Japan has one of the shortest average sleep durations among countries, and this study tests whether a familiar culprit—choosing to stay up later than intended—can be measured reliably in Japanese adults. The study starts with a clear problem: the average sleep duration of Japanese people is shorter than that of people from other countries, and bedtime procrastination is suspected to be one factor contributing to this issue.

Its goal is to develop and validate the Japanese version of the Bedtime Procrastination Scale, or BPS-J. The motivation matters because getting sufficient sleep at the appropriate time is important for maintaining physical and mental health and performance.

Chronic sleep deprivation or delay in the sleep phase increases the prevalence of diabetes, obesity, and cardiovascular diseases. Sleep loss also interferes with cognitive function and emotion regulation, resulting in difficulties in social adjustment. Bedtime procrastination means the behavioral tendency of staying up later at night than intended or needed to go to bed.

Several cross-sectional studies indicate a relationship between bedtime procrastination and sleep loss, with bedtime procrastinators tending to have insufficiently short sleep duration, especially on workdays. The behavior has also been suggested to share a core concept of low self-control and general procrastination.

Among Japanese people, bedtime procrastination is suggested to result in shorter sleep duration and being more prone to an evening chronotype than among non-Japanese individuals. The average sleep duration of Japanese people is much shorter than that of people from sixty-three other countries.

The evening chronotype is also more prevalent in Japanese young adults than in young adults from other countries. The original Bedtime Procrastination Scale is a self-reported outcome measurement consisting of nine items measuring subjective bedtime procrastination.

It is a major measurement tool translated and validated in several languages, including Dutch, Spanish, English, and Korean. But a Japanese version had not yet been developed. After obtaining permission from the developer of the original version, the BPS-J was developed according to the COSMIN reporting guideline in four steps.

Those steps were forward translation, back-translation, cognitive interviews using the questionnaire, and expert checking. The study recruited six hundred individuals aged between twenty and sixty-five years who were daytime workers, working four or more days a week, and had no history of sleep disorders other than ISS.

People whose waking times and social constraints on activities varied from day to day were excluded, in accordance with the definition of bedtime procrastination. The BPS measures the tendency of bedtime procrastination and consists of nine items, with each item evaluated on a five-point scale from one, almost never, to five, almost always.

Higher total scores indicate a greater tendency of bedtime procrastination, and the values for alternatives to inverted items were reversed before calculation. Figure one traces participant selection across two assessment points. Of the six hundred people initially collected, twenty-six were excluded, leaving five hundred seventy-four analyzed at Time one; exclusions included night-shift work, fewer than four working days, and inconsistent answers.

A retest produced two hundred eighty participants at Time two, after two hundred eighty-nine did not complete follow-up and five were excluded. This flow clarifies the sample sizes supporting the validity, reliability, and measurement-error analyses.

For structural validity, confirmatory factor analysis for the one-factor model was performed after the Kaiser–Meyer–Olkin test and Bartlett’s test of Sphericity. The analysis used the Tucker-Lewis index, comparative fit index, and root mean square error of approximation as goodness-of-fit measures.

When the model did not meet the criteria, items with low factor loadings below zero point fifty were eliminated and the model was analyzed again. Out of six hundred participants, data from five hundred seventy-four participants were analyzed to estimate structural and criterion validities, relations to sleep-related variables, internal consistency, and standard error measures.

Among them, data from two hundred eighty participants were retested. The Kaiser–Meyer–Olkin test was zero point eighty-six, and Bartlett’s test was significant, so confirmatory factor analysis was interpreted as appropriate. But the one-factor model had poor fit indices: chi-square divided by degrees of freedom was fifteen point fifty-five, the Tucker-Lewis index was zero point seventy-seven, the comparative fit index was zero point eighty-three, and the root mean square error of approximation was zero point sixteen.

Figure two is a scree plot from the exploratory factor analysis, showing eigenvalues across factor numbers one through nine. The sharp drop after the first factor, followed by a flatter sequence, supports the authors’ conclusion that a two-factor model was appropriate.

This result provides the basis for the subsequent confirmatory factor analyses of the BPS-J structure. Table three reports confirmatory factor analyses for the Japanese Bedtime Procrastination Scale, comparing one-factor and two-factor specifications with different item sets.

In the one-factor full-item model, fit indices were chi-square over degrees of freedom equals fifteen point fifty-five, TLI equals zero point seventy-seven, CFI equals zero point eighty-three, and RMSEA equals zero point sixteen; the table also reports factor loadings below zero point fifty for Items two and three in relevant models.

These results document why the authors examined models with Item two, and then Items two and three, eliminated. Table 4 summarizes the BPS-J’s score characteristics and reliability for the full-item model, the model with Item 2 eliminated, and the model with Items 2 and 3 eliminated.

Average-score ICC and Pearson’s r are 0.78 across all three versions, while sum-score ICC ranges from 0.77 to 0.78 and Pearson’s r remains 0.78. The table also reports measurement error through SEM and MDC or SDC, alongside internal-consistency estimates from Cronbach’s alpha and McDonald’s omega.

Regardless of the model, significant correlations between the BPS-J and the variables were shown, except for the degree of viewing bedtime procrastination as a problem, working time, and work start time. Table five reports Spearman correlations between the BPS-J and related behavioral, sleep, and work variables across the full scale and versions eliminating Item two or Items two and three.

Correlations were significant for most measures, including general procrastination, at zero point four nine for the full item version, and shorter workday sleep, at negative zero point three five. The exceptions were viewing bedtime procrastination as a problem, working time, and work start time, supporting the authors’ summary of the scale’s bivariate validity.

The multivariate regressions found that BPS-J score was significantly related to shorter sleep duration on workdays, more frequent days of feeling daytime fatigue, and more frequent days of feeling sleep loss. BPS-J score was also significantly related to viewing bedtime procrastination as a problem less seriously.

Table seven reports regression analyses linking the BPS-J sum score with sleep-related measures, across the total score and versions with item two or items two and three removed. The authors found significant associations with shorter workday sleep, more frequent daytime fatigue and sleep loss, and the Munich ChronoType Questionnaire measures; for example, seven days of feeling sleep loss had odds ratios of one point three one, one point three two, and one point three eight, all with p less than zero point zero zero one.

These findings support the BPS-J’s criterion validity through its relationship with sleep variables. Except for structural validity, the BPS-J had acceptable validity and reliability, similar to other language versions. The two-factor version that eliminated Item 2 indicated slightly better validity than the full-item version.

The factor loading of Items 2 and 3, especially Item 2, was not as high as that of other items, resulting in consideration of eliminating Item 2 from the BPS-J scale. The validity of the ISS diagnosis is subject to potential inaccuracies because it relies solely on participants’ self-reported accounts.

Participants were limited to daytime workers, and the BPS-J should be applied to other populations, such as students, with caution. Individuals whose bedtime procrastination tendency was so severe that they could not perform daytime work were not included, so additional validation studies including such individuals are necessary.

The BPS-J showed acceptable validity and reliability, while its Japanese data supported a two-factor version with Item 2 removed. It is promising, but should be applied cautiously beyond daytime workers.