Are factor scores measurement invariant?

There has been increased interest in practical methods for integrative analysis of data from multiple studies or samples, and using factor scores to represent constructs has become a popular and practical alternative to latent variable models with all individual items. Although researchers are aware that scores representing the same construct should be on a similar metric across samples-namely they should be measurement invariant-for integrative data analysis, the methodological literature is unclear whether factor scores would satisfy such a requirement. In this note, we show that even when researchers successfully calibrate the latent factors to the same metric across samples, factor scores-which are estimates of the latent factors but not the factors themselves-may not be measurement invariant. Specifically, we prove that factor scores computed based on the popular regression method are generally not measurement invariant. Surprisingly, such scores can be noninvariant even when the items are invariant. We also demonstrate that our conclusions generalize to similar shrinkage scores in item response models for discrete items, namely the expected a posteriori scores and the maximum a posteriori scores. Researchers should be cautious in directly using factor scores for cross-sample analyses, even when such scores are obtained from measurement models that account for noninvariance. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Correcting for Sampling Error in between-Cluster Effects: An Empirical Bayes Cluster-Mean Approach with Finite Population Corrections.

With clustered data, such as where students are nested within schools or employees are nested within organizations, it is often of interest to estimate and compare associations among variables separately for each level. While researchers routinely estimate between-cluster effects using the sample cluster means of a predictor, previous research has shown that such practice leads to biased estimates of coefficients at the between level, and recent research has recommended the use of latent cluster means with the multilevel structural equation modeling framework. However, the latent cluster mean approach may not always be the best choice as it (a) relies on the assumption that the population cluster sizes are close to infinite, (b) requires a relatively large number of clusters, and (c) is currently only implemented in specialized software such as Mplus. In this paper, we show how using empirical Bayes estimates of the cluster means can also lead to consistent estimates of between-level coefficients, and illustrate how the empirical Bayes estimate can incorporate finite population corrections when information on population cluster sizes is available. Through a series of Monte Carlo simulation studies, we show that the empirical Bayes cluster-mean approach performs similarly to the latent cluster mean approach for estimating the between-cluster coefficients in most conditions when the infinite-population assumption holds, and applying the finite population correction provides reasonable point and interval estimates when the population is finite. The performance of EBM can be further improved with restricted maximum likelihood estimation and likelihood-based confidence intervals. We also provide an R function that implements the empirical Bayes cluster-mean approach, and illustrate it using data from the classic High School and Beyond Study.

Evaluating two small-sample corrections for fixed-effects standard errors and inferences in multilevel models with heteroscedastic, unbalanced, clustered data.

Multilevel modeling (MLM) is commonly used in psychological research to model clustered data. However, data in applied research usually violate one of the essential assumptions of MLM-homogeneity of variance. While the fixed-effect estimates produced by the maximum likelihood method remain unbiased, the standard errors for the fixed effects are misestimated, resulting in inaccurate inferences and inflated or deflated type I error rates. To correct the bias in fixed effects standard errors and provide valid inferences, small-sample corrections such as the Kenward-Roger (KR) adjustment and the adjusted cluster-robust standard errors (CR-SEs) with the Satterthwaite approximation for t tests have been used. The current study compares KR with random slope (RS) models and the adjusted CR-SEs with ordinary least squares (OLS), random intercept (RI) and RS models to analyze small, heteroscedastic, clustered data using a Monte Carlo simulation. Results show the KR procedure with RS models has large biases and inflated type I error rates for between-cluster effects in the presence of level 2 heteroscedasticity. In contrast, the adjusted CR-SEs generally yield results with acceptable biases and maintain type I error rates close to the nominal level for all examined models. Thus, when the interest is only in within-cluster effect, any model with the adjusted CR-SEs could be used. However, when the interest is to make accurate inferences of the between-cluster effect, researchers should use the adjusted CR-SEs with RS to have higher power and guard against unmodeled heterogeneity. We reanalyzed an example in Snijders & Bosker (2012) to demonstrate the use of the adjusted CR-SEs with different models.

Does strict invariance matter? Valid group mean comparisons with ordered-categorical items.

Measurement invariance (MI) of a psychometric scale is a prerequisite for valid group comparisons of the measured construct. While the invariance of loadings and intercepts (i.e., scalar invariance) supports comparisons of factor means and observed means with continuous items, a general belief is that the same holds with ordered-categorical (i.e., ordered-polytomous and dichotomous) items. However, as this paper shows, this belief is only partially true-factor mean comparison is permissible in the correctly specified scalar invariance model with ordered-polytomous items but not with dichotomous items. Furthermore, rather than scalar invariance, full strict invariance-invariance of loadings, thresholds, intercepts, and unique factor variances in all items-is needed when comparing observed means with both ordered-polytomous and dichotomous items. In a Monte Carlo simulation study, we found that unique factor noninvariance led to biased estimations and inferences (e.g., with inflated type I error rates of 19.52%) of (a) the observed mean difference for both ordered-polytomous and dichotomous items and (b) the factor mean difference for dichotomous items in the scalar invariance model. We provide a tutorial on invariance testing with ordered-categorical items as well as suggestions on mean comparisons when strict invariance is violated. In general, we recommend testing strict invariance prior to comparing observed means with ordered-categorical items and adjusting for partial invariance to compare factor means if strict invariance fails.

The roles of caregivers and friends in adolescent daily emotion dynamics.

Daily emotion dynamics provide valuable information about individuals' emotion processes as they go about their lives. Emotion dynamics such as emotion levels (mean), emotion variability (degree of fluctuation), and emotion network density (strength of temporal connections among emotions) are associated with risks for various psychopathology in youth and adults. Prior work has shown that caregivers and friends play crucial socializing roles in adolescent emotional well-being, but less is known about their roles in daily emotion dynamics. This study examined whether caregiver emotion coaching, caregiver-adolescent closeness, and friendship quality were associated with adolescents' emotion levels, emotion variability, and emotion network density. Further, we examined whether caregiver-adolescent closeness moderated the associations between coaching and emotion dynamics. Participants were 150 adolescents (61% girls; Mage = 14.75) and one of their caregivers (95% female; Mage = 43.35) who completed a baseline survey and 21 daily surveys. Results showed that caregiver emotion coaching interacted with caregiver-adolescent closeness in predicting emotion levels and variability. Specifically, when closeness was higher, emotion coaching was significantly associated with lower sadness and anger levels, higher happiness levels, and lower happiness variability. Caregiver emotion coaching, independent of closeness, was also associated with lower anxiety levels, lower sadness variability, and lower emotion network density. Friendship quality was significantly associated with lower levels of sadness, anxiety, and anger, higher levels of happiness, and lower variability in anxiety and anger. These findings suggest that caregivers and friends are central to everyday emotion levels and variability and a more flexible emotion system in adolescents.

Measurement Invariance of the Neurobehavioral Symptom Inventory in Male and Female Million Veteran Program Enrollees Completing the Comprehensive Traumatic Brain Injury Evaluation.

This study evaluated measurement invariance across males and females on the Neurobehavioral Symptom Inventory (NSI) in U.S. military veterans enrolled in the VA Million Veteran Program. Participants (N = 17,059; males: n = 15,450; females: n = 1,609) included Veterans who took part in the VA Traumatic Brain Injury (TBI) Screening and Evaluation Program and completed the NSI. Multiple-group confirmatory factor analyses investigated measurement invariance of the NSI 4-factor model. The configural (comparative fit index [CFI] = 0.948, root mean square error of approximation [RMSEA] = 0.060) and metric (CFI = 0.948, RMSEA = 0.058) invariance models showed acceptable fit. There was a minor violation of scalar invariance (Δχ2 = 232.50, p < .001); however, the degree of noninvariance was mild (ΔCFI = -0.002, ΔRMSEA=0.000). Our results demonstrate measurement invariance across sex, suggesting that the NSI 4-factor model can be used to accurately assess symptoms in males and females following TBI. Findings highlight the importance of considering validity of measurement across study groups to increase confidence that a measure is interpreted similarly by respondents from different subgroups.

Sedentary Behavior and Incident Dementia Among Older Adults.

Importance: Sedentary behavior is associated with cardiometabolic disease and mortality, but its association with dementia is unclear. Objective: To investigate whether accelerometer-assessed sedentary behavior is associated with incident dementia. Design, Setting, and Participants: A retrospective study of prospectively collected data from the UK Biobank including 49 841 adults aged 60 years or older without a diagnosis of dementia at the time of wearing the wrist accelerometer and living in England, Scotland, or Wales. Follow-up began at the time of wearing the accelerometer (February 2013 to December 2015) and continued until September 2021 in England, July 2021 in Scotland, and February 2018 in Wales. Exposures: Mean daily sedentary behavior time (included in the primary analysis) and mean daily sedentary bout length, maximum daily sedentary bout length, and mean number of daily sedentary bouts (included in the secondary analyses) were derived from a machine learning-based analysis of 1 week of wrist-worn accelerometer data. Main Outcome and Measures: Incident all-cause dementia diagnosis from inpatient hospital records and death registry data. Cox proportional hazard models with linear and cubic spline terms were used to assess associations. Results: A total of 49 841 older adults (mean age, 67.19 [SD, 4.29] years; 54.7% were female) were followed up for a mean of 6.72 years (SD, 0.95 years). During this time, 414 individuals were diagnosed with incident all-cause dementia. In the fully adjusted models, there was a significant nonlinear association between time spent in sedentary behavior and incident dementia. Relative to a median of 9.27 hours/d for sedentary behavior, the hazard ratios (HRs) for dementia were 1.08 (95% CI, 1.04-1.12, P < .001) for 10 hours/d, 1.63 (95% CI, 1.35-1.97, P < .001) for 12 hours/d, and 3.21 (95% CI, 2.05-5.04, P < .001) for 15 hours/d. The adjusted incidence rate of dementia per 1000 person-years was 7.49 (95% CI, 7.48-7.49) for 9.27 hours/d of sedentary behavior, 8.06 (95% CI, 7.76-8.36) for 10 hours/d, 12.00 (95% CI, 10.00-14.36) for 12 hours/d, and 22.74 (95% CI, 14.92-34.11) for 15 hours/d. Mean daily sedentary bout length (HR, 1.53 [95% CI, 1.03-2.27], P = .04 and 0.65 [95% CI, 0.04-1.57] more dementia cases per 1000 person-years for a 1-hour increase from the mean of 0.48 hours) and maximum daily sedentary bout length (HR, 1.15 [95% CI, 1.02-1.31], P = .02 and 0.19 [95% CI, 0.02-0.38] more dementia cases per 1000 person-years for a 1-hour increase from the mean of 1.95 hours) were significantly associated with higher risk of incident dementia. The number of sedentary bouts per day was not associated with higher risk of incident dementia (HR, 1.00 [95% CI, 0.99-1.01], P = .89). In the sensitivity analyses, after adjustment for time spent in sedentary behavior, the mean daily sedentary bout length and the maximum daily sedentary bout length were no longer significantly associated with incident dementia. Conclusions and Relevance: Among older adults, more time spent in sedentary behaviors was significantly associated with higher incidence of all-cause dementia. Future research is needed to determine whether the association between sedentary behavior and risk of dementia is causal.

The Impacts of the COVID-19 Pandemic on Therapy Utilization Among Racially/Ethnically and Socio-Economically Diverse Autistic Children.

PURPOSE: The purpose of current study was to evaluate change in hours of Applied Behavior Analysis (ABA) therapy utilization for autistic children during the year prior to the COVID-19 pandemic, the first three months of the pandemic (crisis phase), and the following 9 months of the pandemic (mitigation phase). Additionally, this study aimed to evaluate if change in therapy utilization differed based on child race, ethnicity, and primary payer of services. Finally, we aimed to identify potential mechanisms of ABA therapy disruption by interpreting findings using an extended version of Donabedian's structure-process-outcome model. METHODS: Retrospective clinical data on client demographics and therapy utilization (n = 283) were collected from ABA clinics in California and analyzed with four piecewise growth multi-level models. RESULTS: We found that therapy utilization dropped during the first three months of the pandemic (-10.65 h/month; p < .001) and increased during the following 9 months (2.39 h/month; p < .001). Moderator analyses revelated that Asian, Non-Latinx and school-district funded children had significantly different trajectories of change in therapy utilization compared to white, non-Latinx participants and private insurance funded participants, respectively. CONCLUSION: Findings suggest that utilization of ABA therapy was disrupted for a full year following the onset of the COVID-19 pandemic and that child race/ethnicity and primary payer influenced the degree to which autistic children were impacted by service disruption. These findings have implications for autistic children who lost therapy access during key developmental periods and for the ABA care delivery system.

An Examination of Racial/Ethnic Differences on the Neurobehavioral Symptom Inventory Among Veterans Completing the Comprehensive Traumatic Brain Injury Evaluation: A Veterans Affairs Million Veteran Program Study.

OBJECTIVE: The purpose of this study was to explore racial/ethnic differences in neurobehavioral symptom reporting and symptom validity testing among military veterans with a history of traumatic brain injury (TBI). METHOD: Participants of this observational cross-sectional study (N = 9,646) were post-deployed Iraq-/Afghanistan-era veterans enrolled in the VA's Million Veteran Program with a clinician-confirmed history of TBI on the Comprehensive TBI Evaluation (CTBIE). Racial/ethnic groups included White, Black, Hispanic, Asian, Multiracial, Another Race, American Indian or Alaska Native, and Native Hawaiian or Other Pacific Islander. Dependent variables included neurobehavioral symptom domains and symptom validity assessed via the Neurobehavioral Symptom Inventory (NSI) and Validity-10, respectively. RESULTS: Chi-square analyses showed significant racial/ethnic group differences for vestibular, somatic/sensory, and affective symptoms as well as for all Validity-10 cutoff scores examined (≥33, ≥27, ≥26, >22, ≥22, ≥13, and ≥7). Follow-up analyses compared all racial/ethnic groups to one another, adjusting for sociodemographic- and injury-related characteristics. These analyses revealed that the affective symptom domain and the Validity-10 cutoff of ≥13 revealed the greatest number of racial/ethnic differences. CONCLUSIONS: Results showed significant racial/ethnic group differences on neurobehavioral symptom domains and symptom validity testing among veterans who completed the CTBIE. An enhanced understanding of how symptoms vary by race/ethnicity is vital so that clinical care can be appropriately tailored to the unique needs of all veterans. Results highlight the importance of establishing measurement invariance of the NSI across race/ethnicity and underscore the need for ongoing research to determine the most appropriate Validity-10 cutoff score(s) to use across racially/ethnically diverse veterans.

Adjusting for Measurement Noninvariance with Alignment in Growth Modeling.

Longitudinal measurement invariance-the consistency of measurement in data collected over time-is a prerequisite for any meaningful inferences of growth patterns. When one or more items measuring the construct of interest show noninvariant measurement properties over time, it leads to biased parameter estimates and inferences on the growth parameters. In this paper, I extend the recently developed alignment-within-confirmatory factor analysis (AwC) technique to adjust for measurement biases for growth models. The proposed AwC method does not require a priori knowledge of noninvariant items and the iterative searching of noninvariant items in typical longitudinal measurement invariance research. Results of a Monte Carlo simulation study comparing AwC with the partial invariance modeling method show that AwC largely reduces biases in growth parameter estimates and gives good control of Type I error rates, especially when the sample size is at least 1,000. It also outperforms the partial invariance method in conditions when all items are noninvariant. However, all methods give biased growth parameter estimates when the proportion of noninvariant parameters is over 25%. Based on the simulation results, I conclude that AO is a viable alternative to the partial invariance method in growth modeling when it is not clear whether longitudinal measurement invariance holds. The current paper also demonstrates AwC in an example modeling neuroticism over three time points using a public data set, which shows how researchers can compute effect size indices for noninvariance in AwC to assess to what degree invariance holds and whether AwC results are trustworthy.

A Bayesian region of measurement equivalence (ROME) approach for establishing measurement invariance.

Measurement invariance research has focused on identifying biases in test indicators measuring a latent trait across two or more groups. However, relatively little attention has been devoted to the practical implications of noninvariance. An important question is whether noninvariance in indicators or items results in differences in observed composite scores across groups. The current study introduces the Bayesian region of measurement equivalence (ROME) as a framework for visualizing and testing the combined impact of partial invariance on the group difference in observed scores. Under the proposed framework, researchers first compute the highest posterior density intervals (HPDIs)-which contain the most plausible values-for the expected group difference in observed test scores over a range of latent trait levels. By comparing the HPDIs with a predetermined range of values that is practically equivalent to zero (i.e., region of measurement equivalence), researchers can determine whether a test instrument is practically invariant. The proposed ROME method can be used for both continuous indicators and ordinal items. We illustrated ROME using five items measuring mathematics-specific self-efficacy from a nationally representative sample of 10th graders. Whereas conventional invariance testing identifies a partial strict invariance model across gender, the statistically significant noninvariant items were found to have a negligible impact on the comparison of the observed scores. This empirical example demonstrates the utility of the ROME method for assessing practical significance when statistically significant item noninvariance is found. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Associations between public and self-stigma of help-seeking with help-seeking attitudes and intention: A meta-analytic structural equation modeling approach.

The present study examined the association between help-seeking public stigma and help-seeking self-stigma (i.e., internalization of stigma) and the relative association of both types of stigma with help-seeking attitude and intention using a full-information meta-analytic structural equation modeling approach. We also investigated the moderating effect of gender, age, collectivism, and social group in the internalization process. Results from 115 independent samples containing data from 54,793 individuals showed that public stigma of help-seeking was strongly and positively associated with self-stigma of help-seeking. Moreover, after controlling for the effect of each other, self-stigma, but not public stigma, remained significantly associated with help-seeking attitude and help-seeking intention. Gender, age, collectivism, and social group did not significantly moderate the association between public and self-stigma. The findings highlight that people who perceive more stigma of help-seeking from others tend to have higher levels of self-stigma. Compared with help-seeking public stigma, help-seeking self-stigma might have a larger impact on one's help-seeking attitude and intention. Help-seeking promotion campaigns should be devised to tackle both types of stigma to foster positive help-seeking attitude and intention. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Leisure-time sedentary behaviors are differentially associated with all-cause dementia regardless of engagement in physical activity.

Sedentary behavior (SB) is associated with cardiometabolic disease and mortality, but its association with dementia is currently unclear. This study investigates whether SB is associated with incident dementia regardless of engagement in physical activity (PA). A total of 146,651 participants from the UK Biobank who were 60 years or older and did not have a diagnosis of dementia (mean [SD] age: 64.59 [2.84] years) were included. Self-reported leisure-time SBs were divided into two domains: time spent watching television (TV) or time spent using a computer. A total of 3,507 individuals were diagnosed with all-cause dementia over a mean follow-up of 11.87 (±1.17) years. In models adjusted for a wide range of covariates, including time spent in PA, time spent watching TV was associated with increased risk of incident dementia (HR [95% CI] = 1.24 [1.15 to 1.32]) and time spent using a computer was associated with decreased risk of incident dementia (HR [95% CI] = 0.85 [0.81 to 0.90]). In joint associations with PA, TV time and computer time remained significantly associated with dementia risk at all PA levels. Reducing time spent in cognitively passive SB (i.e., TV time) and increasing time spent in cognitively active SB (i.e., computer time) may be effective behavioral modification targets for reducing risk of dementia regardless of engagement in PA.

Effects of Psychological Distress and Coping Resources on Internet Gaming Disorder: Comparison between Chinese and Japanese University Students.

The high prevalence of Internet gaming disorder (IGD) among Asian youth indicates an urgent need to identify protective factors and examine their consistency across Asian cultures in order to facilitate cost-effective interventions. Based on the transactional theory of stress and coping, this study collected data of 1243 online gamers (45% males; 18-25 years) through an anonymous survey from universities in China and Japan and investigated whether three coping resources (i.e., mindfulness, coping flexibility, and social support) serve to protect Chinese and Japanese youth from the impact of psychological distress on IGD tendency. After adjusting for the measurement non-invariance across samples, we found that Japanese students reported higher levels of IGD tendency and psychological distress than Chinese students. The results of multiple-group SEM analyses showed that, after controlling for other predictors, mindfulness served as the strongest protective factor against IGD across samples. Moreover, the buffering effect of mindfulness on the association between psychological distress and IGD tendency of female (but not male) students was observed. Our findings highlighted the cross-cultural invariance of the impact of psychological distress and coping resources on IGD in Chinese and Japanese youth, which can be considered in future IGD prevention programs.

Adjusting for partial invariance in latent parameter estimation: Comparing forward specification search and approximate invariance methods.

Measurement invariance is the condition that an instrument measures a target construct in the same way across subgroups, settings, and time. In psychological measurement, usually only partial, but not full, invariance is achieved, which potentially biases subsequent parameter estimations and statistical inferences. Although existing literature shows that a correctly specified partial invariance model can remove such biases, it ignores the model uncertainty in the specification search step: flagging the wrong items may lead to additional bias and variability in subsequent inferences. On the other hand, several new approaches, including Bayesian approximate invariance and alignment optimization methods, have been proposed; these methods use an approximate invariance model to adjust for partial measurement invariance without the need to directly identify noninvariant items. However, there has been limited research on these methods in situations with a small number of groups. In this paper, we conducted three systematic simulation studies to compare five methods for adjusting partial invariance. While specification search performed reasonably well when the proportion of noninvariant parameters was no more than one-third, alignment optimization overall performed best across conditions in terms of efficiency of parameter estimates, confidence interval coverage, and type I error rates. In addition, the Bayesian version of alignment optimization performed best for estimating latent means and variances in small-sample and low-reliability conditions. We thus recommend the use of the alignment optimization methods for adjusting partial invariance when comparing latent constructs across a few groups.

Two-stage path analysis with definition variables: An alternative framework to account for measurement error.

When estimating path coefficients among psychological constructs measured with error, structural equation modeling (SEM), which simultaneously estimates the measurement and structural parameters, is generally regarded as the gold standard. In practice, however, researchers usually first compute composite scores or factor scores, and use those as observed variables in a path analysis, for purposes of simplifying the model or avoiding model convergence issues. Whereas recent approaches, such as reliability adjustment methods and factor score regression, has been proposed to mitigate the bias induced by ignoring measurement error in composite/factor scores with continuous indicators, those approaches are not yet applicable to models with categorical indicators. In this article, we introduce the two-stage path analysis (2S-PA) with definition variables as a general framework for path modeling to handle categorical indicators, in which estimation of factor scores and path coefficients are separated. It thus allows for different estimation methods in the measurement and the structural path models and easier diagnoses of violations of model assumptions. We conducted three simulation studies, ranging from latent regression to mediation analysis with categorical indicators, and showed that 2S-PA generally produced similar estimates to those using SEM in large samples, but gave better convergence rates, less standard error bias, and better control of Type I error rates in small samples. We illustrate 2S-PA using data from a national data set, and show how researchers can implement it in Mplus and OpenMx. Possible extensions and future directions of 2S-PA are discussed. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Modeling Measurement Errors of the Exogenous Composites From Congeneric Measures in Interaction Models.

We investigated the performance of two single indicator methods: latent moderated structural equation (LMS) and reliability-adjusted product indicator (RAPI) methods, on testing interaction effects with congeneric measures, which vary in factor loadings and error variances under a common factor. Additionally, in the simulation study, we compared the performance of four reliability estimates (Cronbach's alpha, omega composite, Coefficient H, and greatest lower bound [GLB]) to adjust for the exogenous composites' measurement errors. Results from the study showed that: while estimating interaction effects with exogenous composites from congeneric measures, the four reliability estimates performed comparably well. Recommendations on the choice of reliability estimates between the LMS and the RAPI methods under different sample sizes and population reliability conditions are further discussed.

The Contributions of Language Skills and Comprehension Monitoring to Chinese Reading Comprehension: A Longitudinal Investigation.

This study examined how vocabulary, syntactic knowledge, and orthographic knowledge are related to comprehension monitoring and whether comprehension monitoring mediates the relations between these language skills and reading comprehension. Eighty-nine Chinese children were assessed on their vocabulary, syntactic knowledge, orthographic knowledge, and comprehension monitoring in Grade 1. Their reading comprehension skills were assessed in Grade 1 and Grade 3. Results showed that in Grade 1, comprehension monitoring mediated the relations between vocabulary and syntactic knowledge and reading comprehension. For Grade 3 reading comprehension, syntactic knowledge in Grade 1 was the only significant predictor. These findings indicate that multiple language skills make direct and indirect contributions via comprehension monitoring to Chinese reading comprehension, and the relations would change as children's reading skills develop.

The Immigrant Paradox in the Problem Behaviors of Youth in the United States: A Meta-analysis.

This meta-analysis synthesizes the empirical data on problem behaviors among foreign- (G1) and U.S-born (G2+) youth and explores the effects of immigrant status on youth internalizing and externalizing problems. A random effects meta-regression with robust variance estimates summarized effect sizes for internalizing and externalizing problems across 91 studies (N = 179,315, Mage  = 13.98). Results indicated that G1 youth reported significantly more internalizing problems (g = .06), and fewer externalizing problems than G2+ youth (g = -.06). Gender and sample type moderated the effects. The findings provide a first-step toward reconciling mixed support for the immigrant paradox by identifying for whom and under what conditions the immigrant experience serves as a risk or protective factor for youth.

Composite reliability of multilevel data: It's about observed scores and construct meanings.

This article shows how the concept of reliability of composite scores, as defined in classical test theory, can be extended to the context of multilevel modeling. In particular, it discusses the contributions and limitations of the various level-specific reliability indices proposed by Geldhof, Preacher, and Zyphur (2014), denoted as ω̃b and ω̃w (and also α̃b and α̃w). One major limitation of those indices is that they are quantities for latent, unobserved level-specific composite scores, and are not suitable for observed composites at different levels. As illustrated using simulated data in this article, ω̃b can drastically overestimate the true reliability of between-level composite scores (i.e., observed cluster means). Another limitation is that the development of those indices did not consider the recent conceptual development on construct meanings in multilevel modeling (Stapleton & Johnson, 2019; Stapleton, Yang, & Hancock, 2016). To address the second limitation, this article defines reliability indices (ω2l, ωb, ωw, α2l, αb, αw) for three types of multilevel observed composite scores measuring various multilevel constructs: individual, configural, shared, and within-cluster. The article also shows how researchers can obtain sample point and interval estimates using the derived formulas and the provided R and Mplus code. In addition, a large-scale national data set was used to illustrate the proposed methods for estimating reliability for the three types of multilevel composite scores, and practical recommendations on when different indices should be reported are provided. (PsycInfo Database Record (c) 2021 APA, all rights reserved).