Variance Estimation Research Articles

Inclusion of imputed genotypes from non-genotyped dairy cattle in a Thai multibreed genomic-polygenic evaluation.

This study assessed the impact of incorporating imputed SNP information from non-genotyped animals on genomic-polygenic evaluations in a Thai multibreed dairy population under various levels of imputation accuracy. Data encompassed pedigree and phenotypic records for 305-day milk yield (MY), 305-day fat (Fat), and age at first calving (AFC) from 12,859 first-lactation cows, and genotypic records of various densities from 4,364 animals. A set of 64 animals genotyped with GeneSeek Genomic Profiler 80K and with four or more genotyped progenies was defined as target animals to simulate imputation scenarios for non-genotyped individuals. Actual and imputed genotypes were utilized to construct three SNP sets. All SNP Sets contained actual and imputed SNP markers from genotyped animals. SNP Set 1 contained no SNPs from target animals, whereas SNP Set 2 incorporated imputed SNPs from target animals, and SNP Set 3 added actual SNPs from target animals. Genomic-polygenic evaluations were conducted using a 3-trait single-step model that included contemporary group, calving age, and heterozygosity as fixed effects and animal additive genetic and residual as random effects. The imputation accuracy was similar across non-genotyped animals irrespective of the number of genotyped progenies (average: 40.55%; range: 34.68% to 53.82%). Estimates of additive genetic and environmental variances and covariances for MY and AFC varied across SNP sets. SNP Sets 1 and 2 had slightly higher additive genetic and lower environmental variances and covariances than SNP Set 3. Heritabilities and additive genetic, environmental, and phenotypic correlations between MY, Fat, and AFC were similar across all SNP Sets. Spearman rank correlations between genomic-polygenic EBVs from SNP Sets 2 and 3 were high for all traits (0.9990±0.0003). Utilization of phenotypic and pedigree data from imputed non-genotyped animals enhanced the efficiency and cost-effectiveness of the genetic improvement program in the Thai multibreed dairy cattle population.

Improved Variance Estimation From Trimmed Samples

ABSTRACTSeveral methods for the robust estimation of the variance of a normal random variable are based on trimming. For example, the popular estimator is based on trimming the pairs of samples with the largest absolute distance. Another example is least trimmed squares regression, where all unusually large residuals are removed before the estimation of the variance. Here, in this work, we propose two new classes of estimators that use an optimal linear combination of the trimmed samples to achieve a lower mean squared error (MSE). The first class guarantees the smallest variance among all unbiased estimators that are based on linear combinations, while the second class of estimators guarantees the smallest MSE at the cost of some small bias. Through simulations, we demonstrate that our proposed estimators can have considerably smaller MSE than other robust estimators both in the presence and absence of outliers.

Direct Estimation for Commonly Used Pattern-Mixture Models in Clinical Trials

Pattern-mixture models have received increasing attention as they are commonly used to assess treatment effects in primary or sensitivity analyses for clinical trials with nonignorable missing data. Pattern-mixture models have traditionally been implemented using multiple imputation, where the variance estimation may be a challenge because the Rubin’s approach of combining between- and within-imputation variance may not provide consistent variance estimation while bootstrap methods may be time-consuming. Direct likelihood-based approaches have been proposed in the literature and implemented for some pattern-mixture models, but the assumptions are sometimes restrictive and the theoretical framework is fragile. In this article, we propose an analytical framework for an efficient direct likelihood estimation method for commonly used pattern-mixture models corresponding to return-to-baseline, jump-to-reference, placebo washout, and retrieved dropout imputations. A parsimonious tipping point analysis is also discussed and implemented. Results from simulation studies demonstrate that the proposed methods provide consistent estimators. We further illustrate the utility of the proposed methods using data from a clinical trial evaluating a treatment for type 2 diabetes.

Two-stage pseudo maximum likelihood estimation of semiparametric copula-based regression models for semi-competing risks data.

We propose a two-stage estimation procedure for a copula-based model with semi-competing risks data, where the non-terminal event is subject to dependent censoring by the terminal event, and both events are subject to independent censoring. With a copula-based model, the marginal survival functions of individual event times are specified by semiparametric transformation models, and the dependence between the bivariate event times is specified by a parametric copula function. For the estimation procedure, in the first stage, the parameters associated with the marginal of the terminal event are estimated using only the corresponding observed outcomes, and in the second stage, the marginal parameters for the non-terminal event time and the copula parameter are estimated together via maximizing a pseudo-likelihood function based on the joint distribution of the bivariate event times. We derived the asymptotic properties of the proposed estimator and provided an analytic variance estimator for inference. Through simulation studies, we showed that our approach leads to consistent estimates with less computational cost and more robustness than the one-stage procedure developed in Chen YH (Lifetime Data Anal 18:36-57, 2012), where all parameters were estimated simultaneously. In addition, our approach demonstrates more desirable finite-sample performances over another existing two-stage estimation method proposed in Zhu H et al., (Commu Statistics-Theory Methods 51(22):7830-7845, 2021) . An R package PMLE4SCR is developed to implement our proposed method.

Testing for Equal Average Forecast Accuracy in Possibly Unstable Environments *

We consider the issue of testing the null of equal average forecast accuracy in a model where the forecast error loss differential series has a potentially non-constant mean function over time. We show that when time variation is present in the loss differential mean, the standard Diebold and Mariano (1995) test, which was proposed for evaluating forecasts in a stable environment, has an asymptotic size of zero, and, whilst consistent, can have reduced local power. This arises due to inconsistent estimation of the implicit long run variance estimator, which diverges under a time varying mean. We suggest a modified statistic that replaces the standard long run variance estimator based on full-sample demeaning of the loss differential series with one based on nonparametric local demeaning. The new long run variance estimator is consistent under both the null and alternative when the mean function is time varying or constant, and in both cases, the modified test recovers the asymptotic size and power properties associated with the original test in the constant mean case. The modified test therefore provides a robust method for testing the equal average forecast accuracy null, allowing for instability in the loss differential mean. The benefits of our test are demonstrated via Monte Carlo simulation and two empirical applications.

A Separable Bootstrap Variance Estimation Algorithm for Hierarchical Model‐Based Inference of Forest Aboveground Biomass Using Data From NASA's GEDI and Landsat Missions

ABSTRACTThe hierarchical model‐based (HMB) statistical method is currently applied in connection with NASA's Global Ecosystem Dynamics Investigation (GEDI) mission for assessing forest aboveground biomass (AGB) in areas lacking a sufficiently large number of GEDI footprints for employing hybrid inference. This study focuses on variance estimation using a bootstrap procedure that separates the computations into parts, thus considerably reducing the computational time required and making bootstrapping a viable option in this context. The procedure we propose uses a theoretical decomposition of the HMB variance into two parts. Through this decomposition, each variance component can be estimated separately and simultaneously. For demonstrating the proposed procedure, we applied a square‐root‐transformed ordinary least squares (OLS) model, and parametric bootstrapping, in the first modeling step of HMB. In the second step, we applied a random forest model and pairwise bootstrapping. Monte Carlo simulations showed that the proposed variance estimator is approximately unbiased. The study was performed on an artificial copula‐generated population that mimics forest conditions in Oregon, USA, using a dataset comprising AGB, GEDI, and Landsat variables.

Social Relationships and Tooth Loss in Adults Aged 60 Years and Older: A Systematic Review and Meta-Analysis.

To systematically assess current evidence on the extent to which social relationships are associated with tooth loss in adults aged 60 years and older. A systematic literature search was conducted on PubMed, Embase, Web of Science, CINAHL and The Cochrane Library databases to identify relevant studies published from 1966 up to March 2024. Cross-sectional or cohort studies investigating the association between structural, functional and/or combined (structural and functional) components of social relationships and the number of remaining teeth or edentulism among community-dwelling or institutionalised older adults were included. Data were extracted on participants' and study characteristics, including study design, the type of measures used to assess social relationships (structural, functional, and combined), outcome measures and association estimates. The quality of the studies was assessed using the Newcastle-Ottawa Scale (NOS) for cohort studies and the adapted NOS for cross-sectional studies. The reported association between social relationships and the number of remaining teeth or edentulism was summarised using meta-analysis with robust variance estimation. Twenty studies were included in the review and 12 studies (125 553 participants) in the meta-analysis. Across the 12 studies, the average odds ratio (95% confidence interval) was 1.15 (1.01-1.32), indicating a 15% higher likelihood of having a lower number of teeth or edentulism for those with weaker social relationships. The GRADE certainty of the body of evidence was low. Weak social relationships were associated with a lower number of teeth or edentulism in older adults. Our findings may inform potential public health approaches that target and modify social relationships to prevent and address older adults' oral diseases. Still, the directionality and the underlying mechanisms connecting social relationships and tooth loss need to be further explored by longitudinal studies with follow-up long enough for oral health outcomes or changes in social relationships to occur. Protocol Registration: PROSPERO (CRD42023417845).

A General Method for Resampling Autocorrelated Spatial Data

AbstractComparing spatial data sets is a ubiquitous task in data analysis, however the presence of spatial autocorrelation means that standard estimates of variance will be wrong and tend to over‐estimate the statistical significance of correlations and other observations. While there are a number of existing approaches to this problem, none are ideal, requiring detailed analytical calculations, which are hard to generalize or detailed modeling of the data generating process, which may not be straightforward. In this work we propose an approach based on permuting or resampling at fixed spatial autocorrelation, measured by Moran's I, in order to generate a null model that accounts for spatial dependence. Testing on real and synthetic data, we find that, as long as the spatial autocorrelation is not too strong, this approach works just as well as if we knew the data generating process exactly and allows us to compute P‐values with the correct Type‐I error rate.

Group-based versus individual parenting programs: A meta-analysis of effects on parents.

Parenting programs aim to motivate change in parents' behaviors, perceptions, and knowledge to promote desirable changes in children's adjustment and behavior. Parenting programs are typically delivered in a group-based or individual format, and it is unknown which format most effectively supports parents. On the one hand, group-based programs may be more effective since they provide access to peer-based support. On the other hand, individual programs may be more successful since they offer a tailored approach. Therefore, this meta-analysis investigated which delivery format is most effective in optimizing child behavior management (i.e., positive reinforcement and nonviolent discipline), parenting stress, and parental depressive symptoms. Studies were selected from the systematic review by Backhaus et al. (2023). Our sample included evaluations of 121 group-based and 41 individual programs based on social learning theory principles. Robust variance estimation indicated that parents' child behavior management and parenting stress improved in both delivery formats. In contrast, parental depressive symptoms improved only in group-based programs. Our findings suggest that whether parents benefit more from a group-based or individual program in part depends on the outcome examined. Therefore, it is important for clinicians to understand parents' goals to effectively guide them toward the most appropriate parenting program delivery format. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Reaching and stepping respond differently to medication and cueing in Parkinson’s disease

The basal ganglia contribute to internal timekeeping, and dopaminergic medication has been observed to moderate timing deficits associated with Parkinson’s Disease (PD) during single joint movements. However, it is unclear whether similar effects can be observed in multi-joint movements. Twenty-five people with PD and twelve healthy peers performed repetitive reaching and stepping-in-place tasks with and without auditory cues at their self-selected maximal cadence. The PD group was measured ON and OFF medication. Reduced cadence error was found for both groups and tasks when cued, and ON PD exhibited decreased cadence compared to OFF PD. Overall timing variability was no different from controls, but differences were found in estimates of clock and motor variance using the Wing-Kristofferson model of interval timing. A medication and cueing interaction during the reaching task produced increased clock variance in uncued, ON PD. During the stepping task, clock and motor variance of the PD group were unaffected by cues, in contrast to the control group. Serial lag-one correlation was reduced in both groups for cued reaching, but was unaffected by cueing or medication in the PD group when stepping-in-place. These findings suggest that overall timing variability may not capture timing deficits in PD.

Quantitative genetics and breeding strategies in hybrid populations of Pinus patula × Pinus tecunumanii

Forestry companies SAFCOL and Sappi conducted a study in South Africa to examine genetic variation and select improved material in full-sib hybrid families of Pinus patula × Pinus tecunumanii. The results from the study on clonal and seedling hybrid populations of P. patula × P. tecunumanii low-elevation subpopulations (TECL) revealed a significantly larger impact of TECL compared to P. patula. The TECL parents exhibited higher general hybridising ability (GHA) variance for volume gain and modulus of elasticity (MOE). Consequently, future hybrid crosses should focus on increasing the number of TECL parents in the breeding programme to improve these traits. In the hybrid seedling population of P. patula × P. tecunumanii high-elevation subpopulation (TECH), P. patula exerted a substantial influence on volume gain and MOE. However, the TECH parents displayed zero estimates of GHA variance for volume and a low GHA variance for MOE, relative to GHA variance among P. patula parents for MOE. Thus, this study recommends testing more P. patula parents in new intercrosses of this variety to enhance volume and MOE gain.

A new modified estimator of population variance in calibrated survey sampling

In survey statistics, estimating and reducing population variation is crucial. These variations can occur in any sampling design, including stratified random sampling, where stratum weights may increase the variance of estimators. Calibration techniques, which use additional auxiliary information, can help mitigate this issue. This paper examines three calibration-based estimators—calibration variance, calibration ratio, and calibration exponential ratio estimators—within the framework of stratified random sampling. The study generates data from normal, gamma, and exponential distributions to test these estimators. Results demonstrate that the proposed calibration estimators offer more accurate estimates of population variance and outperform existing methods in estimating population variance under stratified random sampling, providing more accurate and reliable estimates.

Design-Based Theory for Lasso Adjustment in Randomized Block Experiments and Rerandomized Experiments

Blocking, a special case of rerandomization, is routinely implemented in the design stage of randomized experiments to balance the baseline covariates. This study proposes a regression adjustment method based on the least absolute shrinkage and selection operator (Lasso) to efficiently estimate the average treatment effect in randomized block experiments with high-dimensional covariates. We derive the asymptotic properties of the proposed estimator and outline the conditions under which this estimator is more efficient than the unadjusted one. We provide a conservative variance estimator to facilitate valid inferences. Our framework allows one treated or control unit in some blocks and heterogeneous propensity scores across blocks, thus including paired experiments and finely stratified experiments as special cases. We further accommodate rerandomized experiments and a combination of blocking and rerandomization. Moreover, our analysis allows both the number of blocks and block sizes to tend to infinity, as well as heterogeneous treatment effects across blocks without assuming a true outcome data-generating model. Simulation studies and two real-data analyses demonstrate the advantages of the proposed method.

Speeding up interval estimation for R2-based mediation effect of high-dimensional mediators via cross-fitting.

Mediation analysis is a useful tool in investigating how molecular phenotypes such as gene expression mediate the effect of exposure on health outcomes. However, commonly used mean-based total mediation effect measures may suffer from cancellation of component-wise mediation effects in opposite directions in the presence of high-dimensional omics mediators. To overcome this limitation, we recently proposed a variance-based R-squared total mediation effect measure that relies on the computationally intensive nonparametric bootstrap for confidence interval estimation. In the work described herein, we formulated a more efficient two-stage, cross-fitted estimation procedure for the R2 measure. To avoid potential bias, we performed iterative Sure Independence Screening (iSIS) in two subsamples to exclude the non-mediators, followed by ordinary least squares regressions for the variance estimation. We then constructed confidence intervals based on the newly derived closed-form asymptotic distribution of the R2 measure. Extensive simulation studies demonstrated that this proposed procedure is much more computationally efficient than the resampling-based method, with comparable coverage probability. Furthermore, when applied to the Framingham Heart Study, the proposed method replicated the established finding of gene expression mediating age-related variation in systolic blood pressure and identified the role of gene expression profiles in the relationship between sex and high-density lipoprotein cholesterol level. The proposed estimation procedure is implemented in R package CFR2M.

Understanding the mechanisms of change in social norms around tobacco use: A systematic review and meta-analysis of interventions.

Tobacco use spreads through social networks influencing social norms around tobacco use. However, the social norms scholarship is extremely diverse and occasionally conflicting, complicating efforts to understand how best to leverage social norms to reduce tobacco use. This study systematically reviewed and meta-analyzed this vast terrain by focusing on social norms measurement and mechanism, and intervention effectiveness and modality aimed at changing social norms around tobacco use and actual tobacco use. We searched Scopus, PubMed, PsycInfo, Clinicaltrials.gov, ProQuest Dissertations, the Cochrane Trial Registry, as well as the websites of the Society for Research on Nicotine and Tobacco, the Open Science Framework, medrXiv and the Truth Initiative for experimental and quasi-experimental evaluation studies of interventions designed to shift tobacco use. We included studies written in English from inception to 30 May 2024. We only included studies which noted social norms or social influence as part of the intervention design or set of measured variables. We excluded studies with only one time point, without an intervention being evaluatedand those not published in English. Study screening followed the PRISMA 2020 guidelines, and was conducted by at least two independent reviewers who resolved discrepancies through discussion and consensus. All included studies were analyzed in a narrative synthesis, and those providing sufficient statistics for tobacco and social norms outcomes were included in meta-analyses, which were performed separately for tobacco outcomes and social norms outcomes. Study outcomes were transformed into a standardized mean difference (Hedges' g) and several meta-regressions were fit to explore sources of heterogeneity using a robust variance estimation specification to handle effect size dependency. A total of 95 studies met inclusion criteria for the narrative synthesis, 200 effect sizes from 86 studies were included in the tobacco outcomes meta-analysis, and 66 effect sizes from 29 studies were included in the social norms outcomes meta-analysis. Nearly 90% of included studies were conducted in high-income settings, with the remainder conducted in middle-income settings. No studies were conducted in Latin America or on the African continent. Social norms change interventions had a statistically significant effect on reducing tobacco use and pro-tobacco social normative perceptions [g = 0.233, 95% confidence interval (CI) = 0.166, 0.301, P < 0.001 and g = 0.292, 95% CI = 0.090, 0.494, P = 0.007, respectively]. Interventions were commonly conducted among schoolchildren in classrooms through multicomponent education sessions, often coupled with regular 'booster' sessions over time. Among adult populations, motivational interviewing and other counselling approaches were used in some cases, and few interventions leveraged digital technologies to change social norms. Social norms change interventions appear to be effective for reducing tobacco use and pro-tobacco social normative perceptions. In particular, leveraging role models appears to be the most effective social norms change mechanisms for tobacco control.

The link between digital gaming and creativity: a meta analysis

ABSTRACT This study explored the relationship between digital gaming and creativity by conducting a meta-analysis of effect sizes using data from 78 effect sizes in 19 studies encompassing 3,172 samples. We employed a two-level multivariate model with a robust variance estimator and conducted separate analyses for between-subject and within-subject designs. For the between-subject designs, the mean effect size was r = .109, 95% [−0.036, 0.249], indicating non-significance. We conducted moderator analyses to explore potential sources of variation in the observed effects, including creativity test modality, creativity test type, divergent thinking index, analytical approach, age, gender, and year of publication. None of the examined moderators were found to be statistically significant. In contrast, we observed a significant mean effect size for within-subject designs based on 29 effects from seven studies (r = .772, 95% CI [0.595, 0.878]). Variation in mean effects by study design warrants caution when making inferences about the impact of digital gaming on creativity. The implications and limitations of the findings are discussed.

Impact of Refeeding Syndrome on the Short-Term Clinical Outcomes of Very-Premature Infants.

Refeeding syndrome (RFS) is a potentially life-threatening condition that can occur in preterm infants if nutritional support is initiated or increased after a period of starvation or malnutrition. The current study aimed to examine the short-term clinical outcomes of RFS in preterm infants born at ≤32 weeks of gestation. Infants with a gestational age of ≤32 weeks and a birth weight of <1500 g who were born and admitted to the level III neonatal intensive care unit and received parenteral nutrition upon admission were retrospectively evaluated. The modified log Poisson regression with generalized linear models and a robust variance estimator was applied to adjust the outcomes of infants. In total, 760 infants met this study's inclusion criteria. Of them, 289 (38%) developed RFS. RFS was significantly associated with a composite outcome of mortality and intraventricular hemorrhage. Based on the multivariate Cox regression analysis adjusted for significant potential confounders, RFS was significantly associated with increased mortality risk, with a hazard ratio for death in infants with RFS being 1.74-fold higher compared to those without RFS. Preterm infants born at ≤32 weeks of gestation who develop RFS within the first week of life are at increased risk for both intraventricular hemorrhage and mortality. This study underscores the need for standardized clinical approaches for managing RFS in the neonatal intensive care unit to improve outcomes. Future research should establish a unified RFS definition and conduct clinical trials to optimize parenteral nutrition strategies for this vulnerable population.

COMPARATIVE STUDY OF RESIDUAL VARIANCE ESTIMATORS CONSTRUCTED WITHOUT OR WITH THE PRESENCE OF ATYPICAL VALUES: CASE OF THE PARTIALLY LINEAR MODEL

Error variance is a measure of the goodness of fit for a regression model. Ahoutou et al. [1] proposed estimators of the error variance under the assumption that the dimension of the nonlinear covariate is one. In this paper, assuming that the dimension of the nonlinear covariate is greater than or equal to 2, we propose estimators of error variance in homogeneous and heterogeneous cases. We propose estimators of the error variance that are constructed without taking atypical values into account and others that are constructed by taking all the data into account. Using simulations, we compare the performance of estimators designed without taking into account atypical values with estimators designed taking into account all the data.

The Effects of Morphological Instruction on Literacy Outcomes for Children in English-Speaking Countries: A Systematic Review and Meta-Analysis

In this pre-registered meta-analysis, we investigated the effectiveness of morphology instruction on literacy outcomes for primary school children in English-speaking countries. We were interested in overall reading and spelling outcomes, but we also looked separately at results for trained and untrained words in order to determine whether there was evidence of transfer to untrained words. Further, we were interested in whether results transferred beyond the word level to reading comprehension outcomes. Our screening process revealed 28 eligible studies, which contributed 177 effect sizes to the analyses. Robust variance estimation methods were used to account for dependence between effect sizes. Overall, effect sizes on reading and spelling outcomes were small to moderate. Effect sizes were larger for trained words than untrained words. There was evidence of transfer to untrained words for spelling outcomes, but not for reading outcomes. There was also no clear evidence of effects on reading comprehension outcomes. In general, the evidence was characterised by large amounts of heterogeneity and imprecision, which was reflective of the wide variety within and between studies in terms of intervention content, outcome measures, intervention dosage and type of control group. We discuss the limitations of the current literature and make recommendations for future research and practice in the field of morphology instruction. (207 words – max 250).

Association of early dysglycemia with intraventricular hemorrhage and mortality in very low birth weight infants.

To investigate the combined effect of hyperglycemia and hypoglycemia on intraventricular hemorrhage (IVH) and mortality recognizing that previous research has predominantly focused on examining these conditions independently. This study included very preterm infants who were born at King Saud Medical City, a tertiary referral center, and admitted to a level 3 neonatal intensive care unit between January 2020 and January 2024. Modified log-Poisson regression with generalized linear models and a robust variance estimator (Huber-White) were used to adjust for potential confounding factors. A total of 554 infants met the inclusion criteria. Hyperglycemia and/or hypoglycemia developed in 75.5% (418) patients within the first postnatal week. During the study period, IVH occurred in 28.5% (N = 158), and severe IVH occurred in 13% (72) infants. In addition, 13.7% (76) of infants died during the study period. The multivariate regression revealed an association between the isolated hyperglycemia, combined exposure of hypo- and hyperglycemia, and development of IVH (relative risk [RR], 2.10; 95% confidence interval [CI], 1.36, 3.25; RR, 2.33; 95% CI, 1.34, 4.06, respectively). Severe IVH was significantly associated with isolated hyperglycemia (adjusted relative risk [aRR], 2.46; 95% CI, 1.16, 5.23). Death was associated with combined hypo- and hyperglycemia (adjusted hazardous ratio [aHR], 3.19; 95% CI, 1.23, 8.26).Conclusion: Combined hyper- and hypoglycemia in the first week of life of premature babies increases the risk and severity of IVH and neonatal mortality.