scholarly journals Making an unknown unknown a known unknown: Missing data in longitudinal neuroimaging studies

2018 ◽  
Author(s):  
Tyler Matta ◽  
John Coleman Flournoy ◽  
Michelle L Byrne
Keyword(s):  
Missing Data ◽  
Incomplete Data ◽  
Statistical Power ◽  
Outcome Data ◽  
Parameter Estimates ◽  
Simulation Studies ◽  
Conceptual Clarity ◽  
Neuroimaging Data ◽  
Research Findings ◽  
Unknown Unknown

The analysis of longitudinal neuroimaging data within the massively univariate framework provides the opportunity to study empirical questions about neurodevelopment. Missing outcome data are an all-too-common feature of any longitudinal study, a feature that, if handled improperly, can reduce statistical power and lead to biased parameter estimates. The goal of this paper is to provide conceptual clarity of the issues and non-issues that arise from analyzing incomplete data in longitudinal studies with particular focus on neuroimaging data. This paper begins with a review of the hierarchy of missing data mechanisms and their relationship to likelihood-based methods, a review that is necessary not just for likelihood-based methods, but also for multiple-imputation methods. Next, the paper provides a series of simulation studies with designs common in longitudinal neuroimaging studies to help illustrate missing data concepts regardless of interpretation. Finally, two applied examples are used to demonstrate the sensitivity of inferences under different missing data assumptions and how this may change the substantive interpretation. The paper concludes with a set of guidelines for analyzing incomplete longitudinal data that can improve the validity of research findings in developmental neuroimaging research.

Missing Data - Better "Not to Have Them", but What If You Do? (Part 1)

Marketing ZFP ◽  
2019 ◽  
Vol 41 (4) ◽  
pp. 21-32
Author(s):  
Dirk Temme ◽  
Sarah Jensen
Keyword(s):  
Missing Data ◽  
Statistical Power ◽  
Missing Values ◽  
Graphical Representation ◽  
Marketing Research ◽  
Likelihood Estimation ◽  
Parameter Estimates ◽  
Full Information Maximum Likelihood ◽  
Definition Of ◽  
Traditional Approaches

Missing values are ubiquitous in empirical marketing research. If missing data are not dealt with properly, this can lead to a loss of statistical power and distorted parameter estimates. While traditional approaches for handling missing data (e.g., listwise deletion) are still widely used, researchers can nowadays choose among various advanced techniques such as multiple imputation analysis or full-information maximum likelihood estimation. Due to the available software, using these modern missing data methods does not pose a major obstacle. Still, their application requires a sound understanding of the prerequisites and limitations of these methods as well as a deeper understanding of the processes that have led to missing values in an empirical study. This article is Part 1 and first introduces Rubin’s classical definition of missing data mechanisms and an alternative, variable-based taxonomy, which provides a graphical representation. Secondly, a selection of visualization tools available in different R packages for the description and exploration of missing data structures is presented.

scholarly journals Evaluation of Four Multiple Imputation Methods for Handling Missing Binary Outcome Data in the Presence of an Interaction between a Dummy and a Continuous Variable

2021 ◽  
Vol 2021 ◽  
pp. 1-14
Author(s):  
Sara Javadi ◽  
Abbas Bahrampour ◽  
Mohammad Mehdi Saber ◽  
Behshid Garrusi ◽  
Mohammad Reza Baneshi
Keyword(s):  
Missing Data ◽  
Multiple Imputation ◽  
Simulation Study ◽  
Recursive Partitioning ◽  
Real Data ◽  
Continuous Variable ◽  
Outcome Data ◽  
Classification And Regression Tree ◽  
Parameter Estimates ◽  
Imputation Model

Multiple imputation by chained equations (MICE) is the most common method for imputing missing data. In the MICE algorithm, imputation can be performed using a variety of parametric and nonparametric methods. The default setting in the implementation of MICE is for imputation models to include variables as linear terms only with no interactions, but omission of interaction terms may lead to biased results. It is investigated, using simulated and real datasets, whether recursive partitioning creates appropriate variability between imputations and unbiased parameter estimates with appropriate confidence intervals. We compared four multiple imputation (MI) methods on a real and a simulated dataset. MI methods included using predictive mean matching with an interaction term in the imputation model in MICE (MICE-interaction), classification and regression tree (CART) for specifying the imputation model in MICE (MICE-CART), the implementation of random forest (RF) in MICE (MICE-RF), and MICE-Stratified method. We first selected secondary data and devised an experimental design that consisted of 40 scenarios (2 × 5 × 4), which differed by the rate of simulated missing data (10%, 20%, 30%, 40%, and 50%), the missing mechanism (MAR and MCAR), and imputation method (MICE-Interaction, MICE-CART, MICE-RF, and MICE-Stratified). First, we randomly drew 700 observations with replacement 300 times, and then the missing data were created. The evaluation was based on raw bias (RB) as well as five other measurements that were averaged over the repetitions. Next, in a simulation study, we generated data 1000 times with a sample size of 700. Then, we created missing data for each dataset once. For all scenarios, the same criteria were used as for real data to evaluate the performance of methods in the simulation study. It is concluded that, when there is an interaction effect between a dummy and a continuous predictor, substantial gains are possible by using recursive partitioning for imputation compared to parametric methods, and also, the MICE-Interaction method is always more efficient and convenient to preserve interaction effects than the other methods.

Benefits from Computerized Adaptive Testing as Seen in Simulation Studies

1999 ◽  
Vol 15 (2) ◽  
pp. 91-98 ◽  
Author(s):  
Lutz F. Hornke
Keyword(s):  
Measurement Error ◽  
Computerized Adaptive Testing ◽  
Test Procedure ◽  
Adaptive Testing ◽  
Parameter Estimates ◽  
Simulation Studies ◽  
Computerized Adaptive Test ◽  
Item Banks ◽  
Item Parameters ◽  
General Reliability

Summary: Item parameters for several hundreds of items were estimated based on empirical data from several thousands of subjects. The logistic one-parameter (1PL) and two-parameter (2PL) model estimates were evaluated. However, model fit showed that only a subset of items complied sufficiently, so that the remaining ones were assembled in well-fitting item banks. In several simulation studies 5000 simulated responses were generated in accordance with a computerized adaptive test procedure along with person parameters. A general reliability of .80 or a standard error of measurement of .44 was used as a stopping rule to end CAT testing. We also recorded how often each item was used by all simulees. Person-parameter estimates based on CAT correlated higher than .90 with true values simulated. For all 1PL fitting item banks most simulees used more than 20 items but less than 30 items to reach the pre-set level of measurement error. However, testing based on item banks that complied to the 2PL revealed that, on average, only 10 items were sufficient to end testing at the same measurement error level. Both clearly demonstrate the precision and economy of computerized adaptive testing. Empirical evaluations from everyday uses will show whether these trends will hold up in practice. If so, CAT will become possible and reasonable with some 150 well-calibrated 2PL items.

Examining Nonnormal Latent Variable Distributions for Non-Ignorable Missing Data

2021 ◽  
Vol 45 (3) ◽  
pp. 159-177
Author(s):  
Chen-Wei Liu
Keyword(s):  
Missing Data ◽  
Latent Variable ◽  
Bivariate Normal Distribution ◽  
Parameter Estimates ◽  
Missing Responses ◽  
Variable Distribution ◽  
Item Responses ◽  
Nonnormal Distribution ◽  
The Impact ◽  
Biased Estimates

Missing not at random (MNAR) modeling for non-ignorable missing responses usually assumes that the latent variable distribution is a bivariate normal distribution. Such an assumption is rarely verified and often employed as a standard in practice. Recent studies for “complete” item responses (i.e., no missing data) have shown that ignoring the nonnormal distribution of a unidimensional latent variable, especially skewed or bimodal, can yield biased estimates and misleading conclusion. However, dealing with the bivariate nonnormal latent variable distribution with present MNAR data has not been looked into. This article proposes to extend unidimensional empirical histogram and Davidian curve methods to simultaneously deal with nonnormal latent variable distribution and MNAR data. A simulation study is carried out to demonstrate the consequence of ignoring bivariate nonnormal distribution on parameter estimates, followed by an empirical analysis of “don’t know” item responses. The results presented in this article show that examining the assumption of bivariate nonnormal latent variable distribution should be considered as a routine for MNAR data to minimize the impact of nonnormality on parameter estimates.

scholarly journals PENDUGAAN DATA HILANG DENGAN METODE YATES DAN ALGORITMA EM PADA RANCANGAN LATTICE SEIMBANG

2015 ◽  
Vol 4 (2) ◽  
pp. 74
Author(s):  
MADE SUSILAWATI ◽  
KARTIKA SARI
Keyword(s):  
Missing Data ◽  
Em Algorithm ◽  
Basic Concept ◽  
Incomplete Data ◽  
Likelihood Function ◽  
Animal Husbandry ◽  
Lattice Design ◽  
The Em Algorithm ◽  
Missing Data Estimation ◽  
Data Estimation

Missing data often occur in agriculture and animal husbandry experiment. The missing data in experimental design makes the information that we get less complete. In this research, the missing data was estimated with Yates method and Expectation Maximization (EM) algorithm. The basic concept of the Yates method is to minimize sum square error (JKG), meanwhile the basic concept of the EM algorithm is to maximize the likelihood function. This research applied Balanced Lattice Design with 9 treatments, 4 replications and 3 group of each repetition. Missing data estimation results showed that the Yates method was better used for two of missing data in the position on a treatment, a column and random, meanwhile the EM algorithm was better used to estimate one of missing data and two of missing data in the position of a group and a replication. The comparison of the result JKG of ANOVA showed that JKG of incomplete data larger than JKG of incomplete data that has been added with estimator of data. This suggest  thatwe need to estimate the missing data.

A new imputation-based incomplete data-driven fuzzy modeling for accuracy improvement in ubiquitous computing applications

2021 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Author(s):  
Sonia Goel ◽  
Meena Tushir
Keyword(s):  
Parameter Estimation ◽  
Missing Data ◽  
Incomplete Data ◽  
Model Identification ◽  
Fuzzy Model ◽  
New Method ◽  
Content Type ◽  
Ubiquitous Environment ◽  
Sugeno Model ◽  
Takagi Sugeno

Purpose In real-world decision-making, high accuracy data analysis is essential in a ubiquitous environment. However, we encounter missing data while collecting user-related data information because of various privacy concerns on account of a user. This paper aims to deal with incomplete data for fuzzy model identification, a new method of parameter estimation of a Takagi–Sugeno model in the presence of missing features. Design/methodology/approach In this work, authors proposed a three-fold approach for fuzzy model identification in which imputation-based linear interpolation technique is used to estimate missing features of the data, and then fuzzy c-means clustering is used for determining optimal number of rules and for the determination of parameters of membership functions of the fuzzy model. Finally, the optimization of the all antecedent and consequent parameters along with the width of the antecedent (Gaussian) membership function is done by gradient descent algorithm based on the minimization of root mean square error. Findings The proposed method is tested on two well-known simulation examples as well as on a real data set, and the performance is compared with some traditional methods. The result analysis and statistical analysis show that the proposed model has achieved a considerable improvement in accuracy in the presence of varying degree of data incompleteness. Originality/value The proposed method works well for fuzzy model identification method, a new method of parameter estimation of a Takagi–Sugeno model in the presence of missing features with varying degree of missing data as compared to some well-known methods.

scholarly journals MGMM: An R Package for fitting Gaussian Mixture Models on Incomplete Data

2019 ◽  
Author(s):  
Zachary R. McCaw ◽  
Hanna Julienne ◽  
Hugues Aschard
Keyword(s):  
Missing Data ◽  
Mixture Models ◽  
Gaussian Mixture Models ◽  
Model Fitting ◽  
Simulated Data ◽  
R Package ◽  
Gaussian Mixture ◽  
Parameter Estimates ◽  
Cluster Assignment ◽  
Underlying Distribution

AbstractAlthough missing data are prevalent in applications, existing implementations of Gaussian mixture models (GMMs) require complete data. Standard practice is to perform complete case analysis or imputation prior to model fitting. Both approaches have serious drawbacks, potentially resulting in biased and unstable parameter estimates. Here we present MGMM, an R package for fitting GMMs in the presence of missing data. Using three case studies on real and simulated data sets, we demonstrate that, when the underlying distribution is near-to a GMM, MGMM is more effective at recovering the true cluster assignments than state of the art imputation followed by standard GMM. Moreover, MGMM provides an accurate assessment of cluster assignment uncertainty even when the generative distribution is not a GMM. This assessment may be used to identify unassignable observations. MGMM is available as an R package on CRAN: https://CRAN.R-project.org/package=MGMM.

scholarly journals Z-Curve

2017 ◽  
Author(s):  
Ulrich Schimmack ◽  
Jerry Brunner
Keyword(s):  
Selection Bias ◽  
Statistical Power ◽  
False Negative ◽  
Simulation Studies ◽  
Replication Studies ◽  
Negative Results ◽  
False Negative Results ◽  
Selection For ◽  
Z Curve ◽  
Substantial Heterogeneity

In recent years, the replicability of original findings published in psychology journals has been questioned. A key concern is that selection for significance inflates observed effect sizes and observed power. If selection bias is severe, replication studies are unlikely to reproduce a significant result. We introduce z-curve as a new method that can estimate the average true power for sets of studies that are selected for significance. We compare this method with p-curve, which has been used for the same purpose. Simulation studies show that both methods perform well when all studies have the same power, but p-curve overestimates power if power varies across studies. Based on these findings, we recommend z-curve to estimate power for sets of studies that are heterogeneous and selected for significance. Application of z-curve to various datasets suggests that the average replicability of published results in psychology is approximately 50%, but there is substantial heterogeneity and many psychological studies remain underpowered and are likely to produce false negative results. To increase replicability and credibility of published results it is important to reduce selection bias and to increase statistical power.

scholarly journals Revisiting Gaussian copulas to handle endogenous regressors

2021 ◽  
Author(s):  
Jan-Michael Becker ◽  
Dorian Proksch ◽  
Christian M. Ringle
Keyword(s):  
Regression Models ◽  
Statistical Power ◽  
Multilevel Models ◽  
Gaussian Copula ◽  
Simulation Studies ◽  
Validity Of Results ◽  
Endogenous Regressors ◽  
Copula Approach ◽  
Original Presentation ◽  
And Performance

AbstractMarketing researchers are increasingly taking advantage of the instrumental variable (IV)-free Gaussian copula approach. They use this method to identify and correct endogeneity when estimating regression models with non-experimental data. The Gaussian copula approach’s original presentation and performance demonstration via a series of simulation studies focused primarily on regression models without intercept. However, marketing and other disciplines’ researchers mainly use regression models with intercept. This research expands our knowledge of the Gaussian copula approach to regression models with intercept and to multilevel models. The results of our simulation studies reveal a fundamental bias and concerns about statistical power at smaller sample sizes and when the approach’s primary assumptions are not fully met. This key finding opposes the method’s potential advantages and raises concerns about its appropriate use in prior studies. As a remedy, we derive boundary conditions and guidelines that contribute to the Gaussian copula approach’s proper use. Thereby, this research contributes to ensuring the validity of results and conclusions of empirical research applying the Gaussian copula approach.

scholarly journals Comparison of Serial and Parallel Computation on Predicting Missing Data with EM Algorithm

2021 ◽  
Vol 18 (1) ◽  
pp. 22-30
Author(s):  
Erna Nurmawati ◽  
Robby Hasan Pangaribuan ◽  
Ibnu Santoso
Keyword(s):  
Parallel Computing ◽  
Missing Data ◽  
Data Processing ◽  
Em Algorithm ◽  
Parallel Computation ◽  
Incomplete Data ◽  
Parallel Programs ◽  
Missing Value ◽  
Speed Up ◽  
Compute Time

One way to deal with the presence of missing value or incomplete data is to impute the data using EM Algorithm. The need for large and fast data processing is necessary to implement parallel computing on EM algorithm serial program. In the parallel program architecture of EM Algorithm in this study, the controller is only related to the EM module whereas the EM module itself uses matrix and vector modules intensively. Parallelization is done by using OpenMP in EM modules which results in faster compute time on parallel programs than serial programs. Parallel computing with a thread of 4 (four) increases speed up, reduces compute time, and reduces efficiency when compared to parallel computing by the number of threads 2 (two).

Sign in / Sign up

Export Citation Format

Share Document