Use Of Multiple Imputation Methods Research Articles

Analysis of Missing Data in Progressed Learners: The Use of Multiple Imputation Methods

The issue of missing data raises concerns in all statistical and educational research. In this study, we focus on missing data in school-based assessment data generated by progressed high school learners (those who did not meet the promotional requirements for their current grades but were allowed to move to the next grade because of policy stipulations). There are a number of approaches available for handling missing data in statistical literature. Multiple imputation is an approach statisticians often recommend for dealing with missing data. We analysed a longitudinal dataset composed of progressed high school learners with missing data points, imputing the missing values with multiple imputation by chained equations and Amelia II. The missing data points in our dataset were due to learners not submitting their tasks, which then impacted negatively on their results. We found a higher proportion of missing values in mathematics/mathematical literacy, science and accounting. Our results showed that progressed learners struggle more with these subjects where knowledge develops cumulatively, and that their gaps in prior knowledge probably hinder them in understanding new concepts. Thus, the policy on progression of learners brings challenges to the already strained educational system and requires a more specialised system of support.

Missing data in FFQs: making assumptions about item non-response.

FFQs are a popular method of capturing dietary information in epidemiological studies and may be used to derive dietary exposures such as nutrient intake or overall dietary patterns and diet quality. As FFQs can involve large numbers of questions, participants may fail to respond to all questions, leaving researchers to decide how to deal with missing data when deriving intake measures. The aim of the present commentary is to discuss the current practice for dealing with item non-response in FFQs and to propose a research agenda for reporting and handling missing data in FFQs. Single imputation techniques, such as zero imputation (assuming no consumption of the item) or mean imputation, are commonly used to deal with item non-response in FFQs. However, single imputation methods make strong assumptions about the missing data mechanism and do not reflect the uncertainty created by the missing data. This can lead to incorrect inference about associations between diet and health outcomes. Although the use of multiple imputation methods in epidemiology has increased, these have seldom been used in the field of nutritional epidemiology to address missing data in FFQs. We discuss methods for dealing with item non-response in FFQs, highlighting the assumptions made under each approach. Researchers analysing FFQs should ensure that missing data are handled appropriately and clearly report how missing data were treated in analyses. Simulation studies are required to enable systematic evaluation of the utility of various methods for handling item non-response in FFQs under different assumptions about the missing data mechanism.

Multiple Imputation to Correct for Nonresponse Bias: Application in Non-communicable Disease Risk Factors Survey.

Background:This study was carried out to use multiple imputation (MI) in order to correct for the potential nonresponse bias in measurements related to variable fasting blood glucose (FBS) in non-communicable disease risk factors survey conducted in Iran in 2007.Methods:Five multiple imputation methods as bootstrap expectation maximization, multivariate normal regression, univariate linear regression, MI by chained equation, and predictive mean matching were applied to impute variable fasting blood sugar. To make FBS consistent with normality assumption natural logarithm (Ln) and Box-Cox (BC) transformations were used prior to imputation. Measurements from which we intended to remove nonresponse bias included mean of FBS and percentage of those with high FBS.Results:For mean of FBS results didn’t considerably change after applying MI methods. Regarding the prevalence of high blood sugar all methods on original scale tended to increase the estimates except for predictive mean matching that along with all methods on BC or Ln transformed data didn’t change the results.Conclusions:FBS-related measurements didn’t change after applying different MI methods. It seems that nonresponse bias was not an important challenge regarding these measurements. However use of MI methods resulted in more efficient estimations. Further studies are encouraged on accuracy of MI methods in these settings.

Missing data sensitivity analysis for recurrent event data using controlled imputation

Statistical analyses of recurrent event data have typically been based on the missing at random assumption. One implication of this is that, if data are collected only when patients are on their randomized treatment, the resulting de jure estimator of treatment effect corresponds to the situation in which the patients adhere to this regime throughout the study. For confirmatory analysis of clinical trials, sensitivity analyses are required to investigate alternative de facto estimands that depart from this assumption. Recent publications have described the use of multiple imputation methods based on pattern mixture models for continuous outcomes, where imputation for the missing data for one treatment arm (e.g. the active arm) is based on the statistical behaviour of outcomes in another arm (e.g. the placebo arm). This has been referred to as controlled imputation or reference-based imputation. In this paper, we use the negative multinomial distribution to apply this approach to analyses of recurrent events and other similar outcomes. The methods are illustrated by a trial in severe asthma where the primary endpoint was rate of exacerbations and the primary analysis was based on the negative binomial model.

Addressing missing data in clinical studies of kidney diseases.

Missing data constitute a problem present in all studies of medical research. The most common approach to handling missing data-complete case analysis-relies on assumptions about missing data that rarely hold in practice. The implications of this approach are biased and inefficient descriptions of relationships of interest. Here, various approaches for handling missing data in clinical studies are described. In particular, this work promotes the use of multiple imputation methods that rely on assumptions about missingness that are more flexible than those assumptions relied on by the most common method in use. Furthermore, multiple imputation methods are becoming increasingly more accessible in mainstream statistical software packages, making them both a sound and practical choice. The use of multiple imputation methods is illustrated with examples pertinent to kidney research, and concrete guidance on their use is provided.

Implementing Welfare Quality® in UK assurance schemes: evaluating the challenges

AbstractThis paper presents an account of a Welfare Quality® assessment of 92 dairy farms carried out by seven experienced assessors. The aim was to evaluate the potential of the Welfare Quality® assessment protocol with respect to its uptake by UK farm assurance schemes. Data collection, and measure aggregation were performed according to the Welfare Quality® protocol for dairy cows. This study examined the data itself, by the testing of how hypothetical interventions might be reflected in changes in the aggregated scores, and also investigated human-related aspects, through inter-assessor standardisation sessions to evaluate reliability, and an assessor focus group to collect feedback. Overall, three main ‘challenges’ were identified. The first challenge related to the large amount of missing data. Unexpectedly, this was such that it was only possible to calculate an overall classification for 7% of farms. The second challenge concerned the way in which aggregated scores did not always reflect hypothetical interventions. The final challenge was inter-assessor reliability, where not all assessors were found to achieve acceptable levels of agreement on a number of outcome measures by the third training session. Suggestions for managing these challenges included, follow-up to assessor training, the use of multiple imputation methods to fill in missing data, and, where applicable, not aggregating the scores. The conclusion of the study was that the protocol provided useful information from which to make an informed selection of measures, but that the challenges, combined with the lengthy assessment time, were too great for its use as a certification tool.

First Use of Multiple Imputation with the National Tuberculosis Surveillance System

The purpose of this study was to compare methods for handling missing data in analysis of the National Tuberculosis Surveillance System of the Centers for Disease Control and Prevention. Because of the high rate of missing human immunodeficiency virus (HIV) infection status in this dataset, we used multiple imputation methods to minimize the bias that may result from less sophisticated methods. We compared analysis based on multiple imputation methods with analysis based on deleting subjects with missing covariate data from regression analysis (case exclusion), and determined whether the use of increasing numbers of imputed datasets would lead to changes in the estimated association between isoniazid resistance and death. Following multiple imputation, the odds ratio for initial isoniazid resistance and death was 2.07 (95% CI 1.30, 3.29); with case exclusion, this odds ratio decreased to 1.53 (95% CI 0.83, 2.83). The use of more than 5 imputed datasets did not substantively change the results. Our experience with the National Tuberculosis Surveillance System dataset supports the use of multiple imputation methods in epidemiologic analysis, but also demonstrates that close attention should be paid to the potential impact of missing covariates at each step of the analysis.

Donor–Recipient Sex Mismatch in Kidney Transplantation

The lack of reliable human proxies for minor (ie, non-HLA) histocompatibility loci hampers the ability to leverage these factors toward improving transplant outcomes. Despite conflicting reports of the effect of donor-recipient sex mismatch on renal allografts, the association between acute rejection of renal allografts and the development of human alloantibodies to the male H-Y antigen suggested to us that donor-recipient sex mismatch deserved re-evaluation. To evaluate whether the relationships between donor sex and allograft failure differed by recipient sex. We studied recipients of deceased-donor (n = 125,369) and living-donor (n = 63,139) transplants in the United States Renal Data System. Using Cox proportional hazards models stratified by donor type, we estimated the association between donor-recipient sex mismatch and death-censored allograft failure with adjustment for known risk factors, with and without the use of multiple imputation methods to account for potential bias and/or loss of efficiency due to missing data. The advantage afforded by male donor kidneys was more pronounced among male than among female recipients (8% vs 2% relative risk reduction; interaction P < 0.01). This difference is of the order of magnitude of several other risk factors affecting donor selection decisions. Donor-recipient sex mismatch affects renal allograft survival in a direction consistent with immune responses to sexually determined minor histocompatibility antigens. Our study provides a paradigm for clinical detection of markers for minor histocompatibility loci.

Imputation of Incident Events in Longitudinal Cohort Studies

Longitudinal cohort studies normally identify and adjudicate incident events detected during follow-up by retrieving medical records. There are several reasons why the adjudication process may not be successfully completed for a suspected event including the inability to retrieve medical records from hospitals and an insufficient time between the suspected event and data analysis. These "incomplete adjudications" are normally assumed not to be events, an approach which may be associated with loss of precision and introduction of bias. In this article, the authors evaluate the use of multiple imputation methods designed to include incomplete adjudications in analysis. Using data from the REasons for Geographic And Racial Differences in Stroke (REGARDS) Study, 2008-2009, they demonstrate that this approach may increase precision and reduce bias in estimates of the relations between risk factors and incident events.

Problems in Statistical Analysis of Attrition in Randomized Controlled Clinical Trials of Antidepressant for Geriatric Depression

Attrition from clinical trials is unavoidable in geriatric psychiatry and beyond. It results in incomplete data and consequently imposes three fundamental challenges: greater bias, reduced power, and less generalizability. In an effort to assess the extent of attrition and the relevance of statistical methods applied to analyze incomplete data in geriatric psychiatry, we reviewed 69 published antidepressant randomized clinical trials conducted since 1975. The median attrition rate estimated from these trials was 26.6%; nevertheless, we found that many trials lack data analytic strategies to address the problem of attrition. Most of the applied statistical analyses involved chi-square tests, t-tests, and analysis of variance (ANOVA), each of which assume that data are missing completely at random. Even when imputation for missing data due to attrition was attempted, only the last observation carried forward (LOCF) method was implemented. The LOCF imputation can actually increase bias of the results in the analysis of repeatedly measured outcomes. In addition, despite the longitudinal nature of repeatedly measured outcomes, the statistical methods used are for analysis of cross-sectional data. Thus, the data analytic strategies did not adequately meet the challenges arising from attrition. We encourage the use of mixed-effects models to reduce the impact of attrition on bias, power and generalizability in antidepressant RCTs for geriatric depression. For imputation, we recommend use of multiple imputation methods instead of LOCF.

Multiple imputation for interval censored data with auxiliary variables

We propose a non-parametric multiple imputation scheme, NPMLE imputation, for the analysis of interval censored survival data. Features of the method are that it converts interval-censored data problems to complete data or right censored data problems to which many standard approaches can be used, and that measures of uncertainty are easily obtained. In addition to the event time of primary interest, there are frequently other auxiliary variables that are associated with the event time. For the goal of estimating the marginal survival distribution, these auxiliary variables may provide some additional information about the event time for the interval censored observations. We extend the imputation methods to incorporate information from auxiliary variables with potentially complex structures. To conduct the imputation, we use a working failure-time proportional hazards model to define an imputing risk set for each censored observation. The imputation schemes consist of using the data in the imputing risk sets to create an exact event time for each interval censored observation. In simulation studies we show that the use of multiple imputation methods can improve the efficiency of estimators and reduce the effect of missing visits when compared to simpler approaches. We apply the approach to cytomegalovirus shedding data from an AIDS clinical trial, in which CD4 count is the auxiliary variable.

Ignoring the Non-ignorables?

Missing or incomplete data on actors’ positions can cause significant problems in political analysis. Research on missing values suggests the use of multiple imputation methods rather than case deletion, but few studies have yet considered the non-ignorable problem - positions that are hidden for strategic purposes. We examine this problem and discuss the advantages and drawbacks of (i) multiple imputation as implemented in AMELIA; (ii) a computationally easy but, in the context of spatial modelling, straightforward measure of indifference and (iii) a conditional averaging algorithm, LDM, which seeks to reasonably fix actors’ positions in the policy space pre- and post-imputation. The analysis suggests that actors biased by the status quo strategically hide their more supportive positions. Although none of the existing methods - which produce quite different results - is perfectly suited for imputing hidden positions, LDM has the highest hit rate for the conjectured more supportive position.