Mixed-effects Location Scale Model Research Articles

Leveraging mixed-effects location scale models to assess the ERP mismatch negativity’s psychometric properties and trial-by-trial neural variability in toddler-mother dyads

Trial-by-trial neural variability, a measure of neural response stability, has been examined in relation to behavioral indicators using summary measures, but these methods do not characterize meaningful processes underlying variability. Mixed-effects location scale models (MELSMs) overcome these limitations by accounting for predictors and covariates of variability but have been rarely used in developmental studies. Here, we applied MELSMs to the ERP auditory mismatch negativity (MMN), a neural measure often related to language and psychopathology. 84 toddlers and 76 mothers completed a speech-syllable MMN paradigm. We extracted early and late MMN mean amplitudes from trial-level waveforms. We first characterized our sample’s psychometric properties using MELSMs and found a wide range of subject-level internal consistency. Next, we examined the relation between toddler MMNs with theoretically relevant child behavioral and maternal variables. MELSMs offered better model fit than analyses that assumed constant variability. We found significant individual differences in trial-by-trial variability but no significant associations between toddler variability and their language, irritability, or mother variability indices. Overall, we illustrate how MELSMs can characterize psychometric properties and answer questions about individual differences in variability. We provide recommendations and resources as well as example code for analyzing trial-by-trial neural variability in future studies.

A Diffusion Model Account of Cognitive Variability in Healthy Aging and Mild Cognitive Impairment

ABSTRACT Within-person variation in cognitive performance is linked to pathological aging. Cognitive fluctuations have not been analyzed using cognitive process models, such as the diffusion model, to characterize which cognitive processes contribute to variability in cognition. We collected 21 daily assessments of attention and personality in younger adults, healthy older adults, and those with mild cognitive impairment. We employed mixed-effects location scale models to analyze group differences on mean diffusion parameters and daily variability. Discussion focuses on how these methods extend our understanding of how cognitive deficits might appear in aging and disease and the moderating influence of daily personality.

Understanding the biopsychosocial knee osteoarthritis pain experience: an ecological momentary assessment.

Psychological, social, and lifestyle factors contribute to the knee osteoarthritis (OA) pain experience. These factors could be measured more accurately using smartphone ecological momentary assessment (EMA). The objective of this study was to characterise the pain experiences of those with knee OA by a smartphone EMA survey and explain how momentary psychological and social states influence knee OA pain experiences. A smartphone EMA survey was designed and piloted. Eligible participants completed smartphone EMA assessing the knee OA pain experience 3 times daily for 2 weeks. Descriptive statistics were used to characterise factors involved in knee OA pain followed by the development of mixed-effects location scale models to explore heterogeneity and relationships between symptoms involved in the knee OA pain experience. Eighty-six community-dwelling volunteers with knee OA were recruited. Pain, psychosocial, and lifestyle factors involved in knee OA pain experience were heterogeneous and variable. Those with greater variability in pain, fatigue, negative affect, and stress had worse levels of these symptoms overall. In addition, fatigue, negative affect, stress, anxiety, loneliness, and joint stiffness demonstrated within-person relationships with knee OA pain outcomes. Knee OA pain is a heterogeneous biopsychosocial condition. Momentary experiences of psychological, social, fatigue, and joint stiffness explain individual and between-individual differences in momentary knee OA pain experiences. Addressing these momentary factors could improve pain and functional outcomes in those with knee OA. Validation studies, including individuals with more severe knee OA presentations, are required to support findings and guide clinical interventions to improve outcomes for those with knee OA.

Demographic factors associated with within-individual variability of lung function for adults with cystic fibrosis: A UK registry study

Background: Lung function is a key outcome used in the evaluation of disease progression in cystic fibrosis. The variability of individual lung function measurements over time (within-individual variability) has been shown to predict subsequent lung function changes. Nevertheless, the association between within-individual lung function variability and demographic and genetic covariates has not been quantified.Methods: We performed a longitudinal analysis of data from a cohort of 7099 adults with cystic fibrosis (between 18 and 49 years old) from the UK cystic fibrosis registry, containing annual review data between 1996 and 2020. A mixed-effects location-scale model is used to quantify mean FEV1 (forced expiratory volume in 1 s) trajectories and FEV1 within-individual variability as a function of sex, age at annual review, diagnosis after first year of life, homozygous F508 genotype and birth cohort.Results: Mean FEV1 decreased with age and lung function variability showed a near-quadratic trend by age. Males showed higher FEV1 mean and variability than females across the whole age range. Earlier diagnosis and homozygous F508 genotype were also associated with higher FEV1 mean and variability. Individuals who died during follow-up showed on average higher lung function variability than those who survived.Conclusions: Key variables known to be linked with mean lung function in cystic fibrosis are also associated with an individual’s lung function variability. This work opens new avenues to understand the role played by lung function variability in disease progression and its utility in predicting key outcomes such as mortality.

Longitudinal within-person variability around personality trajectories.

Decades of research have identified average patterns of normative personality development across the lifespan. However, it is unclear how well these correspond to trajectories of individual development. Past work beyond general personality development might suggest these average patterns are oversimplifications, necessitating novel examinations of how personality develops and consideration of new individual difference metrics. This study uses five longitudinal data sets from Germany, Australia, the Netherlands, and the United States (N = 128,345; Mage = 45.42; 53% female) to examine personality development using mixed-effects location scale models. These models quantify individual differences in within-person residual variability, or sigma, around trajectories-thereby testing if models that assume sigma is homogeneous, unsystematic noise are appropriate. We investigate if there are individual differences in longitudinal within-person variability for Big Five trajectories, if there are variables associated with this heterogeneity, and if person-level sigma values can uniquely predict an outcome. Results indicated that, across all models, there was meaningful heterogeneity in sigma-the magnitude of which was comparable to and often even greater than that of intercepts and slopes. Individual differences in sigma were further associated with covariates central to personality development and had robust predictive utility for health status, an outcome with long-established personality associations. Collectively, these findings underscore the presence, degree, validity, and potential utility of heterogeneity in longitudinal within-person variability and indicate the typical linear model does not adequately depict individual development. We suggest it should become the default to consider this individual difference metric in personality development research. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Pleasure and Satisfaction as Predictors of Future Cigarette and E-cigarette Use: A Novel Two-Stage Modeling Approach.

Subjective experience of e-cigarettes may be an important factor in helping people who use combustible cigarettes switch completely to e-cigarettes to reduce harm from smoking. This paper describes a novel two-stage analysis using pleasure and satisfaction responses from ecological momentary assessments (EMA) of both cigarette and e-cigarette use to predict future cigarette and e-cigarette tobacco use. This observational study included adult users of cigarettes and e-cigarettes who provided 7 days of EMA, capturing cigarette and e-cigarette use, followed by biweekly reports of cigarette and e-cigarette use over 1 year. Participants were 279 adults who provided both cigarette and e-cigarette responses during the EMA. We employed a two-stage analytic approach in which EMA data were used to predict subsequent levels of cigarette and e-cigarette use. In the first stage, EMA responses to cigarette and e-cigarette events were modeled via a mixed-effects location scale model to yield summaries of participants' means and variability on event-related ratings of pleasure and satisfaction. These EMA summaries served as predictors in the second stage analysis of the biweekly post-EMA longitudinal cigarette and e-cigarette use data. EMA pleasure and satisfaction ratings were similar for both products and predicted both longitudinal cigarette and e-cigarette use, even after controlling for baseline cigarette and e-cigarette dependence. Relatively higher levels of satisfaction with e-cigarettes were associated with greater decreases in cigarette use over time. Pleasure and satisfaction are important predictors of subsequent cigarette and e-cigarette use. Experienced subjective pleasure and satisfaction from e-cigarettes relative to cigarettes may be an important factor in helping individuals who smoke to switch completely to e-cigarettes as a harm reduction approach. In order to help sustain complete product switching and reduce dual use or relapse to smoking, e-cigarettes may need to deliver more satisfaction to the user compared to that experienced from cigarettes.

Big Five traits predict between- and within-person variation in loneliness

Past research has linked individual differences in loneliness to Big Five personality traits. However, experience sampling studies also show intrapersonal fluctuations in loneliness. These may reflect situational factors as well as stable individual differences. Here, for the first time, we study the relationship between personality traits and within-person variation in loneliness. In a one-week experience sampling study, n = 285 Nepali participants reported feelings of loneliness three times a day (3597 observations). We use Bayesian mixed-effects location scale models to simultaneously estimate the relationship between Big Five personality traits and (a) mean levels and (b) within-person variability in loneliness. We also test whether these relationships vary depending on whether participants were alone or in the company of others. More neurotic individuals felt lonelier, especially (but not only) when they were alone. These individuals also experienced greater intrapersonal fluctuations in loneliness. These findings extend the differential reactivity hypothesis, according to which individuals vary in loneliness due to differential reactivity to social situations, and accord with the conceptual view of neuroticism as hyperreactivity to social stressors. In addition, we document the role of personality and social context in people’s everyday experience of loneliness in a non-WEIRD population.

Addressing Heteroscedasticity in Correlated Binary Data: A Bayesian Mixed Effects Location Scale Approach

Introduction: The mixed effects logistic regression model is a common model for analysing correlated binary data as longitudinal data. The between and within subject variances are typically considered to be homogeneous but longitudinal data often show heterogeneity in these variances. This study proposes a Bayesian mixed effects location scale model to accommodate heteroscedasticity in binary data analysis.
 Methods: This study was carried out in two stages; first, the simulation study was used to evaluate the accuracy of the proposed model with the Bayesian approach and then the proposed model was applied to a real data. In simulation study, the data were generated from the mixed effects location scale model with different correlations between the random location effect and random scale effect and different sample sizes. In order to evaluate the accuracy of the estimations, the Root Mean Square Error, bias and Coverage Probability were calculated and the deviance information criterion was used to select the appropriate model. At the end we utilized this model to analyse uric acid levels of patients with haematological disorders.
 Results: The simulation results show the accuracy of model parameter estimates as well as the correlation between random location and scale effects. They also display that if a random scale effect is present in the data, it should be accounted for in model. Otherwise, the model is forced to assign the within subject variation due to these subject random effects to the error term. The results of real data are also in line with this. The odds of having normal UA levels increases by a factor of 26% per week. Due to the positive value of the covariance parameter, patients with higher mean of UA levels show higher variation in UA levels. Furthermore, the significance of the covariates in the between subject and within subject variances model, as well as the significance of the random scale variance determines the heterogeneity across subjects.
 Conclusion: Bayesian mixed effects location scale model provides a useful tool for analysing correlated binary data with heteroscedasticity because it considers data correlation and modelling mean and variance simultaneously. Furthermore, it improves the accuracy of statistical inference in longitudinal studies compared to classic mixed effects models.

Relationships between hourly cognitive variability and risk of Alzheimer's disease revealed with mixed-effects location scale models.

Observational studies on aging and Alzheimer's disease (AD) typically focus on mean-level changes in cognitive performance over relatively long periods of time (years or decades). Additionally, some studies have examined how trial-level fluctuations in speeded reaction time are related to both age and AD. The aim of the current project was to describe patterns of variability across repeated days of testing as a function of AD risk in cognitively normal older adults. The current project examined the performance of the Ambulatory Research in Cognition (ARC) smartphone application, a high-frequency remote cognitive assessment paradigm, that administers brief tests of episodic memory, spatial working memory, and processing speed. Bayesian mixed-effects location scale models were used to explore differences in mean cognitive performance and intraindividual variability across 28 repeated sessions over a 1-week assessment interval as function of age and genetic risk of AD, specifically the presence of at least one apolipoprotein E (APOE) ε4 allele. Mean performance on processing speed and working memory was negatively related to age and APOE status. More importantly, e4 carriers exhibited increased session-level variability on a test of processing speed compared to noncarriers. Age and education did not consistently relate to cognitive variability, contrary to expectations. Preclinical AD risk, defined as possessing at least one APOE ε4 allele, is not only associated with mean-level performance differences, but also with increases in variability across repeated testing occasions particularly on a test of processing speed. Thus, cognitive variability may serve as an additional and important indicator of AD risk. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Mixed-Effects Location Scale Models for Joint Modeling School Value-Added Effects on the Mean and Variance of Student Achievement

School value-added models are widely applied to study, monitor, and hold schools to account for school differences in student learning. The traditional model is a mixed-effects linear regression of student current achievement on student prior achievement, background characteristics, and a school random intercept effect. The latter is referred to as the school value-added score and measures the mean student covariate-adjusted achievement in each school. In this article, we argue that further insights may be gained by additionally studying the variance in this quantity in each school. These include the ability to identify both individual schools and school types that exhibit unusually high or low variability in student achievement, even after accounting for differences in student intakes. We explore and illustrate how this can be done via fitting mixed-effects location scale versions of the traditional school value-added model. We discuss the implications of our work for research and school accountability systems.

High-frequency assessment of mood, personality, and cognition in healthy younger, healthy older and adults with cognitive impairment

ABSTRACT Increased variability in cognitive scores, mood or personality traits can be indicative of underlying neurological disorders. Whether variability in cognition is due to changes in mood or personality is unknown. A total of 66 younger adults, 51 healthy older adults and 38 participants with cognitive impairment completed 21 daily sessions of attention, working memory, mood, and personality assessment. Group differences in mean performance and variability were examined using Bayesian mixed effects location scale models. Variability in attention decreased from younger to older adults and then increased again in cognitive impairment. Younger adults were more variable in agreeableness, openness and conscientiousness compared to older adults. The clinically impaired group differed from the healthy older adults in terms of variability on attention, openness, and conscientiousness. Healthy aging results in greater stability in personality traits over short intervals yet this stability is not redundant with increased stability in cognitive scores.

Moody and in love? The role of neuroticism and romantic relationships for momentary affect in adolescence

The affective lives of adolescents are unique in that momentary affect in this age group is more negative and variable. This study examined how neuroticism and romantic relationships (i.e., relationship involvement and relationship quality) relate to adolescents’ daily affective experiences. In a weeklong experience sampling period, 408 German adolescents ( MAge = 16.83) reported up to five times per day on their positive and negative affect. We estimated mixed-effects location scale models to analyze interindividual differences in adolescents’ affect level and variability. Adolescents with higher neuroticism experienced lower levels of positive affect, higher levels of negative affect, and higher variability of positive and negative affect. Whereas adolescents with a romantic partner did not differ from their single peers with regard to affect level, they experienced higher affect variability, although evidence for these effects was weak. Finally, among adolescents who were currently involved in a romantic relationship, those with higher relationship quality experienced more variability in their positive affect if they scored higher in neuroticism. Across models, effect sizes systematically differed between affect level and variability, positive and negative affect, as well as neuroticism facets. We discuss these findings in light of adolescents’ affective dynamics, affective development, and personality-social relationship interactions.

A latent variable mixed-effects location scale model that also considers between-person differences in the autocorrelation.

In public health research an increasing number of studies is conducted in which intensive longitudinal data is collected in an experience sampling or a daily diary design. Typically, the resulting data is analyzed with a mixed-effects model or mixed-effects location scale model because they allow one to examine a host of interesting longitudinal research questions. Here, we introduce an extension of the mixed-effects location scale model in which measurement error of the observed variables is considered by a latent factor model and in which-in addition to the mean-or location-related effects-the residual variance of the latent factor and the parameters of the autoregressive process of this latent factor can differ between persons. We show how to estimate the parameters of the model with a maximum likelihood approach, whose performance is also compared with a Bayesian approach in a small simulation study. We illustrate the models using a real data example and end with a discussion in which we suggest questions for future research.

Detecting influential subjects in intensive longitudinal data using mixed-effects location scale models

BackgroundCollection of intensive longitudinal health outcomes allows joint modeling of their mean (location) and variability (scale). Focusing on the location of the outcome, measures to detect influential subjects in longitudinal data using standard mixed-effects regression models (MRMs) have been widely discussed. However, no existing approach enables the detection of subjects that heavily influence the scale of the outcome.MethodsWe propose applying mixed-effects location scale (MELS) modeling combined with commonly used influence measures such as Cook’s distance and DFBETAS to fill this gap. In this paper, we provide a framework for researchers to follow when trying to detect influential subjects for both the scale and location of the outcome. The framework allows detailed examination of each subject’s influence on model fit as well as point estimates and precision of coefficients in different components of a MELS model.ResultsWe simulated two common scenarios in longitudinal healthcare studies and found that influence measures in our framework successfully capture influential subjects over 99% of the time. We also re-analyzed data from a health behavior study and found 4 particularly influential subjects, among which two cannot be detected by influence analyses via regular MRMs.ConclusionThe proposed framework can help researchers detect influential subject(s) that will be otherwise overlooked by influential analysis using regular MRMs and analyze all data in one model despite influential subjects.

Investigating the day-level associations between affective variability and physical activity using ecological momentary assessment

BackgroundUnderstanding affect as a determinant of physical activity has gained increased attention in health behavior research. Fluctuations in affect intensity from moment-to-moment (i.e., affective variability) may interfere with cognitive and regulatory processes, making it difficult to engage in goal-directed behaviors such as physical activity. Preliminary evidence indicates that those with greater trait-level affective variability engage in lower levels of habitual physical activity. However, the extent to which daily fluctuations in affect variability are associated with same-day physical activity levels is unknown. This study used ecological momentary assessment (EMA) to investigate day-level associations between affective variability (i.e., within-subject variance) and physical activity. MethodsYoung adults (N = 231, M = 23.58 ± 3.02 years) provided three months of smartphone-based EMA and smartwatch-based activity data. Every two weeks, participants completed a 4-day EMA measurement burst (M = 5.17 ± 1.28 bursts per participant). Bursts consisted of hourly randomly-prompted EMA surveys assessing momentary positive-activated (happy, energetic), positive-deactivated (relaxed), negative-activated (tense, stressed), and negative-deactivated (sad, fatigued) affect. Participants continuously wore a smartwatch to measure physical activity across the three months. Mixed-effects location scale modeling examined the day-level associations of affective variability (i.e., positive-activated, positive-deactivated, negative-activated, and negative-deactivated) and physical activity, controlling for covariates such as mean levels of affect, between-subject effects of physical activity, time of day, day of week, day in study, and smartwatch wear time. ResultsThere were 41,546 completed EMA surveys (M = 182.22 ± 69.82 per participant) included in the analyses. Above and beyond mean levels of affect, greater day-level variability in positive-activated affect was associated with greater physical activity on that same day compared to other days (τ = 0.01, p < .001), whereas greater day-level variability in negative-deactivated affect was associated with less physical activity on that same day compared to other days (τ = −0.01, p < .001). Day-level variability in positive-deactivated affect or negative-activated affect were not associated with day-level physical activity (ps > .05) ConclusionsIndividuals were less active on days with greater variability in feeling sad and fatigued but more active on days with greater variability in feeling happy and energetic. Understanding the dynamic relationships of affective variability with day-level physical activity can strengthen physical activity interventions by considering how these processes differ within individuals and unfold within the context of daily life. Future research should examine causal pathways between affective variability and physical activity across the day.

Studying Movement-Related Behavioral Maintenance and Adoption in Real Time: Protocol for an Intensive Ecological Momentary Assessment Study Among Older Adults.

Older adults struggle to maintain newly initiated levels of physical activity (PA) or sedentary behavior (SB) and often regress to baseline levels over time. This is partly because health behavior theories that inform interventions rarely address how the changing contexts of daily life influence the processes regulating PA and SB or how those processes differ across the behavior change continuum. Few studies have focused on motivational processes that regulate the dynamic nature of PA and SB adoption and maintenance on microtimescales (ie, across minutes, hours, or days). The overarching goal of Project Studying Maintenance and Adoption in Real Time (SMART) is to determine the motivational processes that regulate behavioral adoption versus maintenance over microtimescales, using a dual process framework combined with ecological momentary assessment and sensor-based monitoring of behavior. This paper describes the recruitment, enrollment, data collection, and analytics protocols for Project SMART. In Project SMART, older adults engaging in at least 30 minutes of moderate-to-vigorous intensity PA per week complete 3 data collection periods over 1 year, with each data collection period lasting 14 days. Across each data collection period, participants wear an ActiGraph GT3X accelerometer (ActiGraph, LLC) on their nondominant waist and an ActivPAL micro4 accelerometer (PAL Technologies, Ltd) on their anterior thigh to measure PA and SB, respectively. Ecological momentary assessment questionnaires are randomly delivered via smartphone 10 times per day on 4 selected days in each data collection period and assess reflective processes (eg, evaluating one's efficacy and exerting self-control) and reactive processes (eg, contextual cues) within the dual process framework. At the beginning and end of each data collection period, participants complete a computer-based questionnaire to learn more about their typical motivation for PA and SB, physical and mental health, and life events over the course of the study. Recruitment and enrollment began in January 2021; enrollment in the first data collection period was completed by February 2022; and all participants completed their second and third data collection by July 2022 and December 2022, respectively. Data were collected from 202 older adults during the first data collection period, with approximate retention rates of 90.1% (n=182) during the second data collection period and 88.1% (n=178) during the third data collection period. Multilevel models and mixed-effects location scale modeling will be used to evaluate the study aims. Project SMART seeks to predict and model the adoption and maintenance of optimal levels of PA and SB among older adults. In turn, this will inform the future delivery of personalized intervention content under conditions where the content will be most effective to promote sustained behavior change among older adults. DERR1-10.2196/47320.

Mixed-effects location-scale model based on generalized hyperbolic distribution

Motivated by better modeling of intra-individual variability in longitudinal data, we propose a class of location-scale mixed-effects models, in which the data of each individual is modeled by a parameter-varying generalized hyperbolic distribution. We first study the local maximum-likelihood asymptotics and reveal the instability in the numerical optimization of the log-likelihood. Then, we construct an asymptotically efficient estimator based on the Newton–Raphson method based on the original log-likelihood function with the initial estimator being naive least-squares-type. Numerical experiments are conducted to show that the proposed one-step estimator is not only theoretically efficient but also numerically much more stable and much less time-consuming compared with the maximum-likelihood estimator.

Challenges Experienced by Black Women with Breast Cancer During Active Treatment: Relationship to Treatment Adherence.

Although rates of death from breast cancer have declined in the USA for both Black and White women since 1990, mortality rates for Black women remain strikingly higher - 40% higher compared to White women (American Cancer Society 1). The barriers and challenges that may be triggering unfavorable treatment-related outcomes and diminished treatment adherence among Black women are not well understood. We recruited 25 Black women with breast cancer who were to receive surgery and chemotherapy and/or radiation therapy. Through weekly electronic surveys, we assessed types and severity of challenges across various life domains. Because the participants rarely missed treatments or appointments, we examined the impact of severity of weekly challenges on thoughts of skipping treatment or appointment with their cancer care team using a mixed-effects location scale model. Both a higher average severity of challenges and a higher deviation of severity reported across weeks were associated with increased thoughts on skipping treatment or appointment. The correlation between the random location and scale effects was positive; thus, those women that reported more thoughts on skipping a dose of medicine or appointment were also more unpredictable with respect to the severity of challenges reported. Black women with breast cancer are impacted by familial, social, work-related, and medical care factors, and these may in turn affect adherence to treatment. Providers are encouraged to actively screen and communicate with patients regarding life challenges and to build networks of support within the medical care team and social community that can help patients successfully complete treatment as planned.

Heterogeneous heterogeneity by default: Testing categorical moderators in mixed-effects meta-analysis.

Categorical moderators are often included in mixed-effects meta-analysis to explain heterogeneity in effect sizes. An assumption in tests of categorical moderator effects is that of a constant between-study variance across all levels of the moderator. Although it rarely receives serious thought, there can be statistical ramifications to upholding this assumption. We propose that researchers should instead default to assuming unequal between-study variances when analysing categorical moderators. To achieve this, we suggest using a mixed-effects location-scale model (MELSM) to allow group-specific estimates for the between-study variance. In two extensive simulation studies, we show that in terms of Type I error and statistical power, little is lost by using the MELSM for moderator tests, but there can be serious costs when an equal variance mixed-effects model (MEM) is used. Most notably, in scenarios with balanced sample sizes or equal between-study variance, the Type I error and power rates are nearly identical between the MEM and the MELSM. On the other hand, with imbalanced sample sizes and unequal variances, the Type I error rate under the MEM can be grossly inflated or overly conservative, whereas the MELSM does comparatively well in controlling the Type I error across the majority of cases. A notable exception where the MELSM did not clearly outperform the MEM was in the case of few studies (e.g., 5). With respect to power, the MELSM had similar or higher power than the MEM in conditions where the latter produced non-inflated Type 1 error rates. Together, our results support the idea that assuming unequal between-study variances is preferred as a default strategy when testing categorical moderators.

Model construction and sample size planning for mixed-effects location-scale models

Model construction and sample size planning for mixed-effects location-scale models