Multivariate Multilevel Modeling Research Articles

The Temporal and Directional Relationship Between Group-Level Implicit and Explicit Gender Bias

Explicit and implicit gender-science and gender-career biases have shifted toward neutrality in the past decade. Researchers speculate that these changes result from women's increased visibility in the science field and job market, but little is known about how the changes in group-level explicit and implicit gender-science and gender-career bias relate to one another over time. Building on contemporary models of group-level bias, this study investigates the temporal and directional relationship between group-level implicit and explicit gender bias between 2007 to 2016 using multivariate multilevel modeling. We found that lower group-level implicit bias in a previous month predicts lower group-level explicit bias in the following month. We also found evidence that group-level explicit bias in a previous month was positively associated with group-level implicit bias in the following month. These findings have practical and theoretical implications for understanding the bidirectional relationship between group-level implicit and explicit biases over time.

Interpersonal Violence, Racial Discrimination, and Mental Health Symptoms Among Adolescents of Color in the Juvenile-Justice System.

The current study examined how interpersonal racial discrimination experiences operate together with other forms of interpersonal violence to contribute to mental health symptoms among justice-involved adolescents of color. Participants were 118 justice-involved adolescents of color aged 14 to 17 (M = 15.77, SD = 1.08; 52.5% male; 77.1% Black/African American) and their mothers. At baseline, adolescents reported on experiences of interpersonal racial discrimination, harsh parenting, teen dating violence, and exposure to interparental physical intimate partner violence. At baseline and the 3-month follow-up assessment, adolescents reported on trauma symptoms, and adolescents and their mothers reported on the adolescents' externalizing and internalizing symptoms. Multivariate multilevel modeling results indicated that interpersonal racial discrimination experiences contributed additively to adolescent mental health symptoms at both the baseline and 3-month follow-up assessments, after accounting for exposure to other forms of interpersonal violence. The current findings highlight the importance of considering adolescents' experiences of interpersonal racial discrimination, together with other forms of interpersonal violence, in work focused on understanding the mental health symptoms of justice-involved adolescents of color.

Screen time vs. scream time: Developmental interrelations between young children's screen time, negative affect, and effortful control

This study aimed to examine developmental relations of screen time, negative affect and effortful control in children aged 12–36 months. Parents of 462 children up to 3 years of age at the start of the study (M = 1.28 years, SD = 0.61; 50% female) participated in four assessments within 10 months. Parents reported their children's screen time, negative affect, and effortful control at each assessment in a diary study. Results of multivariate Bayesian multilevel growth modeling revealed correlations between growth parameters of (1) children's screen time and their negative affect and (2) children's negative affect and their effortful control but not between growth parameters of (3) children's screen time and their effortful control. Overall, these results indicate that children's screen time may be associated with their negative affect independently of their effortful control. Hence, we found no evidence of displacement in the development of self-regulatory strategies in children of parents with higher levels of education. Future research on this topic should focus on children's excessive screen media use, considering contextual and content-related screen media factors and other factors in the child's immediate and broader environment.

Motivation, Physical Activity, and Affect in Community-Dwelling Stroke Survivors: An Ambulatory Assessment Approach.

Motivation is a frequently reported but far less studied driver for post-stroke physical activity participation. Motivation and physical activity may be important contributors to the prevention management and alleviation of affective symptoms among stroke survivors. To investigate the real-time associations between motivation, physical activity, and affect in the daily lives of community-dwelling stroke survivors using ecological momentary assessment (EMA) and accelerometry. Forty community-dwelling stroke survivors wore an accelerometer on the thigh and completed EMA surveys assessing motivation (autonomous motivation, controlled motivation) and affect (negative affect, positive affect) eight times daily for 7 days. Multivariate regression analysis and multilevel modeling investigated the associations between motivation, physical activity, and affect. Greater autonomous motivation for physical activity was associated with less sedentary behavior (β = -0.40, p = .049) and more moderate-to-vigorous physical activity (β = 0.45, p = .020) participation in daily life. Greater autonomous motivation was momentarily associated with less depressed affect (β = -0.05, p < .001) and greater positive affect (β = 0.13, p < .001). Moreover, greater controlled motivation was momentarily associated with greater depressed affect (β = 0.06, p < .001). More intense physical activity was momentarily associated with greater positive affect (β = 0.13, p = .016). No moderating effect of motivation on the association between physical activity and affect was found. Motivation and physical activity are momentarily associated with affect among stroke survivors. Assessing and fostering autonomous motivation may be beneficial for promoting physical activity and managing positive and depressed affect as stroke survivors return to the community.

Perceived partner responsiveness is associated with sexual well-being in couples with genito-pelvic pain.

Intimacy is vital to romantic relationships, yet is often thwarted by relational challenges, such as sexual difficulties. With prevalence estimates ranging from 10% to 28%, genito-pelvic pain/penetration disorder (GPPPD) is an important sexual problem resulting in negative consequences for affected women and their partners, including significant sexual dysfunction and dissatisfaction. Findings from cross-sectional studies indicate that higher levels of intimacy are associated with better sexuality outcomes in couples coping with GPPPD. However, single-occasion measurements may not capture the daily variations in intimacy that could have important implications for couples' sexual well-being. The present study focused on a key intimacy-building component-perceived partner responsiveness (PPR)-and its daily associations with women's pain and both partners' sexual function and satisfaction. Using daily diaries over an 8-week period, 160 women (Mage = 26.23, SD = 6.26) with GPPPD and their partners (Mage = 27.73, SD = 7.35) reported on PPR, sexual function and satisfaction, as well as women's pain, on days when sexual activity occurred (M = 8.87, SD = 5.87). Drawing on the Actor-Partner Interdependence Model, a multivariate multilevel modeling approach was adopted. Controlling for trait-level PPR, when women and partners reported greater PPR, women reported greater sexual function and satisfaction and partners reported greater sexual function. When partners reported greater PPR, they reported greater sexual satisfaction. No association was found between PPR and women's pain. PPR may facilitate better sexual function and satisfaction and could be an important target in sex and couple therapy for GPPPD. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Associations Between Spirituality, Mindfulness, and Psychological Symptoms Among Advanced Lung Cancer Patients and Their Spousal Caregivers

ContextPatients with metastatic lung cancer and their spousal caregivers are at high risk of psychological symptoms. Mindfulness may improve psychological symptoms via spiritual well-being (SW); yet, this mediation model has not been examined in a dyadic context. ObjectivesWe examined the mediating role of two dimensions of SW (meaning/peace and faith) in the mindfulness-symptoms link in Stage IV lung cancer patients and their spousal caregivers. MethodsWe examined the actor-partner interdependence model of mediation using multivariate multilevel modeling with 78 couples. Four actor-partner interdependence model of mediation analyses were conducted to examine one predictor (mindfulness) × two mediators (meaning/peace and faith) × two psychological symptoms (depressive symptoms and cancer distress). We also tested four alternative models in which mindfulness mediates the associations between SW and psychological symptoms. ResultsThe alternative model (SW → mindfulness → psychological symptoms) was preferred than the original model (mindfulness → SW → psychological symptoms). For patients, meaning/peace was directly associated with their own psychological symptoms, whereas faith was only indirectly associated with their own psychological symptoms via mindfulness. For spouses, meaning/peace was both directly and indirectly associated with their own psychological symptoms, whereas faith was only directly associated with their own depressive symptoms (but not cancer distress). Moreover, spouses' faith was indirectly associated with patients' psychological symptoms through patients' mindfulness. ConclusionSW is associated with patients' and spouses' psychological symptoms both directly and indirectly through mindfulness. Thus, interventions that target SW, particularly meaning and peace, along with mindfulness may be beneficial to the psychological management of patients facing a terminal disease and their spousal caregivers.

Changes of Occupational Noise-Induced Hearing Loss due to Working in the Steel Industry and Associated Effective Factors: Application of Bayesian Multivariate Multilevel Modeling using Skew Distribution : Rate of NIHL in Steel Workers

Background:There have been a limited number of studies on the relationships between time and occupational and demographical variables with the mean changes of low-frequency hearing (LFH) and high-frequency hearing (HFH). Objectives: This study investigated the rate of occupational noise-induced hearing loss (NIHL) due to working in the steel industry and associated effective factors. Materials and Methods: This historical cohort study was conducted within 2000 to 2010. Two LFH and HFH definitions of NIHL were used in this study. The average changes of LFH and HFL were considered the response variables. In addition, time and occupational (i.e., shift work and work experience) and demographical (i.e., age and educational level) variables were regarded as the independent variables. For data analysis, Bayesian multivariate multilevel modeling using skew distribution and OpenBUGS (version 3.2.2) and R (version 2.13.2) software were used in this study. Results: The present study was performed on 1,959 male workers with a mean age of 36.64±3.92 years. Among these subjects, 913 (46.6%), 134 (6.8%), and 912 (46.6%) participants were day workers, weekly-rotating shift workers, and routinely-rotating shift workers, respectively. The obtained results showed that age, work experience, educational level, and shift work had significant relationships with the changes of LFH and HFH. Conclusion:Overall, the findings of this 10-year historical cohort study demonstrated a relationship between time and demographical and occupational variables with the changes of LFH and HFL. Therefore, it is recommended to design preventive measures to reduce the deleterious effects of such variables on LFH and HFL.

Promoting primary school pupils’ language achievement: investigating the impact of school-based language policies

ABSTRACT School-based language policies (SLP) are expected to have a positive effect on student achievement. To date, few studies have empirically examined the impact of such policies on student outcomes. This study investigates to what extent SLPs are related to pupils’ reading performance from an educational effectiveness perspective. In addition, the study investigates whether SLPs have differential effects for pupils at risk of underachievement. Data from a cross-sectional study involving 3,000 pupils in the first, third, and sixth grades of 28 elementary schools in Flanders, the northern Dutch-speaking part of Belgium, were analyzed using multivariate multilevel modeling. Results indicate that one SLP component, team reflective capacity for language instruction, positively affects students’ decoding skills in reading. No differential effects for low- versus high-risk pupils were found.

Multivariate multilevel modeling of quality of life dynamics of HIV infected patients

BackgroundLongitudinal quality of life (QoL) is an important outcome in many chronic illness studies aiming to evaluate the efficiency of care both at the patient and health system level. Although many QoL studies involve multiple correlated hierarchical outcome measures, very few of them use multivariate modeling. In this work, we modeled the long-term dynamics of QoL scores accounting for the correlation between the QoL scores in a multilevel multivariate framework and to compare the effects of covariates across the outcomes.MethodsThe data is from an ongoing prospective cohort study conducted amongst adult women who were HIV-infected and on the treatment in Kwazulu-Natal, South Africa. Independent and related QoL outcome multivariate multilevel models were presented and compared.ResultsThe analysis showed that related outcome multivariate multilevel models fit better for our data used. Our analyses also revealed that higher educational levels, middle age, stable sex partners and higher weights had a significant effect on better improvements in the rate of change of QoL scores of HIV infected patients. Similarly, patients without TB co-infection, without thrombocytopenia, with lower viral load, with higher CD4 cell count levels, with higher electrolytes component score, with higher red blood cell (RBC) component score and with lower liver abnormality component score, were associated with significantly improved the rate of change of QoL, amongst HIV infected patients.ConclusionIt is hoped that the article will help applied researchers to familiarize themselves with the models and including interpretation of results. Furthermore, three issues are highlighted: model building of multivariate multilevel outcomes, how this model can be used to assess multivariate assumptions, involving fixed effects (for example, to examine the size of the covariate effect varying across QoL domain scores) and random effects (for example, to examine the rate of change in one response variable associated to changes in the other).

Comparing organizational content and fan interaction on Twitter and Facebook in United States professional sport

ABSTRACT Rationale/Purpose: The purpose of this study was to extend social media research by comparing how professional sport organizations in the United States used Facebook and Twitter in the off-season and how interaction on these networks differs. Design/Methodology/Approach: Data were collected from professional sport teams’ Facebook and Twitter pages and deductive content analysis was used. Interaction data were then analyzed using multivariate multilevel modeling. Findings: Results suggested both networks were used most often for player and personnel promotion, and fans interacted most often with this content. However, Twitter was used more for information dissemination and Facebook was used more for organizational promotion. Overall, interaction on Facebook was significantly greater than interaction on Twitter. Practical Implications: Sport marketers and researchers can utilize these results to make strategic content plans to effectively engage fans. Research Contribution: This study used statistical modeling to compare and test differences between networks and focused on off-season content.

Positive Affect Is Associated With Decreased Symptom Severity in the Daily Lives of Individuals With Borderline Personality Disorder.

Previous research has repeatedly demonstrated positive associations between negative affect (NA) and borderline personality disorder (BPD) symptoms in daily life, but studies have rarely addressed potential effects of positive affect (PA). Consequently, little is known about how PA in daily life covaries with symptoms of BPD. The authors assessed the effects of both PA and NA levels on BPD symptom severity in a sample of 81 treatment-seeking women diagnosed with BPD over a period of 21 days, employing a daily diary design. Using multivariate multilevel modeling, the authors obtained negative associations between PA levels and daily BPD severity in total and at the level of the individual symptoms inappropriate anger, affective instability, emptiness, identity disturbance, and paranoid ideation/dissociation. Moreover, the authors replicated previously reported positive associations between NA and BPD severity for all nine symptoms. Future research can address whether increasing PA in the treatment of BPD may potentially help reduce symptom burden.

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.

An exploration of Illinois students' attitudes toward science using multivariate multilevel modeling with a cross‐sectional sample of responses from grades 5 through 10

AbstractThe present study examined students' attitudes toward science and associated constructs, based on the theories of reasoned action and planned behavior, and explored relationships between individual and school‐related variables common to the research literature. Responses from 1,291 students in Grades 5 through 10 were collected using the 30‐item Behaviors, Related Attitudes, and Intentions toward Science (BRAINS) Survey along with background information questions. Additional self‐report data were collected from teachers (n = 56; 82.4%) in participating schools (n = 68) to obtain information about their education and experience, characteristics and practices, as well as other classroom variables, which could influence students' outlook. Student information, teacher data collected, and other data compiled about participating schools, were used to explore patterns in students' attitudes, beliefs, and intentions. These variables were used to generate multivariate multilevel models through a forward construction process. The final model presented favors individual variables to explain differences in students' responses on all five of the BRAINS subscales, more than group‐level variables captured. Of the predictor variables explored, students' perceived science ability and frequency of talk with family were influential on all subscales, and increasing these variables had a positive effect on the estimated mean scores according to the final model presented. Findings from this study also include commonly observed relationships, such as the decline in attitudes over time, but these were found to be less pervasive in this sample. The paper concludes with a discussion about the comparative ineffectiveness of teacher and school‐related variables in explaining students' attitudes toward science in this study, in light of design decisions and limitations, to guide future investigations.

A comparison of nonsense-word fluency and curriculum-based measurement of reading to measure response to phonics instruction.

Student response to instruction is a key piece of information that school psychologists use to make instructional decisions. Curriculum-based measures (CBMs) are the most commonly used and researched family of academic progress-monitoring assessments. There are a variety of reading CBMs that differ in the type and specificity of skills they assess. The purpose of this study was to determine the degree to which the CBM of oral reading (CBM-R) progress-monitoring data differed from nonsense-word fluency (NWF) progress-monitoring data in the presence of a common intervention. We used multivariate multilevel modeling to compare growth trajectories from CBM-R and NWF progress-monitoring data from a geographically diverse sample of 3,000 1st-grade students receiving Tier-2 phonics interventions. We also evaluated differences in sensitivity to improvement and reliability of improvement from each measure. Improvement on CBM-R was statistically, but not practically, significantly greater than NWF. Although CBM-R was not as direct a measure of decoding, it still captured student response to phonics instruction similarly to NWF. NWF demonstrated slightly better sensitivity to growth, but CBM-R yielded more reliable growth estimates. (PsycINFO Database Record (c) 2018 APA, all rights reserved).

The effects of school contexts and student characteristics on cognitive and affective achievement in South Korea

This study investigated the effects of educational context variables on two types of achievement: cognitive domain and affective domain. To assess the effects of educational context variables at student and school levels, the National Assessment of Educational Achievement data were used, which were collected by the Korea Educational Development Institute. Participants were in the second year of a high school. Multivariate multilevel modeling was employed to model the two related dependent variables: cognitive and affective domains. Results showed that school characteristics were generally more influential for cognitive achievement than for affective achievement. Moreover, it was found that psychological stability and satisfaction with the facilities at school played an important role in predicting affective achievement for the subjects of Korean, English, and mathematics. Korea’s Educational Broadcasting System was found to exert a positive effect on affective achievement, but negative effects on cognitive achievement. Shadow education was found to be generally helpful in improving cognitive/affective achievement. Programs aimed at improving academic achievement at school were found to have had very limited success in accomplishing their aim. Teacher climate, in particular, adversely affected students’ cognitive achievement. Schools must be equipped with good facilities and counseling centers for student to improve student achievement.

Verifying the ecological model of peer aggression on Croatian students

AbstractUsing the ecological approach, the goal of this study was to determine the predictors of physical and verbal peer aggressive behavior. The participants were 880 school students from the fifth to eighth grade (48% boys and 52% girls) and the same number of parents (19% fathers and 61% mothers) as well as 107 teachers. The main analysis was performed using multivariate–multilevel modeling. The following significant predictors of physical peer aggression were obtained: Boys, a greater degree of impulsivity, more parental punishment, poorer school success, more time spent with the media, and the perception of great neighborhood dangerousness. For verbal peer aggression, the significant predictors were: A greater degree of impulsivity, lower level of affective empathy, more parental punishment, lack of parental supervision, lesser peer acceptance, large differences in family income, more time spent with the media, and the perception of great neighborhood dangerousness. A moderating effect of neighborhood dangerousness and parental supervision was found. The results were interpreted within Bronfenbrenner's ecological model.

Interpersonal behavior in couple therapy: Concurrent and prospective associations with depressive symptoms and relationship distress

Objective: This study investigated associations between couples’ interpersonal behavior, depressive symptoms, and relationship distress over the course of couple psychotherapy. Method: After every other session of Integrative Systemic Therapy (M = 13 sessions), N = 100 individuals within 50 couples rated their in-session affiliation and autonomy behavior using the circumplex-based Structural Analysis of Social Behavior Intrex. Concurrent and prospective associations of interpersonal behavior with depressive symptoms and relationship distress were evaluated via multivariate multilevel modeling using the Actor-Partner Interdependence Model. Results: An individual’s hostility, as well as the partner’s hostility, positively predicted an individual’s concurrent depressive symptoms and relationship distress, as well as his or her relationship distress at the following session. Partner hostility during one session predicted an individual’s subsequent depressive symptoms. During sessions in which individuals controlled the partner, and separated themselves from the partner, they reported more concurrent depressive symptoms and relationship distress, and more subsequent relationship distress. When partners separated themselves, individuals reported more concurrent depressive symptoms and relationship distress, and more subsequent relationship distress. Conclusions: Results underscore the importance of couples’ in-session affiliation and autonomy behavior in the treatment of depressive symptoms and relationship distress within couple therapy.

Tasks, interaction and role perception of management accountants: evidence from Germany

The role of management accountants (MAs) has been discussed for more than a decade. Scholars as well as practitioners advocate a new role for MAs: the business partner who is involved in managerial decisions and strategic tasks. This is contrasted with the “bean counter” doing more repetitive and operational tasks. Different tasks and accountant–manager interactions are important indicators of MAs’ roles. An empirical assessment of them would allow us to understand what roles MAs fulfill and which contingency factors impact roles. Contrasting results on tasks and interaction with self-perceived roles of MAs helps to understand the extent of self-serving perception. The sample consists of 183 German MAs. Data were analyzed using multivariate multilevel modeling and ordinal regression. On a descriptive level, the results indicate an extension of tasks from operational into strategic areas. That supports the notion of role extension and so-called hybrid MAs. The main contingency factors affecting roles are the position of MAs, the hierarchical level of the MA function, group affiliation and capital market orientation. The second result is that role perception of respondents is independent of what they do and how intensively they interact with managers. It follows from the study that assuming clear-cut roles and transition to one role is too simplistic. A great variety of tasks and interactions of MAs exists. Future studies should take this complexity fully into account. Additionally, self-rating and self-perceived roles do not seem to be coherent with responses of MAs on what they do and how intensively they interact with managers, which is explained by self-serving bias.

Comparison of progress monitoring data from general outcome measures and specific subskill mastery measures for reading

Interventionists often monitor the progress of students receiving supplemental interventions with general outcome measures (GOMs) such as curriculum-based measurement of reading (CBM-R). However, some researchers have suggested that interventionists should collect data more closely related to instructional targets, specific subskill mastery measures (SSMMs) because outcomes from GOMs such as CBM-R may not be sufficiently sensitive to gauge intervention effects. In turn, interventionists may prematurely terminate an effective intervention or continue to deliver an ineffective intervention if they do not monitor student progress with the appropriate measure. However, such recommendations are based upon expert opinion or studies with serious methodological shortcomings. We used multi-variate multilevel modeling to compare pre-intervention intercepts and intervention slopes between GOM and SSMM data collected concurrently in a sample of 96 first, 44 second, and 53 third grade students receiving tier 2 phonics interventions. Statistically significant differences were observed between slopes from SSMM consonant-vowel-consonant words and CBM-R data. Statistically significant differences in slopes were not observed for consonant blend, digraph or consonant-vowel-consonant-silent e (CVCe) SSMMs. Results suggest that using word lists to monitor student response to instruction for early struggling readers is beneficial but as students are exposed to more complex phonetic patterns, the distinction between SSMMs and CBM-R become less meaningful.

Multivariate multilevel modeling of age related diseases

Multivariate multilevel modeling of age related diseases