A comparative study of estimators in multilevel linear models

Multilevel Models are widely used in organizational research, educational research, epidemiology, psychology, biology and medical fields. In this paper, we recommend the situations where Bootstrap procedures through Minimum Norm Quadratic Unbiased Estimator (MINQUE) can be extremely handy than that of Restricted Maximum Likelihood (REML) in multilevel level linear regression models. In our simulation study the bootstrap by means of MINQUE is superior to REML in conditions where normality does not hold. Moreover, the real data application also supports our findings in terms of accuracy of estimates and their standard errors.

A teacher like me? Different approaches to examining personality similarity between teachers and students

Research suggests that people tend to like others more if they are similar rather than dissimilar to themselves. Likewise, students may tend to prefer teachers with whom they share similar personality characteristics. To test this hypothesis, we examined the role of personality similarity between students and teachers in predicting how much students liked their teachers. Secondary school students ( N = 634) provided self-reports and reported on their teachers’ personality using a Big Five personality scale. Their teachers ( N = 31) also provided self-reports. These reports were then used to compute three indices of similarity; i.e. perceptual similarity, actual similarity, and perceptual accuracy which were used to explain teacher liking. Multilevel linear models showed that perceptual similarity (computed as the profile agreement across student self- and teacher-ascribed Big Five traits) had the largest effect on teacher liking. Teachers described as more agreeable and conscientious were liked by their students more. Findings highlight the importance of considering students’ perceptions of personality similarity with their teachers for understanding how students feel “connected” to their teachers and positively interact with them.

Associations between Sociodemographic, Dietary, and Substance Use Factors with Self-Reported 24-Hour Movement Behaviors in a Sample of Brazilian Adolescents

We aimed to identify sociodemographic, dietary, and substance use factors associated with self-reported sleep duration, physical activity (PA), and sedentary behavior (SB) indicators in a sample of Brazilian adolescents. Adolescents (n = 731, 51% female, mean age: 16.4 years) answered a questionnaire. The volume of total PA, sports, non-sports, total SB, leisure-time SB, involuntary SB, sleep duration, dietary behaviors, sociodemographic, and substance use indicators were self-reported. Multilevel linear models were fitted. Females engaged in less total PA, sports, total SB, and leisure-time SB, but in more involuntary SB than males. Age was positively associated with non-sports and involuntary SB. Socioeconomic status was positively associated with total PA. Adolescents who lived with the mother only practiced more sports compared to those living with two parents. Unprocessed food was positively associated with total PA and sports. Processed food was inversely associated with total PA and non-sports, and positively associated with total SB and leisure-time SB. Alcohol use was positively associated with total PA, and tobacco smoking was negatively associated with total PA. No associations were observed for sleep duration. In conclusion, sociodemographic, dietary, and substance use factors are associated with the 24 h movement behaviors among Brazilian adolescents, and some associations are type specific.

The development of a standard procedure for the optimal reliability-feasibility trade-off in Multilevel Linear Models analyses in Psychology and Neuroscience.

The use of Multilevel Linear Models (MLMs) is increasing in Psychology and Neuroscience research. A key aspect of MLMs is choosing a random effects structure according to the experimental needs. To date, opposite suggestions are present in the literature, spanning from keeping all random effects, which produces several singularity and convergence issues and often requires high computational resources, to removing random effects until the best fit is found, with the risk of inflating first-type error. However, defining the random structure to fit a non-singular and convergent model is not straightforward. Moreover, the lack of a standard approach may lead the researcher to make decisions that potentially inflate first-type errors and generate distortions in the estimates. To date, how to deal with singular and non-converging models is an ongoing debate.We introduce a new way to control for first-type error inflation during model reduction, namely transforming random slopes in complex random intercepts (CRIs). These are multiple random intercepts that represent the complexity of the factors within a given grouping factor. We demonstrate that CRIs can produce reliable results, require less computational and timing resources, and we provide a straightforward procedure to use CRIs in MLMs reduction. Importantly we outline a few criteria and clear recommendations on how and when scholars should reduce singular and non-converging models.We validated this approach by extensive simulations, a real-case application, and a comprehensive procedure that defines new solutions to avoid overinflated results and potentially standardise the use of MLMs in Psychology and Neuroscience.

Is the Urine Cannabinoid Level Measured via a Commercial Point-of-Care Semiquantitative Immunoassay a Cannabis Withdrawal Syndrome Severity Predictor?

Background: For cannabis-dependent subjects, the relationship between cannabis withdrawal syndrome (CWS) severity and the urine cannabinoid concentrations are unclear; we investigated this using a commercial point-of-care (POC) enzyme immunoassay detecting 11-nor-9-carboxy-Delta-9-tetrahydrocannabinol (THC-COOH).Methods: Observational study of 78 adult chronic cannabis-dependent subjects assessed over a 24-day inpatient detoxification treatment, with 13 serial measurement days. Repeated Measures Correlation and Multilevel Linear Models were employed.Results: Absolute urinary THC-COOH levels significantly correlated with Marijuana Withdrawal Checklist (MWC) scores across the entire study duration (r = 0.248; p < 0.001). Correlation between serial creatinine-adjusted THC-COOH ratios and serial MWC scores emerged as significant only in the sample with higher MWC scores (>11 points) at admission (n = 21; r = 0.247; p = 0.002). The aforementioned significant relationships have persisted when replacing the absolute THC-COOH-levels with the (relative) day-to-day change in urinary THC-COOH levels. MWC scores were significantly correlated with the Clinical Global Impression-Severity (CGI-S; r = 0.812; p < 0.001). Females showed a significantly slower decline in urine THC-COOH levels and prolonged CWS course characterized by substantial illness severity (per CGI-S), occurring in nearly 30% of cases.Conclusion: Urine cannabinoid levels (THC-COOH) determined by POC assay significantly predicted CWS severity (moderate correlation), guiding detoxification treatment duration. In patients with MWC > 11 points upon admission, creatinine-adjusted THC-COOH ratios also significantly predicted CWS severity—again with moderate effect size. Females showed prolonged urinary THC-COOH elimination and cannabis withdrawal.

Knowledge of Dental Academics about the COVID-19 Pandemic: A Multi-Country Online Survey

Background COVID-19 is a global pandemic affecting all aspects of life in all countries. We assessed COVID-19 knowledge and associated factors among dental academics in various countries. Method We invited dental academics to participate in a cross-sectional, multi-country, online survey from March to April 2020. The survey assessed knowledge of COVID-19 regarding the mode of transmission, symptoms, diagnosis, treatment, protection, and dental treatment precautions as well as participants’ background variables. The analysis was based on multilevel linear models to assess the association between knowledge and factors at individual levels (personal and professional) and country-level (number of COVID-19 cases/ million population), accounting for random variation among countries. Results Two thousand forty-five academics from 26 countries participated in the survey (response rate= 14.3%, with 54.7% female and 67% younger than 46 years of age). The mean (SD) knowledge percent score was 73.2% (11.2), and the knowledge of symptoms score was significantly lower than the diagnostic methods score (53.1% and 85.4%). Knowledge was significantly higher among those living with partner/spouse than those living alone (regression coefficient (B)= 0.48); those with PhD than those with BDS (B= 0.48), those seeing 21 to 30 patients daily than those seeing no patients (B= 0.65) and those from countries with a higher number of COVID-19 cases/million population (B= 0.0007). Conclusions Dental academics had poorer knowledge of COVID-19 symptoms than diagnostic methods. Living arrangements, academic degrees, patient load, and magnitude of epidemic in the country were associated with COVD-19 knowledge among dental academics. COVID-19 training can be designed using these factors to target academics with the greatest need.

Urinary Sodium Excretion and Blood Pressure Relationship across Methods of Evaluating the Completeness of 24-h Urine Collections

We compared the sodium intake and systolic blood pressure (SBP) relationship from complete 24-h urine samples determined by several methods: self-reported no-missed urine, creatinine index ≥0.7, measured 24-h urine creatinine (mCER) within 25% and 15% of Kawasaki predicted urine creatinine, and sex-specific mCER ranges (mCER 15–25 mg/kg/24-h for men; 10–20 mg/kg/24-h for women). We pooled 10,031 BP and 24-h urine sodium data from 2143 participants. We implemented multilevel linear models to illustrate the shape of the sodium–BP relationship using the restricted cubic spline (RCS) plots, and to assess the difference in mean SBP for a 100 mmol increase in 24-h urine sodium. The RCS plot illustrated an initial steep positive sodium–SBP relationship for all methods, followed by a less steep positive relationship for self-reported no-missed urine, creatinine index ≥0.7, and sex-specific mCER ranges; and a plateaued relationship for the two Kawasaki methods. Each 100 mmol/24-h increase in urinary sodium was associated with 0.64 (95% CI: 0.34, 0.94) mmHg higher SBP for self-reported no-missed urine, 0.68 (95% CI: 0.27, 1.08) mmHg higher SBP for creatinine index ≥0.7, 0.87 (95% CI: 0.07, 1.67) mmHg higher SBP for mCER within 25% Kawasaki predicted urine creatinine, 0.98 (95% CI: −0.07, 2.02) mmHg change in SBP for mCER within 15% Kawasaki predicted urine creatinine, and 1.96 (95% CI: 0.93, 2.99) mmHg higher SBP for sex-specific mCER ranges. Studies examining 24-h urine sodium in relation to health outcomes will have different results based on how urine collections are deemed as complete.

Knowledge of Dental Academics about the COVID-19 Pandemic: A Multi-Country Online Survey

Background: COVID-19 is a global pandemic affecting all aspects of life in all countries. We assessed COVID-19 knowledge and associated factors among dental academics in various countries. Method: We invited dental academics to participate in a cross-sectional, multi-country, online survey from March to April 2020. The survey assessed knowledge of COVID-19 regarding the mode of transmission, symptoms, diagnosis, treatment, protection, and dental treatment precautions as well as participants’ background variables. The analysis was based on multilevel linear models to assess the association between knowledge and factors at individual levels (personal and professional) and country-level (number of COVID-19 cases/ million population), accounting for random variation among countries. Results: Two thousand forty-five academics from 26 countries participated in the survey (response rate= 14.3%, with 54.7% female and 67% younger than 46 years of age). The mean (SD) knowledge percent score was 73.2% (11.2), and the knowledge of symptoms score was significantly lower than the diagnostic methods score (53.1% and 85.4%). Knowledge was significantly higher among those living with partner/spouse than those living alone (regression coefficient (B)= 0.48); those with PhD than those with BDS (B= 0.48), those seeing 21 to 30 patients daily than those seeing no patients (B= 0.65) and those from countries with a higher number of COVID-19 cases/million population (B= 0.0007). Conclusions: Dental academics had poorer knowledge of COVID-19 symptoms than diagnostic methods. Living arrangements, academic degrees, patient load, and magnitude of epidemic in the country were associated with COVD-19 knowledge among dental academics. COVID-19 training can be designed using these factors to target academics with the greatest need.

Trust in government and its associations with health behaviour and prosocial behaviour during the COVID-19 pandemic

Previous studies suggested that public trust in government is vital for implementations of social policies that rely on public's behavioural responses. This study examined associations of trust in government regarding COVID-19 control with recommended health behaviours and prosocial behaviours. Data from an international survey with representative samples (N=23,733) of 23 countries were analysed. Specification curve analysis showed that higher trust in government was significantly associated with higher adoption of health and prosocial behaviours in all reasonable specifications of multilevel linear models (median standardised β=0.173 and 0.244, P<0.001). We further used structural equation modelling to explore potential determinants of trust in government regarding pandemic control. Governments perceived as well organised, disseminating clear messages and knowledge on COVID-19, and perceived fairness were positively associated with trust in government (standardised β=0.358, 0.230, 0.055, and 0.250, P<0.01). These results highlighted the importance of trust in government in the control of COVID-19.