Social diversity and social cohesion in Britain

We use data from a large-scale and nationally representative survey to examine whether there is in Britain a trade-off between social diversity and social cohesion. Using six separate measures of social cohesion (generalised trust, volunteering, giving to charity, inter-ethnic friendship, and two neighbourhood cohesion scales) and four measures of social diversity (ethnic fractionalisation, religious fractionalisation, percentage Muslim, and percentage foreign-born), we show that, net of individual covariates, there is a negative association between social diversity and most measures of social cohesion. But these associations largely disappear when neighbourhood deprivation is taken into account. These results are robust to alternative definitions of neighbourhood. We also investigate the possibility that the diversity--cohesion trade-off is found in more segregated neighbourhoods. But we find very little evidence to support that claim. Overall, it is material deprivation, not diversity, that undermines social cohesion.

Detecting Cohort Effects in Accelerated Longitudinal Designs Using Multilevel Models

Accelerated longitudinal designs allow researchers to efficiently collect longitudinal data covering a time span much longer than the study duration. One important assumption of these designs is that each cohort (a group defined by their age of entry into the study) shares the same longitudinal trajectory. Although previous research has examined the impact of violating this assumption when each cohort is defined by a single age of entry, it is possible that each cohort is instead defined by a range of ages, such as groups that experience a particular historical event. In this paper we examined how including cohort membership in linear and quadratic multilevel models performed in detecting and controlling for cohort effects in this scenario. Using a Monte Carlo simulation study, we assessed the performance of this approach under conditions related to the number of cohorts, the overlap between cohorts, the strength of the cohort effect, the number of affected parameters, and the sample size. Our results indicate that models including a proxy variable for cohort membership based on age at study entry performed comparably to using true cohort membership in detecting cohort effects accurately and returning unbiased parameter estimates. This indicates that researchers can control for cohort effects even when true cohort membership is unknown.

Workers' emotional exhaustion and mental well-being over the COVID-19 pandemic: a Dynamic Structural Equation Modeling (DSEM) approach

The COVID-19 pandemic has presented significant challenges to the workforce, particularly concerning emotional and mental well-being. Given the prolonged periods of work-related stress, unexpected organizational changes, and uncertainties about work faced during the pandemic, it becomes imperative to study occupational health constructs under a dynamic methodological perspective, to understand their stable and unstable characteristics better. In this study, drawing on the Dynamic Structural Equation Modeling (DSEM) framework, we used a combination of multilevel AR(1) models, Residual-DSEM (RDSEM), multilevel bivariate VAR(1) models, and multilevel location-scale models to investigate the autoregression, trend, and (residual) cross-lagged relationships between emotional exhaustion (EmEx) and mental well-being (MWB) over the COVID-19 pandemic. Data were collected weekly on 533 workers from Germany (91.18%) and Italy (8.82%) who completed a self-reported battery (total number of observations = 3,946). Consistent with our hypotheses, results were as follows: (a) regarding autoregression, the autoregressive component for both EmEx and MWB was positive and significant, as well as it was their associated between-level variability; (b) regarding trend, over time EmEx significantly increased, while MWB significantly declined, furthermore both changes had a significant between-level variability; (c) regarding the longitudinal bivariate (cross-lagged) relationships, EmEx and MWB negatively and significantly affected each other from week to week, furthermore both cross-lagged relationships showed to have significant between-level variance. Overall, our study pointed attention to the vicious cycle between EmEx and MWB, even after controlling for their autoregressive component and trend, and supported the utility of DSEM in occupational health psychology studies.

Do regional labor market conditions matter for temporary work disability duration? A multilevel analysis of workers' compensation claims in five Canadian provinces

BACKGROUND

Research has shown how regionally varying labor market conditions are associated with differences in work disability duration. However, the majority of these studies have not used multilevel models to appropriately account for the hierarchical clustering of individuals nested within contextual units (e.g., regions). Studies that have used multilevel models have focused on privately insured workers or on disability not specifically caused by work-related injury or illness.

METHODS

Using claims data from five Canadian provincial workers' compensation systems, linear random-intercept models were used to estimate how much variance in temporary work disability duration ('work disability duration' for brevity) for work-related injuries and musculoskeletal disorders was due to differences between economic regions, what economic region-level labor market characteristics were associated with work disability duration, and what characteristics best explained economic region differences in work disability duration.

RESULTS

Economic region characteristics, such as the unemployment rate and proportion of goods-producing employment, were independently associated with individual-level work disability duration. However, economic region variation only accounted for 1.5%-2% of total variation in work disability duration. The majority (71%) of economic region-level variation was explained by the provincial jurisdiction where the worker lived and was injured. Regional variation tended to be greater for female workers than males.

CONCLUSIONS

The findings suggest that while regional labor market conditions matter for work disability duration, system-level differences in workers' compensation and health care are more important factors influencing work disability duration. Furthermore, while this study includes both temporary and permanent disability claims, the work disability duration measure only captures temporary disability.

Developmental trajectories of anger and sadness dysregulation in childhood differentially predict later borderline symptoms

Difficulties with emotion regulation are integral to borderline personality disorder (BPD) and its hypothesized developmental pathway. Here, we prospectively assess trajectories of emotion processing across childhood, how BPD symptoms impact these trajectories, and whether developmental changes are transdiagnostic or specific to BPD, as major depressive (MDD) and conduct disorders (CD) are also characterized by emotion regulation difficulties. This study included 187 children enriched for those with early symptoms of depression and disruptive behaviors from a longitudinal study. We created multilevel models of multiple components of emotional processing from mean ages 9.05 to 18.55 years, and assessed the effect of late adolescent BPD, MDD, and CD symptoms on these trajectories. Linear trajectories of coping with sadness and anger, and quadratic trajectories of dysregulated expressions of sadness and anger were transdiagnostic, but also exhibited independent relationships with BPD symptoms. Only inhibition of sadness was related to BPD symptoms. The quadratic trajectories of poor emotional awareness and emotional reluctance were also independently related to BPD. Findings support examining separable components of emotion processing across development as potential precursors to BPD, underscoring the importance of understanding these trajectories as not only a marker of potential risk but also potential targets for prevention and intervention.

Do positive memory characteristics influence daily-level trajectories of posttraumatic stress disorder symptoms? an exploratory daily diary study

BACKGROUND AND OBJECTIVES

Evidence links posttraumatic stress disorder (PTSD) with positive memory characteristics. To extend this research, we utilized daily diary data to examine (1) concurrent/lagged associations between daily PTSD symptom severity and positive memory vividness/accessibility; and (2) associations between baseline-assessed positive memory characteristics and changes in PTSD symptom severity over time.

DESIGN AND METHODS

A sample of 238 trauma-exposed participants (Mage = 21.19 years; 86% women) completed baseline and 10 daily measures of PTSD symptoms and positive memory characteristics. Multilevel models covaried for gender, number of trauma types, and number of completed surveys.

RESULTS

Days with greater PTSD symptom severity than an individual's average associated with less vividness (b = -0.02, p < .001) and accessibility (b = -0.02, p < .001) of the positive memory on the same day. Days with greater positive memory vividness (b = -1.06, p < .001) and accessibility (b = -0.93, p < .001) than an individual's average associated with less PTSD symptom severity on the same day. There were no significant lagged associations between these constructs. There were significant interactions between baseline-assessed psychological distance and time (b = -0.04, p  = .042) and between baseline-assessed visual perspective and time (b = 0.05, p = .023) on PTSD symptom severity across days.

CONCLUSIONS

Findings inform positive memory intervention targets for PTSD and provide impetus for longitudinal investigations on their inter-relations.

Multilevel SEM with random slopes in discrete data using the pairwise maximum likelihood

Pairwise maximum likelihood (PML) estimation is a promising method for multilevel models with discrete responses. Multilevel models take into account that units within a cluster tend to be more alike than units from different clusters. The pairwise likelihood is then obtained as the product of bivariate likelihoods for all within-cluster pairs of units and items. In this study, we investigate the PML estimation method with computationally intensive multilevel random intercept and random slope structural equation models (SEM) in discrete data. In pursuing this, we first reconsidered the general 'wide format' (WF) approach for SEM models and then extend the WF approach with random slopes. In a small simulation study we the determine accuracy and efficiency of the PML estimation method by varying the sample size (250, 500, 1000, 2000), response scales (two-point, four-point), and data-generating model (mediation model with three random slopes, factor model with one and two random slopes). Overall, results show that the PML estimation method is capable of estimating computationally intensive random intercept and random slopes multilevel models in the SEM framework with discrete data and many (six or more) latent variables with satisfactory accuracy and efficiency. However, the condition with 250 clusters combined with a two-point response scale shows more bias.

[Age-specific regional characteristics of COVID-19 mortality in 2021]

INTRODUCTION

In most countries, COVID-19 mortality increases exponentially with age, but the growth rate varies considerably between countries. The different progression of mortality may reflect differences in population health, the quality of health care or coding practices.

OBJECTIVE

In this study, we investigated differences in age-specific county characteristics of COVID-19 mortality in the second year of the pandemic.

METHOD

Age-specific patterns of COVID-19 adult mortality were estimated according to county level and sex using a Gompertz function with multilevel models.

RESULTS

The Gompertz function is suitable for describing age patterns of COVID-19 adult mortality at county level. We did not find significant differences in the age progression of mortality between counties, but there were significant spatial differences in the level of mortality. The mortality level showed a relationship with socioeconomic and health care indicators with the expected sign, but with different strengths.

DISCUSSION AND CONCLUSION

The COVID-19 pandemic in 2021 resulted in a decline in life expectancy in Hungary not seen since World War II. The study highlights the importance of healthcare in addition to social vulnerability. It also points out that understanding age patterns will help to mitigate the consequences of the epidemic. Orv Hetil. 2023; 164(17): 643-650.

Physical activity, sleep and affective wellbeing on the following day: An experience sampling study

This study aims to investigate the temporal links between physical activity, sleep and affective wellbeing in young adults. In particular, the aim was to examine whether physical activity is associated with sleep indicators in subsequent nights and, in turn, whether sleep was associated with improved affective wellbeing the next morning. Relatedly, moderation by baseline levels of depressive symptoms, sleep quality, habitual physical activity and gender was analysed. One-hundred and forty-seven individuals (85.7% female) aged 18-25 years old participated in an experience sampling study over 14 consecutive days. Participants received seven prompts per day, and answered questions about their physical activity and affective states. Every morning, participants also reported their sleep. Physical activity throughout the day was not related to sleep during the following night or to affective wellbeing the next morning. An exception to that pattern was that physical activity before 14:00 hours was associated with longer subsequent sleep duration. Better subjective sleep quality predicted affective wellbeing the next morning. Associations of physical activity, sleep and affective wellbeing were not moderated by baseline depressive symptoms, sleep quality or habitual physical activity. However, investigation of gender as a moderator revealed that moderate physical activity was associated with better subsequent sleep quality for males, but not for females. Overall, we found that physical activity is associated with better subsequent sleep for males, but not for females. Also, our study provides further evidence that better sleep quality is associated with the next morning's affective wellbeing.

The effect of the rate of increase of estrogen replacement therapy on bone mineral density accrual in young patients with Turner syndrome

BACKGROUND

Turner syndrome (TS) is caused by partial/complete X-chromosome monosomy with variable phenotypes, characterized by hypogonadism and short stature. To achieve pubertal changes, up to 50-79% of patients with TS require estrogen replacement therapy (ERT) and 80% have low bone mineral density (BMD). Studies show that pubertal delays are associated with decreased BMD. Currently, guidelines suggest ERT start at 12 years, increasing slowly, simulating pubertal progression. Many studies show that ERT increases BMD in TS adolescents, but uncertainty remains as to how the rate of increase in ERT affects BMD.

METHODS

IRB approval was obtained from our institution for this retrospective chart review from 1991-2020. Charts were requested for the database using ICD 9-10 codes for TS and patients undergoing dual-energy X-ray absorptiometry (DEXA). Biometric data, medical and treatment histories were extracted from charts. Multilevel random effects models were constructed to assess the time dependent associations between ERT and bone density parameters. The primary independent variable of interest was the rate at which patients went from initiating ERT to reaching final doses. The primary dependent variables measured were total body BMD (tbBMD) and corresponding z-scores, calculated using DEXA techniques. Analyses were done with SAS software (version 9.4, Cary, NC).

RESULTS

28 patients met inclusion criteria. Mean age of TS diagnosis was 6.9 years; 8 patients had monosomy X, 16 had mosaic karyotypes, 4 had unknown karyotypes. The average age for starting HRT was 14.1 years. 13 patients had spontaneous pubertal onset before starting HRT. tbBMD increased significantly with age (p = 0.03). However, change in BMD by age does not vary between patients who reached final adult doses of ERT within 0-2.5 years, compared to patients who took 2.5-5.5 years (p=0.7). Patients who took 2.5-5.5 years to reach final adult doses of ERT had a more negative trend in z-scores (-2.144) in comparison to patients who took 0-2.5 years (-1.776), although this difference did not reach statistical significance (p=.15). Future larger studies are needed to better understand the relationship between duration of ERT use and tbBMD.

CONCLUSIONS

BMD in TS adolescents increases with age. Neither absolute tbBMD values nor tbBMD z-scores increased faster when ERT doses were maximized within 2.5 years. This study has identified a cohort of children under 12 years with TS who have not had any ERT or BMD measurements - a potential population for future larger prospective studies.

Fast Multilevel Functional Principal Component Analysis

We introduce fast multilevel functional principal component analysis (fast MFPCA), which scales up to high dimensional functional data measured at multiple visits. The new approach is orders of magnitude faster than and achieves comparable estimation accuracy with the original MFPCA (Di et al., 2009). Methods are motivated by the National Health and Nutritional Examination Survey (NHANES), which contains minute-level physical activity information of more than 10000 participants over multiple days and 1440 observations per day. While MFPCA takes more than five days to analyze these data, fast MFPCA takes less than five minutes. A theoretical study of the proposed method is also provided. The associated function mfpca.face() is available in the R package refund.

A Multilevel Analysis of the Associated and Determining Factors of TB among Adults in South Africa: Results from National Income Dynamics Surveys 2008 to 2017

TB is preventable and treatable but remains the leading cause of death in South Africa. The deaths due to TB have declined, but in 2017, around 322,000 new cases were reported in the country. The need to eradicate the disease through research is increasing. This study used population-based National Income Dynamics Survey data (Wave 1 to Wave 5) from 2008 to 2017. By determining the simultaneous multilevel and individual-level predictors of TB, this research examined the factors associated with TB-diagnosed individuals and to what extent the factors vary across such individuals belonging to the same province in South Africa for the five waves. Multilevel logistic regression models were fitted using frequentist and Bayesian techniques, and the results were presented as odds ratios with statistical significance set at p < 0.05. The results obtained from the two approaches were compared and discussed. Findings reveal that the TB factors that prevailed consistently from wave 1 to wave 5 were marital status, age, gender, education, smoking, suffering from other diseases, and consultation with a health practitioner. Also, over the years, the single males aged 30-44 years suffering from other diseases with no education were highly associated with TB between 2008 and 2017. The methodological findings were that the frequentist and Bayesian models resulted in the same TB factors. Both models showed that some form of variation in TB infections is due to the different provinces these individuals belonged. Variation in TB patients within the same province over the waves was minimal. We conclude that demographic and behavioural factors also drive TB infections in South Africa. This research supports the existing findings that controlling the social determinants of health will help eradicate TB.

Subgrain Coalescence Simulation by Means of an Advanced Statistical Model of Inelastic Deformation

The development of technological methods for processing and manufacturing of functional (with a priori targeted properties) polycrystalline materials and products made of these materials still remains an acute problem. A multilevel modeling approach offers researchers the opportunity to describe inelastic deformation by applying internal variables that give an effective characterization of the material structure at different structural scale levels. High temperature plastic deformation is accompanied by these processes, which leads to a significant rearrangement of the meso- and microstructure of the material. The most substantial contribution to changing the properties of polycrystals is made by the evolution of grain and defect structures at the expense of dynamic recrystallization, which significantly depends on dynamic recovery. In this paper, we consider the problem of the coalescence of subgrains undergoing rotation during inelastic hot deformation. This process is called subgrain coalescence, and it is one of the dynamic recovery mechanisms responsible for changes in the fine subgrain structure. Under applied thermomechanical loads, the coalescence process promotes the formation of recrystallization nuclei and their subsequent growth, which can greatly change the grain structure of a polycrystal. The problem was solved in terms of the advanced statistical model of inelastic deformation, modified to describe the subgrain coalescence process. The model takes into account the local interactions between contacting structural elements (subgrains). These have to be considered so that the grain coalescence caused by a decrease in subboundary energies during their progressive merging can be adequately analyzed. For this purpose, a subgrain structure quite similar to the real structure was modeled using Laguerre polyhedra. Subgrain rotations were investigated using the developed model, which relies on the consideration of the excess density edge component of the same sign dislocations on incidental subgrain boundaries. The results of modeling of a copper polycrystal are presented, and the effects of temperature and strain rate on the subgrain coalescence process is demonstrated.

Fixed or random? On the reliability of mixed-effects models for a small number of levels in grouping variables

Biological data are often intrinsically hierarchical (e.g., species from different genera, plants within different mountain regions), which made mixed‐effects models a common analysis tool in ecology and evolution because they can account for the non‐independence. Many questions around their practical applications are solved but one is still debated: Should we treat a grouping variable with a low number of levels as a random or fixed effect? In such situations, the variance estimate of the random effect can be imprecise, but it is unknown if this affects statistical power and type I error rates of the fixed effects of interest. Here, we analyzed the consequences of treating a grouping variable with 2–8 levels as fixed or random effect in correctly specified and alternative models (under‐ or overparametrized models). We calculated type I error rates and statistical power for all‐model specifications and quantified the influences of study design on these quantities. We found no influence of model choice on type I error rate and power on the population‐level effect (slope) for random intercept‐only models. However, with varying intercepts and slopes in the data‐generating process, using a random slope and intercept model, and switching to a fixed‐effects model, in case of a singular fit, avoids overconfidence in the results. Additionally, the number and difference between levels strongly influences power and type I error. We conclude that inferring the correct random‐effect structure is of great importance to obtain correct type I error rates. We encourage to start with a mixed‐effects model independent of the number of levels in the grouping variable and switch to a fixed‐effects model only in case of a singular fit. With these recommendations, we allow for more informative choices about study design and data analysis and make ecological inference with mixed‐effects models more robust for small number of levels.

Accounting for Heteroskedasticity Resulting from Between-Group Differences in Multilevel Models

Homogeneity of variance (HOV) is a well-known but often untested assumption in the context of multilevel models (MLMs). However, depending on how large the violation is, how different group sizes are, and the variance pairing, standard errors can be over or underestimated even when using MLMs, resulting in questionable inferential tests. We evaluate several tests (e.g., the H statistic, Breusch Pagan, Levene's test) that can be used with MLMs to assess violations of HOV. Although the traditional robust standard errors used with MLMs require at least 50 clusters to be effective, we assess a robust standard error adjustment (i.e., the CR2 estimator) that can be used even with a few clusters. Findings are assessed using a Monte Carlo simulation and are further illustrated using an applied example. We show that explicitly modeling the heterogenous variance structures or using the CR2 estimator are both effective at ameliorating the issues associated with the fixed effects of the regression model related to violations of HOV resulting from between-group differences.

Risks and Opportunities to Ensure Equity in the Application of Big Data Research in Public Health

The big data revolution presents an exciting frontier to expand public health research, broadening the scope of research and increasing the precision of answers. Despite these advances, scientists must be vigilant against also advancing potential harms toward marginalized communities. In this review, we provide examples in which big data applications have (unintentionally) perpetuated discriminatory practices, while also highlighting opportunities for big data applications to advance equity in public health. Here, big data is framed in the context of the five Vs (volume, velocity, veracity, variety, and value), and we propose a sixth V, virtuosity, which incorporates equity and justice frameworks. Analytic approaches to improving equity are presented using social computational big data, fairness in machine learning algorithms, medical claims data, and data augmentation as illustrations. Throughout, we emphasize the biasing influence of data absenteeism and positionality and conclude with recommendations for incorporating an equity lens into big data research.

The Importance of Random Slopes in Mixed Models for Bayesian Hypothesis Testing

Mixed models are gaining popularity in psychology. For frequentist mixed models, previous research showed that excluding random slopes-differences between individuals in the direction and size of an effect-from a model when they are in the data can lead to a substantial increase in false-positive conclusions in null-hypothesis tests. Here, I demonstrated through five simulations that the same is true for Bayesian hypothesis testing with mixed models, which often yield Bayes factors reflecting very strong evidence for a mean effect on the population level even if there was no such effect. Including random slopes in the model largely eliminates the risk of strong false positives but reduces the chance of obtaining strong evidence for true effects. I recommend starting analysis by testing the support for random slopes in the data and removing them from the models only if there is clear evidence against them.

Individual differences in postoperative recovery trajectories for adult symptomatic lumbar scoliosis

OBJECTIVE

The Adult Symptomatic Lumbar Scoliosis-1 (ASLS-1) trial demonstrated the benefit of adult symptomatic lumbar scoliosis (ASLS) surgery. However, the extent to which individuals differ in their postoperative recovery trajectories is unknown. This study's objective was to evaluate variability in and factors moderating recovery trajectories after ASLS surgery.

METHODS

The authors used longitudinal, multilevel models to analyze postoperative recovery trajectories following ASLS surgery. Study outcomes included the Oswestry Disability Index (ODI) score and Scoliosis Research Society-22 (SRS-22) subscore, which were measured every 3 months until 2 years postoperatively. The authors evaluated the influence of preoperative disability level, along with other potential trajectory moderators, including radiographic, comorbidity, pain/function, demographic, and surgical factors. The impact of different parameters was measured using the R2, which represented the amount of variability in ODI/SRS-22 explained by each model. The R2 ranged from 0 (no variability explained) to 1 (100% of variability explained).

RESULTS

Among 178 patients, there was substantial variability in recovery trajectories. Applying the average trajectory to each patient explained only 15% of the variability in ODI and 21% of the variability in SRS-22 subscore. Differences in preoperative disability (ODI/SRS-22) had the strongest influence on recovery trajectories, with patients having moderate disability experiencing the greatest and most rapid improvement after surgery. Reflecting this impact, accounting for the preoperative ODI/SRS-22 level explained an additional 56%-57% of variability in recovery trajectory, while differences in the rate of postoperative change explained another 7%-9%. Among the effect moderators tested, pain/function variables-such as visual analog scale back pain score-had the biggest impact, explaining 21%-25% of variability in trajectories. Radiographic parameters were the least influential, explaining only 3%-6% more variance than models with time alone. The authors identified several significant trajectory moderators in the final model, such as significant adverse events and the number of levels fused.

CONCLUSIONS

ASLS patients have highly variable postoperative recovery trajectories, although most reach steady state at 12 months. Preoperative disability was the most important influence, although other factors, such as number of levels fused, also impacted recovery.

Streamlined variational inference for higher level group-specific curve models

A two-level group-specific curve model is such that the mean response of each member of a group is a separate smooth function of a predictor of interest. The three-level extension is such that one grouping variable is nested within another one, and higher level extensions are analogous. Streamlined variational inference for higher level group-specific curve models is a challenging problem. We confront it by systematically working through two-level and then three-level cases and making use of the higher level sparse matrix infrastructure laid down in Nolan and Wand (2019). A motivation is analysis of data from ultrasound technology for which three-level group-specific curve models are appropriate. Whilst extension to the number of levels exceeding three is not covered explicitly, the pattern established by our systematic approach sheds light on what is required for even higher level group-specific curve models.

Innovative moments in recovered cases treated with the unified protocol for transdiagnostic treatment of emotional disorders

OBJECTIVE

Innovative moments (IMs) are moments in which the previous problematic pattern of meaning is challenged. Studies have shown that IMs are associated with good psychotherapy outcomes. A three-level hierarchy of IMs was observed in recent studies, with level 1 IMs being more elementary and levels 2 and 3 being more complex and associated with treatment success. However, studies with manualized protocol treatments are thus far lacking. This study analyzed the longitudinal progression of IMs in the Unified Protocol for Transdiagnostic Treatment of Emotional Disorders (UP) and explored its associations with changes in psychological distress.

METHODS

Data were collected from a Portuguese university-based outpatient clinic and included 18 cases with positive outcomes. Nine sessions of each case were coded with the IM coding system (N=162).

RESULTS

Multilevel analyses showed a significant increase in all IM levels across treatments. The decrease in psychological distress predicted an increase in level 2 IMs in the same session.

CONCLUSION

The evolution of IMs is similar to what was found previously in other studies. Contrary to what was found in previous studies, IMs did not predict outcomes in the following session, whereas the reduction in psychological distress predicted the emergence of level 2 IMs.

Occupational hazard perceptions as factors influencing students' intentions to engage in part-time jobs

Purpose. The purpose of this study was to investigate the joined effects of students' perceived importance of occupational hazards and financial gains on intentions to engage in part-time jobs. The role of students' maturity value levels was also investigated. We used conservation of resources theory (COR) and job demands-resources (JD-R) theory as theoretical frameworks to test our hypotheses. Method. We conducted a factorial survey experiment using a sample of 858 students. Multilevel regression analyses with random effects were used to investigate effects of hazards perceptions and financial gains on intentions to engage in part-time jobs. The direct and moderating effects of students' maturity values were investigated. Results. Students considered financial gains more important in developing their intentions to take a part-time job, compared to occupational hazards. We also found interactions between occupational hazards and between occupational hazards and financial gains in predicting intentions. Students' maturity scores were negatively related to intentions and moderated the effects of financial gains and perceptions of psychological hazards on intentions. Conclusion. The study provided insights into students' perceptions of occupational hazards and financial benefits and their intention to take a part-time job. The results also indicate the direct and moderating effects of students' maturity values.

The Rise, Demise, and Reprise of the Increasingly Protracted APA Journal Article?

Are APA journal articles getting longer or shorter over time? Earlier work that examined changes over time in article length in 24 APA journals (1986-2005) found that pages per article increased over time for the average journal, peaking around 2001, and then plateaued thereafter. But have these trends continued during the past 14 years? The current research extends prior work by adding additional years (1986-2019) and four additional journals (28 journals, 34 years, and 865 total observations). Multilevel growth curve analyses revealed a cubic effect of time on average article length, showing an increase in the 1980s and 1990s, a plateau or slight decline in the 2000s, and a slight increase again in the 2010s. Journal impact factors (JIFs) moderated linear growth over time; journals with higher JIFs had larger linear increases in article length. Exploratory multilevel interrupted time-series analyses suggested that the average linear increase in pages per article over time was greater after the start of psychology's credibility crisis (2012-2019) than before it (1986-2011), which may relate to an increased emphasis on reporting details and transparency. We discuss implications for article length in the contexts of publishing and psychology's ongoing credibility crisis.

Egocentric network characteristics of persons with Type 1 diabetes and their relationships to perceived social support and well-being

Objectives: The size of one's support network is positively related to health and well-being. It is therefore important to understand this association in people with Type 1 diabetes, as this could inform interventions. Moreover, the type of support (emotional, instrumental, informational) offered likely varies by gender of both the person seeking support and offering support. We thus examine the relationship between the composition of (perceived) social support networks and well-being in a sample of 121 persons with Type 1 diabetes. Design: An egocentric social network survey, combined with survey measures. Main outcome(s): The size and composition of support networks and well-being. Measures: Participants indicated the type of support individuals in their contact network offered and their gender, alongside measures of perceived social support and well-being. They indicated which individuals offered which types of support (emotional, instrumental, informational). Results: Perceived support was associated with the actual size of the emotional support network. Further, the size of the emotional support network was associated with well-being. Using multilevel models we examined assortment by gender in social support networks. Compared to women, men were more inclined to list the opposite gender as support, especially for emotional and informational support. Conclusion: Mapping out an individual's multidimensional support network paints a more complete picture of support than single item measures of support. We therefore recommend relying on a social network methodology to gain a more complete understanding of support networks. The findings highlight that an association exists between emotional network size and wellbeing. Given the potential implications of this finding for the quality of life of diabetes patients, it is important to establish the causality of this relationship.

Association of adiposity with diabetes: A national research among Chinese adults

BACKGROUND

Adiposity is an established risk factor for diabetes. The different measurements of adiposity for predicting diabetes have been compared in recent studies in Western countries. However, similar researches among Chinese adults are limited.

METHODS

Data were collected from a national survey conducted during September 2014 and May 2015 Among Chinese adults aged 40 years and older across 30 China's provinces. Multilevel model analysis was performed to examine the impacts of different obesity indices [body mass index (BMI), waist circumference (WC), lipid accumulation product index (LAP), visceral adiposity index (VAI), and body adiposity index (BAI)] on the risk of diabetes.

RESULTS

A total of 162 880 participants were included in this study. Of them, 54.47% were female. With an increase in BMI, WC, LAP, VAI, and BAI, the prevalence of diabetes significantly grew (P < 0.001). The multilevel model analysis showed that WC has the strongest impact on diabetes prevalence, while BAI was the weakest. For one SD increment in BMI, WC, LAP, VAI, and BAI, the prevalence of diabetes increased by 27.0% (Odds Ratio (OR) = 1.270, 95% Confidence interval (CI) = 1.251-1.289), 37.4% (OR = 1.374, 95% CI = 1.346-1.401), 28.1% (OR = 1.281, 95% CI = 1.266-1.297), 22.0% (OR = 1.220, 95% CI = 1.204-1.236), and 17.4% (OR = 1.174, 95% CI = 1.151-1.192), respectively.

CONCLUSION

Obesity indicators of BMI, WC, LAP, VAI, and BAI have significant positive relationships with the risk of diabetes. WC has the strongest impact on diabetes, while BAI has the weakest.

Power and Sample Size Determination for Multilevel Mediation in Three-Level Cluster-Randomized Trials

Mediation analyses supply a principal lens to probe the pathways through which a treatment acts upon an outcome because they can dismantle and test the core components of treatments and test how these components function as a coordinated system or theory of action. Experimental evaluation of mediation effects in addition to total effects has become increasingly common but literature has developed only limited guidance on how to plan mediation studies with multi-tiered hierarchical or clustered structures. In this study, we provide methods for computing the power to detect mediation effects in three-level cluster-randomized designs that examine individual- (level one), intermediate- (level two) or cluster-level (level three) mediators. We assess the methods using a simulation and provide examples of a three-level clinic-randomized study (individuals nested within therapists nested within clinics) probing an individual-, intermediate- or cluster-level mediator using the R package PowerUpR and its Shiny application.

Identification of Vulnerable Populations and Areas at Higher Risk of COVID-19-Related Mortality during the Early Stage of the Epidemic in the United States

We characterized vulnerable populations located in areas at higher risk of COVID-19-related mortality and low critical healthcare capacity during the early stage of the epidemic in the United States. We analyze data obtained from a Johns Hopkins University COVID-19 database to assess the county-level spatial variation of COVID-19-related mortality risk during the early stage of the epidemic in relation to health determinants and health infrastructure. Overall, we identified highly populated and polluted areas, regional air hub areas, race minorities (non-white population), and Hispanic or Latino population with an increased risk of COVID-19-related death during the first phase of the epidemic. The 10 highest COVID-19 mortality risk areas in highly populated counties had on average a lower proportion of white population (48.0%) and higher proportions of black population (18.7%) and other races (33.3%) compared to the national averages of 83.0%, 9.1%, and 7.9%, respectively. The Hispanic and Latino population proportion was higher in these 10 counties (29.3%, compared to the national average of 9.3%). Counties with major air hubs had a 31% increase in mortality risk compared to counties with no airport connectivity. Sixty-eight percent of the counties with high COVID-19-related mortality risk also had lower critical care capacity than the national average. The disparity in health and environmental risk factors might have exacerbated the COVID-19-related mortality risk in vulnerable groups during the early stage of the epidemic.

The Importance of Contextualized Psychosocial Risk Indicators in Workplace Stress Assessment: Evidence from the Healthcare Sector

The routine assessment of workplace stress is mostly based on standardized self-report tools, including generic psychosocial risk indicators (G-PRIs) designed to fit very heterogeneous occupational sectors. However, the use "by default" of such indicators might be inadequate when they fail to characterize the specificity of the work environment; hence, the inclusion of more contextualized indicators (C-PRIs) has been recommended. We aimed at evaluating the additional contribution of three C-PRIs (Work-Family Conflict, Emotional Demands, and Excessive Demands from Patients) in predicting individual outcomes (Emotional Exhaustion, Turnover Intentions) compared to commonly used G-PRIs (e.g., Demand, Control, Support), in a sample of 787 healthcare workers involved in a routine workplace stress assessment. Multilevel hierarchical regression supported the additional contributions of C-PRIs in predicting both outcomes over G-PRIs, sex, age and shift work. More robust and consistent evidence emerged for Emotional Exhaustion, which was significantly predicted by all C-PRIs, whereas Turnover Intentions was only predicted by the C-PRI Emotional Demands. Importantly, not all G-PRIs showed a relationship with the two outcomes. Taken together, our results support the importance of including C-PRIs in workplace stress assessment carried out by organizations, which should be selected based on literature search and discussion with the stakeholders.

State-Level Alcohol Environments and Sexual Identity Disparities in Binge Drinking in the Behavioral Risk Factor Surveillance System

Purpose: Binge drinking disparities by sexual identity are well documented. Stronger alcohol policy environments reduce binge drinking in the general population. We examined whether state-level alcohol policy environments have the same association with binge drinking among lesbian, gay, and bisexual (LGB) adults as among heterosexual adults. Methods: Binge drinking, sexual identity, and demographic characteristics were extracted from the 2015 to 2018 Behavioral Risk Factor Surveillance System. The strength of the alcohol policy environment was measured by using the Alcohol Policy Scale (APS) score. We estimated the association between APS score and binge drinking by using logistic regression and included an interaction term between APS score and sexual identity. Results: The interaction between APS score and sexual identity was not significant, and findings differed between women and men. Among women, a higher APS score was associated with lower odds of binge drinking (adjusted odds ratio [aOR]: 0.96, 95% confidence interval [CI]: 0.94-0.99). Differences in binge drinking by sexual identity remained after adjusting for individual and state-level factors (e.g., the percentage of LGB adults in the state). Compared with heterosexual women, the odds of binge drinking were 43% higher (aOR: 1.43, 95% CI: 1.17-1.75) among lesbian women and 58% higher (aOR: 1.58, 95% CI: 1.40-1.79) among bisexual women. A higher APS score was not associated with binge drinking among men. Conclusion: Stronger state-level alcohol policy environments were associated with lower binge drinking among women. Lesbian and bisexual women were still more likely to engage in binge drinking compared with heterosexual women even in states with stronger alcohol policy environments.

Suicide as an interpersonal phenomenon: Dyadic methodological and statistical considerations in suicide research

Interpersonal risk and resilience factors are prominent in current conceptual models of suicide. A growing body of empirical evidence links suicidal thoughts and behaviors to a range of interpersonal phenomenon adding further support to the value of this line of inquiry. At present, research on interpersonal phenomenon focuses on assessing individuals' perceptions of interpersonal phenomenon, such as appraisals of burdensomeness, experienced loneliness, and thwarted belongingness. As this line of research continues to develop, we argue that it would be valuable to consider incorporating conceptual models of interpersonal phenomenon and corresponding methodological approaches from closely allied fields. After providing a brief overview of interpersonal models of suicide, we present an introduction to conceptual models of interpersonal phenomenon developed in relationship science, describe how these models can be applied to the study of interpersonal phenomenon in suicide research, and close with a guided tutorial on data collection and statistical analysis methods for testing hypotheses derived from these conceptual approaches. References for additional reading are provided, and the Appendix S1 provides simulated data sets and statistical code for the analyses in the tutorial section.

Facial and Vocal Expressions During Clinical Interviews Suggest an Emotional Modulation Paradox in Borderline Personality Disorder: An Explorative Study

Videotape recordings obtained during an initial and conventional psychiatric interview were used to assess possible emotional differences in facial expressions and acoustic parameters of the voice between Borderline Personality Disorder (BPD) female patients and matched controls. The incidence of seven basic emotion expressions, emotional valence, heart rate, and vocal frequency (f0), and intensity (dB) of the discourse adjectives and interjections were determined through the application of computational software to the visual (FaceReader) and sound (PRAAT) tracks of the videotape recordings. The extensive data obtained were analyzed by three statistical strategies: linear multilevel modeling, correlation matrices, and exploratory network analysis. In comparison with healthy controls, BPD patients express a third less sadness and show a higher number of positive correlations (14 vs. 8) and a cluster of related nodes among the prosodic parameters and the facial expressions of anger, disgust, and contempt. In contrast, control subjects showed negative or null correlations between such facial expressions and prosodic parameters. It seems feasible that BPD patients restrain the facial expression of specific emotions in an attempt to achieve social acceptance. Moreover, the confluence of prosodic and facial expressions of negative emotions reflects a sympathetic activation which is opposed to the social engagement system. Such BPD imbalance reflects an emotional alteration and a dysfunctional behavioral strategy that may constitute a useful biobehavioral indicator of the severity and clinical course of the disorder. This face/voice/heart rate emotional expression assessment (EMEX) may be used in the search for reliable biobehavioral correlates of other psychopathological conditions.

A Revision of the Current Software and Usage Examples for Researchers

Although modern lines for dealing with missing data are well established from the 1970s, today there is a challenge when researchers encounter this problem in multilevel models. First, there is a variety of existing software to handle missing data based on multiple imputation (MI), currently pointed out by experts as the most promising strategy. Second, the two principal paradigms of MI are joint modelling (JM) and fully conditional specification (FCS), one more complication because they are not equally useful depending on the combination of multilevel model and the estimated parameters affected by missing data. Technical literature do not contribute to ease the number of decisions that researcher has to do. Given these inconveniences, the present paper has three objectives. (1) To present a thorough revision of the most recently developed software and functions about multiple imputation in multilevel models. (2) We derive a set of suggestions, recommendations, and guides for helping researchers to handle missing data. We list a number of key questions to consider when analyzing multilevel models. (3) Finally, based on the previous relevant questions, we present two detailed examples using the recommended R packages to be easy for the researcher applying multiple imputation in multilevel models.

Using the "Hidden" genome to improve classification of cancer types

It is increasingly common clinically for cancer specimens to be examined using techniques that identify somatic mutations. In principle, these mutational profiles can be used to diagnose the tissue of origin, a critical task for the 3% to 5% of tumors that have an unknown primary site. Diagnosis of primary site is also critical for screening tests that employ circulating DNA. However, most mutations observed in any new tumor are very rarely occurring mutations, and indeed the preponderance of these may never have been observed in any previous recorded tumor. To create a viable diagnostic tool we need to harness the information content in this "hidden genome" of variants for which no direct information is available. To accomplish this we propose a multilevel meta-feature regression to extract the critical information from rare variants in the training data in a way that permits us to also extract diagnostic information from any previously unobserved variants in the new tumor sample. A scalable implementation of the model is obtained by combining a high-dimensional feature screening approach with a group-lasso penalized maximum likelihood approach based on an equivalent mixed-effect representation of the multilevel model. We apply the method to the Cancer Genome Atlas whole-exome sequencing data set including 3702 tumor samples across seven common cancer sites. Results show that our multilevel approach can harness substantial diagnostic information from the hidden genome.

Occupational inequalities in volunteering participation: Using detailed data on jobs to explore the influence of habits and circumstances

In this paper we present empirical results that show that detailed occupations have distinctive patterns of association with voluntary participation. We draw upon data from four secondary survey datasets from the UK (coverage 1972-2012). Occupations are shown to link to volunteering in a wide range of scenarios and in individual, household, and longitudinal contexts. We argue that these linkages provide insight into social inequalities in volunteering, and that they can help us to understand the relative influence of "circumstance" and "habits" in enabling or inhibiting voluntary participation.

Differential trajectories of neurocognitive functioning in females versus males following treatment for pediatric brain tumors

BACKGROUND

Female and male trajectories of cerebellar and lobar brain structures are sexually dimorphic, making sex a potential candidate moderator of neurocognitive late effects from radiation treatment. We sought to evaluate longitudinal neurocognitive functioning in male versus female children treated for posterior fossa brain tumors.

METHODS

Fifty-one female and 63 male survivors of posterior fossa tumors completed neuropsychological testing at 2 timepoints. We included patients treated with surgical resection, chemotherapy, and radiation therapy. Multilevel mixed modeling was used to predict IQ score as a function of patient sex following treatment (~2 or ~4 years post treatment). Effect sizes were used as a measure of clinical significance.

RESULTS

Multilevel models resulted in a significant sex by time interaction (F = 6.69, P = 0.011). Females' cognitive scores were considerably higher compared with males at 4 years posttreatment. Females demonstrated an average improvement of 7.61 standard score IQ points compared with a decline of 2.97 points for males at 4 years follow-up. Effect sizes for female IQ compared with male IQ at 4 years posttreatment were between 0.8 and 0.9.

CONCLUSION

Trajectories of neurocognitive functioning following posterior fossa tumor treatment differed between female and male children. Sexual dimorphism in radiation late effects may alter treatment decisions in children. Research into sex-specific neuroprotective mechanisms underlying neurocognitive development following pediatric brain tumor treatments is warranted.

Alcohol Use and Harm to Children by Parents and Other Adults

This study used mixed methods to examine parent-reported harm to children (lack of supervision or physical harm) due to alcohol use by themselves or someone else as well as parent beliefs about alcohol, parenting, and harms to children. We conducted a cross-sectional general population survey of 1,599 parents who were primary caregivers to a child aged 10 or younger and follow-up interviews with 23 parents who responded "yes" to one of the questions about alcohol use causing harm their child. Survey data were analyzed using multilevel logistic models. Compared to abstainers, parents who drank at least once a year were less likely to report that someone else's drinking caused a lack of supervision or physical harm to their child. Higher continued volumes of drinking were related to fewer reports of not watching a child closely enough. Social companionship support (having people to go out with) was related to greater odds that a parent's drinking would cause physical harm to his or her child(ren). In the qualitative analysis, four relevant themes emerged: perceived effects of alcohol use, tangible support and child supervision, immediate and distal harm, and turning points in drinking behaviors.

On the Interpretation of Parameters in Multivariate Multilevel Models Across Different Combinations of Model Specification and Estimation

The increasing availability of software with which to estimate multivariate multilevel models (also called multilevel structural equation models) makes it easier than ever before to leverage these powerful techniques to answer research questions at multiple levels of analysis simultaneously. However, interpretation can be tricky given that different choices for centering model predictors can lead to different versions of what appear to be the same parameters; this is especially the case when the predictors are latent variables created through model-estimated variance components. A further complication is a recent change to Mplus (Version 8.1), a popular software program for estimating multivariate multilevel models, in which the selection of Bayesian estimation instead of maximum likelihood results in different lower-level predictors when random slopes are requested. This article provides a detailed explication of how the parameters of multilevel models differ as a function of the analyst's decisions regarding centering and the form of lower-level predictors (i.e., observed or latent), the method of estimation, and the variant of program syntax used. After explaining how different methods of centering lower-level observed predictor variables result in different higher-level effects within univariate multilevel models, this article uses simulated data to demonstrate how these same concepts apply in specifying multivariate multilevel models with latent lower-level predictor variables. Complete data, input, and output files for all of the example models have been made available online to further aid readers in accurately translating these central tenets of multivariate multilevel modeling into practice.

Relationships Among Children's Independent Mobility, Active Transportation, and Physical Activity: A Multisite Cross-Sectional Study

PURPOSE

Children's independent mobility (IM) may facilitate both active transportation (AT) and physical activity (PA), but previous studies examining these associations were conducted in single regions that provided limited geographical variability.

METHOD

We recruited 1699 children (55.0% girls) in 37 schools stratified by level of urbanization and socioeconomic status in 3 regions of Canada: Ottawa, Trois-Rivières, and Vancouver. Participants wore a SC-StepRx pedometer for 7 days and completed a validated questionnaire from which we derived a 6-point IM index, the number of AT trips over a week, and the volume of AT to/from school (in kilometer per week). We investigated relationships among measures of IM, AT, and PA employing linear mixed models or generalized linear mixed models adjusted for site, urbanization, and socioeconomic status.

RESULTS

Each unit increase in IM was associated with 9% more AT trips, 19% higher AT volume, and 147 more steps per day, with consistent results across genders. Both measures of AT were associated with marginally higher PA when pooling boys' and girls' data. Children in Vancouver engaged in more AT. PA did not vary across site, urbanization, or socioeconomic status.

CONCLUSION

IM was associated with more AT and PA regardless of where children lived, underscoring a need for IM interventions.

Discrete Survival Time Constructions for Studying Marital Formation and Dissolution in Rural South Africa

Introduction: Marriage formation and dissolution are important life-course events which impact psychological well-being and health of adults and children experiencing the events. Family studies have usually concentrated on analyzing single transitions including Never Married to Married and Married to Divorced. This does not allow understanding and interrogation of dynamics of these life changing events and their effects on individuals and their families. The objective of this study was to assess determinants associated with transitions between and within marital states in South Africa.

Methods: The population-based data available for this study consists of over 55, 000 subjects representing over 340, 000 person-years exposure from the Africa Health Research Institute (AHRI) in rural KwaZulu-Natal, South Africa. It was collected from 1 January 2004 to 31 December 2016. Multilevel multinomial, binary and competing risks regression models were used to model marital state occupation, transitions between marital states as well as investigate determinants of marital dissolution, respectively.

Results: Between the years 2006 and 2007, a subject was more likely to be married than never married when compared to years 2004 − 2005. After 2007, subjects were less likely to be married than never married and the trend reduced over the years up to 2016 [with OR=0.86, CI=(0.78; 0.94), OR=0.71, CI=(0.64; 0.78), OR=0.60, CI=(0.54; 0.67), OR=0.50, CI=(0.44; 0.56), and OR = 0.43, CI = (0.38; 0.48)] for periods 2008 − 2009, 2010 − 2011, 2012 − 2013, 2014 − 2015, and 2016, respectively. In 2008 − 2009, subjects were more likely to experience a marital dissolution than in the period 2004 − 2005 and the trend slightly reduces from 2010 until 2013 [OR=24.49, CI=(5.53; 108.37)]. Raising age at first sexual debut was found to be inversely associated with a marital dissolution [OR = 0.97;CI = (0.95; 0.99)]. Highly educated subjects were more likely to stay in one marital state than those who never went to school [OR=6.43, CI=(4.89; 8.47), OR=18.86, CI=(1.14; 53.31), and OR=2.96, CI=(1.96; 4.46) for being married, separated and widowed, respectively, among subjects with tertiary education]. As the age at first marriage increased, subjects became less likely to experience a marital separation [OR = 0.06, CI = (0.00; 1.11), OR = 0.05, CI = (0.00; 0.91), and OR = 0.04, CI = (0.00; 0.76) for subjects who entered a first marriage at ages 18 − 22, 23 − 29, and 30 − 40, respectively].

Conclusion: The study found that marrying at later ages is associated with a lower rate of marital dissolution while more educated subjects tend to stay longer in one marital state. Sexual debut at later ages was associated with a lower likelihood of experiencing a marital dissolution. There could, however, be some factors that are not accounted for in the model that may lead to heterogeneity in these dynamics in our model specification which are captured by the random effects in the model. Nonetheless, we may postulate that existing programs that encourage delay in onset of sexual activity for HIV risk reduction for example, may also have a positive impact on lowering rates of marital dissolution, thus ultimately improving psychological and physical health.

Outcomes of In Situ Training for Disclosure as a Standalone and a Booster to a Child Protective Behaviors Education Program

In this study, the effectiveness of the Observed Protective Behaviors behaviors test, a single-session, disclosure-focused, in situ skills training (IST), was evaluated as a standalone program (IST only) or as a booster to the child protective education program, Learn to be safe with Emmy and friends ^™ (program + IST). Participants included 281 Year 1 children (5-7 years; 52% male), randomly assigned to IST only, program + IST, program only or waitlist, and followed across 6 months. At each assessment, children completed interviews to assess their intention and confidence to disclose unsafe situations (disclosure intentions and confidence) and their ability to identify unsafe situations (safety identification skills). Children also reported their anxiety symptoms to assess for a possible iatrogenic effect. The IST-only condition was effective, with children showing increased disclosure intentions relative to waitlist children. The program + IST condition was also effective, with children showing increased disclosure intentions relative to children in the waitlist or program-only conditions as well as greater increases in disclosure confidence relative to waitlist children. No differences were observed between conditions in children's safety identification skills, and no iatrogenic effect on anxiety was found. Future research may seek to develop an IST that will also boost children's safety identification skills.

Bayesian hierarchical modeling of substate area estimates from the Medicare CAHPS survey

Each year, surveys are conducted to assess the quality of care for Medicare beneficiaries, using instruments from the Consumer Assessment of Healthcare Providers and Systems (CAHPS®) program. Currently, survey measures presented for Fee-for-Service beneficiaries are either pooled at the state level or unpooled for smaller substate areas nested within the state; the choice in each state is based on statistical tests of measure heterogeneity across areas within state. We fit spatial-temporal Bayesian random-effects models using a flexible parameterization to estimate mean scores for each of the domains formed by 94 areas in 32 states measured over 5 years. A Bayesian hat matrix provides a heuristic interpretation of the way the model combines information for estimates in these domains. The model can be used to choose between reporting of state- or substate-level direct estimates in each state, or as a source of alternative small-area estimates superior to either direct estimate. We compare several candidate models using log pseudomarginal likelihood and posterior predictive checks. Results from the best-performing model for 8 measures surveyed from 2012 to 2016 show substantial reductions in mean squared error (MSE) over direct estimates.

Distinguishing between-individual from within-individual predictors of gun carrying among Black and White males across adolescence

Longitudinal studies have found that male adolescents who deal drugs, associate with delinquent peers, and engage in aggressive behavior are at increased risk for carrying a gun (between-individual risks). However, it is unclear whether changes in these risk factors help to explain fluctuations in youth gun carrying across adolescence (within-individual risks). The current study examined this issue using a community sample of 970 adolescent males (58% Black, 42% White) assessed annually from ages 14 to 18. Multilevel models examined the extent to which between-individual differences and within-individual changes in drug dealing, peer delinquency, aggressive behavior, and neighborhood disadvantage were associated with gun carrying across adolescence. Each of these predictors, except for disadvantage, exerted a between-individual and within-individual influence for Black youth. For White youth, drug dealing was significant on both levels, peer delinquency was a significant between-individual predictor, and aggression was a significant within-individual predictor. Neighborhood disadvantage did not significantly predict gun carrying in the model, on either the between- or within-individual level, for Black or White youth. These results stress the importance of examining race-specific predictors of gun carrying among Black and White adolescents and point to drug dealing as a robust predictor of gun carrying, at both the between-individual and within-individual levels for youth of either race. Efforts to prevent drug market involvement and reduce aggressive behaviors in adolescence may in turn prove useful for preventing firearm violence. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Estimating Multilevel Models on Data Streams

Social scientists are often faced with data that have a nested structure: pupils are nested within schools, employees are nested within companies, or repeated measurements are nested within individuals. Nested data are typically analyzed using multilevel models. However, when data sets are extremely large or when new data continuously augment the data set, estimating multilevel models can be challenging: the current algorithms used to fit multilevel models repeatedly revisit all data points and end up consuming much time and computer memory. This is especially troublesome when predictions are needed in real time and observations keep streaming in. We address this problem by introducing the Streaming Expectation Maximization Approximation (SEMA) algorithm for fitting multilevel models online (or "row-by-row"). In an extensive simulation study, we demonstrate the performance of SEMA compared to traditional methods of fitting multilevel models. Next, SEMA is used to analyze an empirical data stream. The accuracy of SEMA is competitive to current state-of-the-art methods while being orders of magnitude faster.

Understanding the Influence of Text Complexity and Question Type on Reading Outcomes

In the current study, we examined how student characteristics and cognitive skills, differing levels of text complexity (cohesion, decoding, vocabulary, and syntax), and reading comprehension question types (literal, inferential, critical analysis, and reading strategy) affected different types of reading outcomes (multiple-choice reading comprehension questions, free recall, and oral reading fluency) in a sample of 181 native English-speaking adolescents (9 to 14.83 years). Results from item response theory one-parameter models and multilevel models suggested that different cognitive skills predicted performance across the three reading outcomes. After controlling for student characteristics and cognitive skills, text complexity negatively impacted reading outcomes, particularly oral reading fluency and free recall. Critical analysis and inferential questions emerged as the most difficult types of comprehension questions. The implications of these findings are discussed.

Homework and Academic Achievement in Latin America: A Multilevel Approach

The relationship between homework and academic results has been widely researched. Most of that research has used English-speaking, European or Asian samples, and to date there have been no detailed studies into that relationship in Latin America and the Caribbean. The aim of this study is to examine the effect of quantitative homework characteristics on achievement in science. The sample comprised 61,938 students at 2,955 schools in the 15 Latin American countries (plus the Mexican state of New Leon) which participated in the Third Regional Comparative and Explanatory Study (TERCE), carried out by the Latin American Laboratory for Educational Quality (LLECE) in 2013. The mean age was 12.42 years old (±0.94). Within each country, three hierarchical-linear models were applied at two levels: student and school. The individual level considered time spent doing homework and the school level considered the amount and frequency of homework assignment. In addition, ten control variables were included in order to control the net effect of the characteristics of the homework on the result. The results confirmed that homework is widely assigned in the Latin American region. At the individual level, time spent on homework had little effect on academic performance, while in the quantitative homework characteristics it was the frequency of homework assignment which demonstrated a clearer effect rather than the amount of homework assigned.

A discussion of statistical methods to characterise early growth and its impact on bone mineral content later in childhood

BACKGROUND

Many statistical methods are available to model longitudinal growth data and relate derived summary measures to later outcomes.

AIM

To apply and compare commonly used methods to a realistic scenario including pre- and postnatal data, missing data, and confounders.

SUBJECTS AND METHODS

Data were collected from 753 offspring in the Southampton Women's Survey with measurements of bone mineral content (BMC) at age 6 years. Ultrasound measures included crown-rump length (11 weeks' gestation) and femur length (19 and 34 weeks' gestation); postnatally, infant length (birth, 6 and 12 months) and height (2 and 3 years) were measured. A residual growth model, two-stage multilevel linear spline model, joint multilevel linear spline model, SITAR and a growth mixture model were used to relate growth to 6-year BMC.

RESULTS

Results from the residual growth, two-stage and joint multilevel linear spline models were most comparable: an increase in length at all ages was positively associated with BMC, the strongest association being with later growth. Both SITAR and the growth mixture model demonstrated that length was positively associated with BMC.

CONCLUSIONS

Similarities and differences in results from a variety of analytic strategies need to be understood in the context of each statistical methodology.

Multilevel Modeling of Cognitive Diagnostic Assessment: The Multilevel DINA Example

Many multilevel linear and item response theory models have been developed to account for multilevel data structures. However, most existing cognitive diagnostic models (CDMs) are unilevel in nature and become inapplicable when data have a multilevel structure. In this study, using the log-linear CDM as the item-level model, multilevel CDMs were developed based on the latent continuous variable approach and the multivariate Bernoulli distribution approach. In a series of simulations, the newly developed multilevel deterministic input, noisy, and gate (DINA) model was used as an example to evaluate the parameter recovery and consequences of ignoring the multilevel structures. The results indicated that all parameters in the new multilevel DINA were recovered fairly well by using the freeware Just Another Gibbs Sampler (JAGS) and that ignoring multilevel structures by fitting the standard unilevel DINA model resulted in poor estimates for the student-level covariates and underestimated standard errors, as well as led to poor recovery for the latent attribute profiles for individuals. An empirical example using the 2003 Trends in International Mathematics and Science Study eighth-grade mathematical test was provided.

Extending the Simple View of Reading to Account for Variation Within Readers and Across Texts: The Complete View of Reading (CVRi)

This study leverages advances in multivariate cross-classified random effects models to extend the Simple View of Reading to account for variation within readers and across texts, allowing for both the personalization of the reading function and the integration of the component skills and text and discourse frameworks for reading research. We illustrate the Complete View of Reading (CVRi) using data from an intensive longitudinal design study with a large sample of typical (N = 648) and struggling readers (N = 865) in middle school and using oral reading fluency as a proxy for comprehension. To illustrate the utility of the CVRi, we present a model with cross-classified random intercepts for students and passages and random slopes for growth, Lexile difficulty, and expository text type at the student level. We highlight differences between typical and struggling readers and differences across students in different grades. The model illustrates that readers develop differently and approach the reading task differently, showing differential impact of text features on their fluency. To be complete, a model of reading must be able to reflect this heterogeneity at the person and passage level, and the CVRi is a step in that direction. Implications for reading interventions and 21st century reading research in the era of "Big Data" and interest in phenotypic characterization are discussed.

A Comparison of Multilevel Imputation Schemes for Random Coefficient Models: Fully Conditional Specification and Joint Model Imputation with Random Covariance Matrices

Literature addressing missing data handling for random coefficient models is particularly scant, and the few studies to date have focused on the fully conditional specification framework and "reverse random coefficient" imputation. Although it has not received much attention in the literature, a joint modeling strategy that uses random within-cluster covariance matrices to preserve cluster-specific associations is a promising alternative for random coefficient analyses. This study is apparently the first to directly compare these procedures. Analytic results suggest that both imputation procedures can introduce bias-inducing incompatibilities with a random coefficient analysis model. Problems with fully conditional specification result from an incorrect distributional assumption, whereas joint imputation uses an underparameterized model that assumes uncorrelated intercepts and slopes. Monte Carlo simulations suggest that biases from these issues are tolerable if the missing data rate is 10% or lower and the sample is composed of at least 30 clusters with 15 observations per group. Furthermore, fully conditional specification tends to be superior with intraclass correlations that are typical of crosssectional data (e.g., ICC = .10), whereas the joint model is preferable with high values typical of longitudinal designs (e.g., ICC = .50).

Child Maltreatment and Physical Victimization: Does Heavy Drinking Mediate the Relationship?

Past studies examining the child maltreatment (CM)/victimization pathway have been limited by their focus on sexual victimization, narrow windows of assessment, and failure to examine gender differences. In the current study, we sought to examine (1) the impact of CM on physical victimization (PV) trajectories from adolescence to young adulthood and (2) the extent to which heavy drinking mediated the relationship between CM and later PV. Using three waves of the National Longitudinal Study of Adolescent to Adult Health, we found that CM was associated with a 69% greater odds of later PV for both genders, after the inclusion of control variables, and that the risk continued into adulthood. Further, heavy drinking was found to mediate the CM/victimization pathway at Wave I, but not at later waves. When mediation was examined separately for men and women, support for mediation was found for men and women. The current study suggests that CM represents a liability for interpersonal violence for both genders and highlights the importance of looking at victimization across time.