developer. Two native Japanese-speaking researchers in clinical psychology independently translated the original English version into Japanese. Subsequently, the translated version was back-translated from Japanese to English by an additional independent bilingual translator. No marked differences were observed between the back-translated and original versions. Finally, the BPS developer assessed the original and back-translated versions for uniformity, and recommended revisions were made. Finally, the developer rechecked the revised version and confirmed the absence of semantic differences between the original and back-translated versions. The final version of the scale can be found in the Supplementary Material (Online Resource 1).

In addition to the BPS, an anchor item was utilized to examine the test–retest reliability of the Japanese version of the BPS. AtTime 2, participants in Sample 1 responded to the question, “Has your frequency of BP increased in the past two weeks since the last survey?” on an 11-point scale (-5: decreased considerably; 0: unchanged; 5: increased considerably). To test the criterion-related validity of the BPS, scores between the group that answered “yes” and the group that answered “no” to the item “Do you have trouble with bedtime procrastination?” were compared.

The pure procrastination scale [25] was used to confirm the validity of the BPS. We used the Japanese version of this scale [26]; a 5-point scale ranging from 1 (not at all) to 5 (always), with 12 items to assess general procrastination. Cronbach’s α and ω coefficients were 0.94 and 0.95, respectively.

To confirm the validity of the BPS, we used the short version of the Self-Control Scale [27]. We used the Japanese version of the Self-Control Scale [28]; a 5-point scale ranging from 1 (not at all) to 5 (very much), with 13 items to assess self-control. Cronbach’s α and ω coefficients were 0.87 and 0.89, respectively.

The Athens Insomnia Scale [29] was used to evaluate sleep quality at least three times a week during the past month. This study used the Japanese version of this scale [30]; a 4-point scale requiring responses for eight items. Cronbach’s α and ω coefficients were 0.86 and 0.90, respectively.

The questionnaire includes questions on the time of sleep onset, time of awakening, latency to fall asleep, and actual sleep duration. The questionnaire items were developed from the Japanese version [31] of the Pittsburgh Sleep Quality Index [32], with eight items for weekdays and holidays (sleep before holidays). We calculated sleep efficiency using these items; dividing actual sleep time by the difference between the time of falling asleep and that of waking. Among the participants in Sample 2, 90 whose sleep efficiency was not between 0 and 1 were excluded from the analysis of sleep efficiency.

Participants in Sample 1 were asked to complete the BPS and answer questions about the frequency of BP. Those in Sample 2 were asked to complete the BPS, Pure Procrastination Scale, Self-Control Scale, and Athens Insomnia Scale. Sleep outcome and schedule data were also provided. The study protocol was approved by the Research Ethics Review Committee of the Graduate School of Education of Tohoku University (21-1-049). The study protocol was preregistered with the Open Science Framework, where all study data are available [33].

Analyses regarding the reliability and validity of the BPS were preregistered and performed accordingly. The details of the preregistered analyses and hypotheses are shown in Table 3. To assess internal consistency, we calculated the α and ω coefficient of the BPS. To assess factorial validity, we conducted an exploratory factor analysis in Sample 1 and a confirmatory factor analysis in Sample 2. Before conducting factor analyses, in addition to the preregistered analysis, we conducted the Kaiser–Meyer–Olkin (KMO) test to determine how suited the data were for factor analysis. For construct validity, we assessed the correlations between BPS and general procrastination, self-control, sleep quantity, and sleep quality. To assess test–retest reliability, we used an anchor item and ICC. To examine criterion-related validity, we compared the BPS scores between individuals that answered “yes” and those that answered “no” to the item “Do you have trouble with bedtime procrastination?” using a Student’s t test.

To investigate the relationship of BP with demographic variables and insomnia symptoms, the correlation coefficients were calculated in Sample 2. Additionally, BPS scores were compared between age groups, genders, and employment statuses, with analysis of varianceFootnote 1 and follow-up analysis with Tukey-HSD correction. Further, to assess the level of BPS scores that would indicate maladaptive BP; the cutoff BPS score was determined via ROC analysis using the item “Do you have trouble with bedtime procrastination?” The R 4.2.2 statistical software was used for statistical analysis of the results.