Childhood Environmental Unpredictability and Bedtime Procrastination Among Undergraduate Medical Students in South India: A Cross-sectional Study

Article information

Chronobiol Med. 2026;8(3):176-181
Publication date (electronic) : 2026 September 30
doi : https://doi.org/10.33069/cim.2026.0038
1Undergraduate Medical Student, Alluri Sitarama Raju Academy of Medical Sciences, Eluru, India
2Department of Biochemistry, Alluri Sitarama Raju Academy of Medical Sciences, Eluru, India
3Department of Community Medicine, Alluri Sitarama Raju Academy of Medical Sciences, Eluru, India
Corresponding author: Sai Kiran Santosh Pissey, MD, Department of Community Medicine, Alluri Sitarama Raju Academy of Medical Sciences, Malkapuram, NH-16, Eluru, Andhra Pradesh 534005, India. Tel: 91-8074939372, E-mail: pisseysaikiransantosh@gmail.com
Received 2026 July 21; Revised 2026 August 12; Accepted 2026 August 15.

Abstract

Objective

To examine the association between retrospectively reported childhood environmental unpredictability and bedtime procrastination among undergraduate medical students in South India.

Methods

This single-center cross-sectional online survey included 200 undergraduate medical students. Bedtime procrastination was assessed using the nine-item Bedtime Procrastination Scale (BPS; range, 9–45), and childhood unpredictability using the 38-item Questionnaire of Unpredictability in Childhood (QUIC; range, 0–38). Because the BPS has no validated diagnostic cutoff, scores were analyzed continuously. The primary analysis used Spearman rank correlation with bootstrap confidence intervals (CI), and robust linear regression was used to quantify the association. An additional multivariable model adjusted for age, sex, academic year, residence, and socioeconomic category.

Results

The mean age was 18.8 years (standard deviation [SD], 1.1), and 67.0% of participants were women. Mean BPS and QUIC scores were 27.93 (SD, 7.19) and 9.39 (SD, 5.53), respectively. Overall QUIC score was weakly associated with BPS score (Spearman rho=0.242; 95% bootstrap confidence interval, 0.111–0.366; p<0.001). After adjustment for age, sex, academic year, residence, and socioeconomic category, a five-point higher QUIC score was associated with a 1.48-point higher mean BPS score (95% CI, 0.52–2.43; p=0.003). Three parental domains showed small exploratory associations after false-discovery-rate adjustment.

Conclusion

Childhood environmental unpredictability showed a small positive association with bedtime procrastination. The cross-sectional, retrospective findings are hypothesis-generating and do not establish causality.

INTRODUCTION

Bedtime procrastination refers to going to bed later than intended despite the absence of external circumstances that require the delay. It is commonly conceptualized as a self-regulation problem in which immediately rewarding activities are prioritized over the longer-term benefits of adequate sleep [1,2]. The nine-item Bedtime Procrastination Scale (BPS) was developed to measure the tendency as a continuous construct [1]. Subsequent work has distinguished delay before entering bed from delay that occurs while already in bed, indicating that sleep procrastination may encompass more than one behavioral stage [3].

Young adults and students may be especially vulnerable because academic demands, irregular schedules, digital media use and greater autonomy over daily routines converge during this period. Bedtime mobile-phone use and problematic smartphone use have been associated with poorer sleep outcomes and bedtime procrastination [4,5]. In an Indian university sample, the mean BPS score was 29.1; that study also noted that population norms and validated cutoffs for the BPS were unavailable and raised concerns about the interpretation of reverse-scored items in Indian students [6]. Accordingly, BPS scores are best treated as continuous unless a validated threshold is available.

Most research has focused on proximal correlates such as self-control, stress, chronotype and digital behavior. A developmental perspective proposes that unpredictability in early caregiving, routines and the physical environment may shape perceived control, present-oriented decision-making and later self-regulation [7]. Childhood unpredictability is conceptually distinct from harshness. It can be assessed using the Questionnaire of Unpredictability in Childhood (QUIC), a 38-item instrument covering parental monitoring and involvement, parental predictability, parental environment, physical environment, and safety and security [8-10].

Evidence directly linking childhood unpredictability with bedtime procrastination remains limited and has largely come from non-Indian student samples using narrower indicators of instability [7]. The present exploratory study therefore examined the association between retrospectively reported childhood environmental unpredictability and bedtime procrastination among undergraduate medical students in South India. We hypothesized that higher overall QUIC scores would be associated with higher BPS scores. Associations with the five QUIC domains were examined as secondary, exploratory analyses.

METHODS

Study design and setting

This analytical cross-sectional study was conducted during 2025 among undergraduate medical students at a single private medical college in Andhra Pradesh, South India. Reporting was structured with reference to the Strengthening the Reporting of Observational Studies in Epidemiology recommendations for cross-sectional studies [11].

Participants and recruitment

Students enrolled in the first through fourth academic years were eligible. Those who provided electronic informed consent and submitted a complete questionnaire were included; students who declined consent or submitted incomplete responses were excluded. The survey link was circulated to students in each academic year through class communication channels. Participation was voluntary, and convenience sampling was used until 200 complete responses had been obtained. Because the invitation was distributed through group channels, the number of students who viewed it was not available, so a conventional response rate could not be calculated.

Sample size

The original pre-recruitment calculation used a study-defined high BPS-score proportion of 20.2% reported in an Indian university sample [6]. With 95% confidence and 6% absolute precision, the minimum sample size was 172. After allowing 10% for non-response, the target was 189 and was rounded to 200. Because that BPS categorization is pragmatic rather than a validated diagnostic threshold, the present analysis does not report prevalence. A sensitivity calculation indicated that 200 participants provided approximately 80% power, with a two-sided alpha of 0.05, to detect a correlation of approximately 0.20.

Measures

Bedtime procrastination

The BPS contains nine items rated from 1 (rarely) to 5 (almost always). Items 2, 3, 7 and 9 are reverse scored. Total scores range from 9 to 45, with higher scores indicating greater bedtime procrastination [1]. Because no validated population cutoff or severity classification is available, the total score was analyzed as a continuous variable, and no prevalence estimate was generated.

Childhood unpredictability

The QUIC is a 38-item retrospective self-report instrument with dichotomous responses. Selected items are reverse scored and summed to create five domain scores: parental monitoring and involvement (0–9), parental predictability (0–12), parental environment (0–7), physical environment (0–7), and safety and security (0–3). The overall score ranges from 0 to 38; higher scores indicate greater reported childhood unpredictability [8].

Both the BPS and QUIC were administered in their English-language versions.

Other variables

Age, sex, academic year, residence (hostel or day scholar), and socioeconomic category were collected for each participant. Socioeconomic status was classified using the Modified B.G. Prasad approach current at the time of data collection and collapsed into lower, middle, and upper groups. For multivariable modelling, age was entered as a continuous variable; sex, academic year, residence, and socioeconomic category were entered as indicator variables, with male sex, first academic year, hostel residence, and lower socioeconomic category as the reference groups.

Data handling and statistical analysis

All 200 included records had complete total BPS and QUIC scores; no imputation was performed. Continuous variables were summarized using mean and standard deviation and, where useful, median, interquartile range, and observed range. Categorical variables were reported as counts and percentages. The primary association between overall QUIC and BPS scores was estimated using Spearman rank correlation. A 95% percentile bootstrap confidence interval (CI) was obtained from 20,000 resamples. For clinical interpretability, a sensitivity analysis used ordinary least-squares regression with heteroscedasticity-consistent type 3 (HC3) standard errors; the coefficient was expressed per five-point increase in QUIC score. An additional multivariable linear regression model was fitted with BPS total score as the dependent variable and overall QUIC score as the primary exposure, adjusting simultaneously for age, sex, academic year, residence, and socioeconomic category. Age was entered continuously and categorical covariates were represented using indicator variables. Heteroscedasticity-consistent HC3 standard errors were used, and the QUIC coefficient was expressed per five-point increase in the overall QUIC score. Internal consistency of the BPS and overall QUIC in the present sample was assessed using Cronbach’s alpha after application of the prescribed scoring procedures. Residual normality, linearity, and homoscedasticity were examined for the total-score model. Correlations between BPS and the five QUIC domains were secondary and exploratory; Benjamini-Hochberg false-discovery-rate adjustment was applied across these five tests [12]. For completeness, conventional ordinary least-squares models were also tabulated: a simple model using the overall QUIC score and an exploratory simultaneous model containing the five QUIC domains. These secondary domain-model p-values were not adjusted for multiplicity and were interpreted cautiously. All tests were two-sided. Data management was undertaken in IBM SPSS Statistics version 26.0 (IBM Corp.), and the reproducible total-score reanalysis was performed using Python version 3.13 (https://www.python.org/downloads/release/python-3130/), SciPy version 1.17 (https://docs.scipy.org/doc/scipy/release/1.17.0-notes.html), and statsmodels version 0.14.6 (https://www.statsmodels.org/v0.14.6/release/version0.14.6.html).

Ethical considerations

The study was approved by the Institutional Ethics Committee of Alluri Sitarama Raju Academy of Medical Sciences (Approval No. 181/2024; May 20, 2024). Electronic informed consent was obtained before participation. No personally identifying information was included in the analytic dataset.

RESULTS

All 200 complete responses were analyzed. The mean age was 18.8 years (standard deviation [SD], 1.1; range, 16–24 years), 134 participants (67.0%) were women, and 134 (67.0%) lived in a hostel. Representation across academic years ranged from 22.0% to 28.0% (Table 1). Internal consistency was good for both instruments, with Cronbach’s alpha=0.829 for the BPS and alpha=0.810 for the overall QUIC.

Participant characteristics (n=200)

The mean BPS score was 27.93 (SD, 7.19), with a median of 28.0 (interquartile range, 22.75–33.00) and an observed range of 10–44. The mean overall QUIC score was 9.39 (SD, 5.53), with a median of 9.0 (interquartile range, 5.0–12.0) and an observed range of 0–27 (Table 2).

Distribution of bedtime procrastination and childhood unpredictability scores

In exploratory domain analyses, greater unpredictability in parental monitoring and involvement (rho=0.236; false-discovery-rate-adjusted q=0.004), parental predictability (rho=0.211; q=0.007), and parental environment (rho=0.174; q=0.023) was associated with higher BPS scores. Correlations for physical environment (rho=0.100; q=0.199) and safety and security (rho=0.037; q=0.603) were not statistically significant (Table 3). The complete intercorrelation matrix for bedtime procrastination, overall QUIC score, and the five QUIC domains is presented in Supplementary Table 1. In the exploratory simultaneous-domain regression, the model explained 8.6% of BPS variance (adjusted R2=0.062; p=0.004). Parental monitoring and involvement showed a nominal association (B=0.572; P=0.046), whereas the remaining domain coefficients were not statistically significant (Table 4).

Associations of overall and domain-specific childhood unpredictability with bedtime procrastination

Conventional linear regression analyses predicting bedtime procrastination

Overall childhood unpredictability had a small positive correlation with bedtime procrastination (Spearman rho=0.242; 95% bootstrap CI, 0.111–0.366; p<0.001) (Figure 1). In the robust linear sensitivity analysis, each five-point higher QUIC score was associated with a 1.57-point higher mean BPS score (95% CI, 0.64–2.49; p=0.001). The model accounted for 5.8% of the observed variance in BPS score. The corresponding conventional simple-regression estimate was similar (B=0.313 per one-point increase in QUIC) (Table 4).

Figure 1.

Relationship between overall childhood unpredictability and bedtime procrastination. Bubble size represents the number of participants with identical score pairs. The solid line represents the ordinary least-squares regression fit, and the shaded band represents the 95% confidence interval for the mean response. The regression line and confidence band are provided for visualization only; the primary inferential analysis was based on Spearman rank correlation. QUIC is shown on the horizontal axis and BPS on the vertical axis to reflect the exposure–outcome ordering used in the analysis. BPS, Bedtime Procrastination Scale; QUIC, Questionnaire of Unpredictability in Childhood.

In the covariate-adjusted analysis, the association between childhood unpredictability and bedtime procrastination remained statistically significant. After simultaneous adjustment for age, sex, academic year, residence, and socioeconomic category, a fivepoint higher overall QUIC score was associated with a 1.48-point higher mean BPS score (95% CI, 0.52–2.43; p=0.003). The model explained 9.3% of the variance in BPS score (R2=0.093; adjusted R2=0.050) (Table 5). The adjusted effect estimate was similar to the unadjusted robust estimate, indicating that these measured demographic characteristics did not materially account for the observed association.

Covariate-adjusted robust linear regression predicting bedtime procrastination

DISCUSSION

In this exploratory sample of undergraduate medical students, retrospectively reported childhood unpredictability was positively associated with bedtime procrastination, but the magnitude was small. A five-point difference in overall QUIC score corresponded to an estimated 1.57-point difference on the 9–45 BPS scale in the unadjusted robust model, and the estimate remained similar after adjustment for age, sex, academic year, residence, and socioeconomic category (1.48 points; 95% CI, 0.52–2.43). The unadjusted total-score model explained 5.8% of variance, while the fully adjusted model explained 9.3% (adjusted R2=0.050). The findings therefore support a modest association rather than a strong developmental determination of current bedtime behavior.

The direction of association is consistent with a study of Chinese college students in which childhood environmental harshness and unpredictability were related to bedtime procrastination through life-history strategy and sense of control [7]. The present study extends that observation to an Indian medical-student sample and uses the broader QUIC measure, which captures routine, caregiving, and environmental instability rather than only major life events [8]. Nevertheless, the cross-sectional and retrospective design does not test the proposed pathway or establish temporal causation.

The mean BPS score of 27.93 was close to the mean of 29.1 reported among Indian university students by Shukla and Andrade [6]. The earlier study explicitly noted that the BPS lacks population norms and validated diagnostic cutoffs. The present interpretation concerns score level and association, not prevalence or clinical severity.

The exploratory domain pattern suggests that inconsistency in parental monitoring, parental behavior, and the wider parental environment may be more closely related to later bedtime self-regulation than physical instability or safety concerns. This interpretation is biologically and behaviorally plausible because predictable routines and consistent caregiving may support the development of behavioral structure and perceived control [9,10]. However, domain correlations were weak, the domains were self-reported retrospectively, and the study did not measure potential mediators. The findings should therefore be treated as signals for future hypothesis testing rather than evidence for a specific mechanism. The simultaneous-domain regression did not provide strong evidence of independent effects across domains; the nominal finding for parental monitoring and involvement was secondary, unadjusted for multiplicity, and close to the conventional significance threshold.

Bedtime procrastination is likely to be strongly shaped by proximal factors not measured here, including chronotype, sleep opportunity, examination schedules, stress, mood symptoms, selfcontrol, and late-night digital engagement [1,2,4,5]. The small R² value is consistent with this multifactorial model. The present data do not justify screening students for childhood adversity on the basis of bedtime behavior or introducing trauma-focused interventions. More immediate and modifiable determinants should be measured alongside developmental exposures in future work.

These findings also need to be viewed in the broader context of sleep and health in young adults. Childhood unpredictability has been linked with later health-related outcomes, suggesting that early environmental experiences may continue to influence behaviour beyond childhood [13]. At the same time, sleep health depends on several dimensions, including adequate duration, regularity and appropriate timing [14], and current recommendations emphasise the importance of sufficient sleep for maintaining health in adults [15]. College students are particularly prone to irregular sleep patterns because of academic demands, lifestyle factors and changing daily routines [16]. Previous work has also shown that bedtime procrastination is related to sleep behaviours and demographic characteristics [17]. Taken together, these findings suggest that bedtime procrastination is likely to reflect the combined influence of earlier developmental experiences and more immediate behavioural and lifestyle factors.

This study has several strengths: it used established instruments, treated the BPS as a continuous construct, prespecified an overall-score hypothesis, reported an effect-size CI, and controlled the false-discovery rate for exploratory domain analyses. Important limitations remain. The single-center convenience sample limits generalizability and may be affected by self-selection. Both exposure and outcome were self-reported in the same survey, creating potential common-method bias, and childhood experiences were recalled retrospectively. Residual and unmeasured confounding remains possible. Although age, sex, academic year, residence, and socioeconomic category were included in the adjusted analysis, several potentially important proximal determinants of bedtime procrastination were not measured, including chronotype, sleep duration and quality, insomnia symptoms, smartphone and other late-night digital-media use, academic stress, anxiety, and depression. These factors may influence bedtime procrastination and could partly account for the observed relationship with childhood unpredictability. Consequently, the observed association should not be interpreted as evidence of an independent developmental or causal effect. Although internal consistency was good for both instruments in the present sample, broader psychometric properties, including factor structure and the performance of reverse-scored BPS items, were not evaluated and warrant further study in Indian student populations. The cross-sectional design precludes causal inference.

Future multicenter studies should recruit probability-based or consecutively invited samples, evaluate broader local psychometric properties including factor structure and item performance, and include sleep, psychological, and digital-behavioral covariates. Longitudinal designs could examine whether perceived control or self-regulatory capacity mediates any association between early-life unpredictability and later bedtime behaviour.

In conclusion, retrospectively reported childhood environmental unpredictability was weakly associated with higher bedtime procrastination scores among undergraduate medical students. The association was statistically reliable but small and cannot be interpreted causally. These findings support further longitudinal, multicenter research integrating developmental exposures with current sleep, mental-health, and digital-behavioral factors.

Supplementary Materials

The Supplement is available with this article at https://doi.org/10.33069/cim.2026.0038.

Supplementary Table 1.

Spearman correlation matrix between bedtime procrastination and childhood unpredictability measures

cim-2026-0038-Supplementary-Table-1.pdf

Notes

The authors have no potential conflicts of interest to disclose.

Availability of Data and Material

Data supporting the findings are available from the corresponding author upon reasonable request and subject to institutional and ethical permissions.

Author Contributions

Conceptualization: Nafisa Nikhat Shaik, Sameera Ketavarapu. Data curation: Nafisa Nikhat Shaik. Formal analysis: Sai Kiran Santosh Pissey. Investigation: Nafisa Nikhat Shaik, Sai Kiran Santosh Pissey, Rama Lakshmi Makineedi. Methodology: Sameera Ketavarapu, Sai Kiran Santosh Pissey. Project administration: Sameera Ketavarapu. Supervision: Sameera Ketavarapu. Visualization: Sai Kiran Santosh Pissey. Writing—original draft: Nafisa Nikhat Shaik. Writing—review & editing: Sameera Ketavarapu, Sai Kiran Santosh Pissey, Rama Lakshmi Makineedi.

Funding Statement

None

Acknowledgments

None

References

1. Kroese FM, De Ridder DT, Evers C, Adriaanse MA. Bedtime procrastination: introducing a new area of procrastination. Front Psychol 2014;5:611.
2. Kroese FM, Evers C, Adriaanse MA, de Ridder DTD. Bedtime procrastination: a self-regulation perspective on sleep insufficiency in the general population. J Health Psychol 2016;21:853–862.
3. Magalhães P, Cruz V, Teixeira S, Fuentes S, Rosário P. An exploratory study on sleep procrastination: bedtime vs. while-in-bed procrastination. Int J Environ Res Public Health 2020;17:5892.
4. Exelmans L, Van den Bulck J. Bedtime mobile phone use and sleep in adults. Soc Sci Med 2016;148:93–101.
5. Correa-Iriarte S, Hidalgo-Fuentes S, Martí-Vilar M. Relationship between problematic smartphone use, sleep quality and bedtime procrastination: a mediation analysis. Behav Sci (Basel) 2023;13:839.
6. Shukla A, Andrade C. Prevalence of bedtime procrastination in university students and reexamination of the bedtime procrastination scale. Prim Care Companion CNS Disord 2023;25:22m03334.
7. Shao L, Jin J, Yu G. Childhood environmental risk and youth bedtime procrastination: a path model with life history strategy and sense of control as mediators. Child Abuse Negl 2024;150:106137.
8. Glynn LM, Stern HS, Howland MA, Risbrough VB, Baker DG, Nievergelt CM, et al. Measuring novel antecedents of mental illness: the questionnaire of unpredictability in childhood. Neuropsychopharmacology 2019;44:876–882.
9. Ellis BJ, Figueredo AJ, Brumbach BH, Schlomer GL. Fundamental dimensions of environmental risk: the impact of harsh versus unpredictable environments on the evolution and development of life history strategies. Hum Nat 2009;20:204–268.
10. Doom JR, Han D, Rivera KM, Tseten T. Childhood unpredictability research within the developmental psychopathology framework: advances, implications, and future directions. Dev Psychopathol 2024;36:2452–2463.
11. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med 2007;4e296.
12. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol 1995;57:289–300.
13. Maner JK, Hasty CR, Martinez JL, Ehrlich KB, Gerend MA. The role of childhood unpredictability in adult health. J Behav Med 2023;46:417–428.
14. Buysse DJ. Sleep health: can we define it? Does it matter? Sleep 2014;37:9–17.
15. Watson NF, Badr MS, Belenky G, Bliwise DL, Buxton OM, Buysse D, et al. Recommended amount of sleep for a healthy adult: a joint consensus statement of the American Academy of Sleep Medicine and Sleep Research Society. Sleep 2015;38:843–844.
16. Lund HG, Reider BD, Whiting AB, Prichard JR. Sleep patterns and predictors of disturbed sleep in a large population of college students. J Adolesc Health 2010;46:124–132.
17. Herzog-Krzywoszanska R, Krzywoszanski L. Bedtime procrastination, sleep-related behaviors, and demographic factors in an online survey on a Polish sample. Front Neurosci 2019;13:963.

Article information Continued

Figure 1.

Relationship between overall childhood unpredictability and bedtime procrastination. Bubble size represents the number of participants with identical score pairs. The solid line represents the ordinary least-squares regression fit, and the shaded band represents the 95% confidence interval for the mean response. The regression line and confidence band are provided for visualization only; the primary inferential analysis was based on Spearman rank correlation. QUIC is shown on the horizontal axis and BPS on the vertical axis to reflect the exposure–outcome ordering used in the analysis. BPS, Bedtime Procrastination Scale; QUIC, Questionnaire of Unpredictability in Childhood.

Table 1.

Participant characteristics (n=200)

Variable Value
Age (yr) 18.8±1.1 (16–24)
Sex
 Men 66 (33.0)
 Women 134 (67.0)
Academic year
 First 48 (24.0)
 Second 52 (26.0)
 Third 56 (28.0)
 Fourth 44 (22.0)
Residence
 Hostel 134 (67.0)
 Day scholar 66 (33.0)
Socioeconomic category
 Lower 38 (19.0)
 Middle 112 (56.0)
 Upper 50 (25.0)

Data are presented as mean±standard deviation (range) or n (%).

Table 2.

Distribution of bedtime procrastination and childhood unpredictability scores

Measure Theoretical range Mean±SD Median Observed range
BPS total 9–45 27.93±7.19 28.0 10–44
QUIC overall 0–38 9.39±5.53 9.0 0–27
Parental monitoring and involvement 0–9 1.84±1.86 1.0 0–8
Parental predictability 0–12 4.31±2.61 4.0 0–12
Parental environment 0–7 1.09±1.25 1.0 0–6
Physical environment 0–7 1.50±1.39 1.0 0–6
Safety and security 0–3 0.60±0.84 0.0 0–3

SD, standard deviation; BPS, Bedtime Procrastination Scale; QUIC, Questionnaire of Unpredictability in Childhood.

Table 3.

Associations of overall and domain-specific childhood unpredictability with bedtime procrastination

QUIC exposure Spearman rho with BPS Unadjusted p FDR-adjusted q
Overall QUIC score (primary) 0.242 <0.001 Not applicable
Parental monitoring and involvement 0.236 <0.001 0.004
Parental predictability 0.211 0.003 0.007
Parental environment 0.174 0.014 0.023
Physical environment 0.100 0.159 0.199
Safety and security 0.037 0.603 0.603

The overall QUIC score was the prespecified primary analysis. FDR adjustment was applied across the five domain-level tests only. QUIC, Questionnaire of Unpredictability in Childhood; BPS, Bedtime Procrastination Scale; FDR, false discovery rate.

Table 4.

Conventional linear regression analyses predicting bedtime procrastination

Model/predictor B SE 95% CI Beta t p
Simple linear regression
 Constant 24.984 0.975 23.061–26.908 - 25.620 <0.001
 Overall childhood unpredictability 0.313 0.089 0.137–0.490 0.241 3.500 <0.001
Multiple linear regression
 Constant 24.777 1.011 22.783–26.771 - 24.505 <0.001
 Parental monitoring and involvement 0.572 0.284 0.011–1.133 0.148 2.011 0.046
 Parental predictability 0.310 0.223 -0.129–0.749 0.113 1.391 0.166
 Parental environment 0.938 0.499 -0.047–1.922 0.163 1.879 0.062
 Physical environment 0.345 0.421 -0.485–1.175 0.067 0.821 0.413
 Safety and security -1.317 0.732 -2.759–0.126 -0.154 -1.800 0.073

Conventional ordinary least-squares estimates are shown. For the overall QUIC score, the HC3 robust unadjusted estimate was B=1.57 per 5-point increase in QUIC (95% CI, 0.64–2.49; p=0.001). The covariate-adjusted HC3 model including age, sex, academic year, residence, and socioeconomic category is presented in Table 5. The multiple regression model shown here was exploratory, included all five QUIC domains simultaneously, and its p-values were not adjusted for multiple testing. B, unstandardized regression coefficient; SE, standard error; CI, confidence interval; Beta, standardized coefficient; QUIC, Questionnaire of Unpredictability in Childhood; HC3, heteroscedasticity-consistent type 3.

Table 5.

Covariate-adjusted robust linear regression predicting bedtime procrastination

Predictor B Robust SE 95% CI p Reference/scale
Intercept 16.727 9.051 -1.013–34.466 0.065 -
Overall QUIC score 1.475 0.489 0.518–2.433 0.003 Per 5-point increase
Age 0.480 0.475 -0.451–1.412 0.312 Per year
Sex: female -0.892 1.113 -3.073–1.289 0.423 Male reference
Academic year: second 3.775 2.814 -1.739–9.290 0.180 First year reference
Academic year: third 1.954 2.979 -3.885–7.794 0.512 First year reference
Academic year: fourth 1.234 3.499 -5.624–8.093 0.724 First year reference
Residence: day scholar 0.080 1.913 -3.670–3.829 0.967 Hostel reference
Socioeconomic category: middle -1.961 2.952 -7.746–3.824 0.506 Lower reference
Socioeconomic category: upper -2.857 3.635 -9.982–4.268 0.432 Lower reference

The dependent variable was total Bedtime Procrastination Scale score. Coefficients were estimated using ordinary least squares with heteroscedasticity-consistent HC3 standard errors. Age was entered continuously; sex, academic year, residence, and socioeconomic category were represented using indicator variables. The QUIC coefficient is expressed per five-point increase. The multiple linear regression model showed R=0.093, adjusted R2= 0.050, with an overall p-value of 0.019 (n=200). B, unstandardized regression coefficient; SE, standard error; CI, confidence interval; QUIC, Questionnaire of Unpredictability in Childhood; HC3, heteroscedasticity-consistent type 3.