Measuring the Impact of Nonignorable Missingness Using the R Package isni

Hui Xie, Weihua Gao, Baodong Xing, Daniel F. Heitjan, Donald Hedeker, Chengbo Yuan

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Background and Objective: The popular assumption of ignorability simplifies analyses with incomplete data, but if it is not satisfied, results may be incorrect. Therefore it is necessary to assess the sensitivity of empirical findings to this assumption. We have created a user-friendly and freely available software program to conduct such analyses. Method: One can evaluate the dependence of inferences on the assumption of ignorability by measuring their sensitivity to its violation. One tool for such an analysis is the index of local sensitivity to nonignorability (ISNI), which evaluates the rate of change of parameter estimates to the assumed degree of nonignorability in the neighborhood of an ignorable model. Computation of ISNI avoids the need to estimate a nonignorable model or to posit a specific magnitude of nonignorability. Our new R package, named isni, implements ISNI analysis for some common data structures and corresponding statistical models. Result: The isni package computes ISNI in the generalized linear model for independent data, and in the marginal multivariate Gaussian model and the linear mixed model for longitudinal/clustered data. It allows for arbitrary patterns of missingness caused by dropout and/or intermittent missingness. Examples illustrate its use and features. Conclusions: The R package isni enables a systematic and efficient sensitivity analysis that informs evaluations of reliability and validity of empirical findings from incomplete data.

Original languageEnglish (US)
Pages (from-to)207-220
Number of pages14
JournalComputer Methods and Programs in Biomedicine
Volume164
DOIs
StatePublished - Oct 1 2018

Fingerprint

Linear Models
Statistical Models
Reproducibility of Results
Software
Sensitivity analysis
Data structures

Keywords

  • Analytical reliability
  • Data quality
  • Missing data
  • Missing not at random
  • Multivariate normal
  • Selection model

ASJC Scopus subject areas

  • Software
  • Computer Science Applications
  • Health Informatics

Cite this

Measuring the Impact of Nonignorable Missingness Using the R Package isni. / Xie, Hui; Gao, Weihua; Xing, Baodong; Heitjan, Daniel F.; Hedeker, Donald; Yuan, Chengbo.

In: Computer Methods and Programs in Biomedicine, Vol. 164, 01.10.2018, p. 207-220.

Research output: Contribution to journalArticle

Xie, Hui ; Gao, Weihua ; Xing, Baodong ; Heitjan, Daniel F. ; Hedeker, Donald ; Yuan, Chengbo. / Measuring the Impact of Nonignorable Missingness Using the R Package isni. In: Computer Methods and Programs in Biomedicine. 2018 ; Vol. 164. pp. 207-220.
@article{5201c6f070654deebaedfc54d750b768,
title = "Measuring the Impact of Nonignorable Missingness Using the R Package isni",
abstract = "Background and Objective: The popular assumption of ignorability simplifies analyses with incomplete data, but if it is not satisfied, results may be incorrect. Therefore it is necessary to assess the sensitivity of empirical findings to this assumption. We have created a user-friendly and freely available software program to conduct such analyses. Method: One can evaluate the dependence of inferences on the assumption of ignorability by measuring their sensitivity to its violation. One tool for such an analysis is the index of local sensitivity to nonignorability (ISNI), which evaluates the rate of change of parameter estimates to the assumed degree of nonignorability in the neighborhood of an ignorable model. Computation of ISNI avoids the need to estimate a nonignorable model or to posit a specific magnitude of nonignorability. Our new R package, named isni, implements ISNI analysis for some common data structures and corresponding statistical models. Result: The isni package computes ISNI in the generalized linear model for independent data, and in the marginal multivariate Gaussian model and the linear mixed model for longitudinal/clustered data. It allows for arbitrary patterns of missingness caused by dropout and/or intermittent missingness. Examples illustrate its use and features. Conclusions: The R package isni enables a systematic and efficient sensitivity analysis that informs evaluations of reliability and validity of empirical findings from incomplete data.",
keywords = "Analytical reliability, Data quality, Missing data, Missing not at random, Multivariate normal, Selection model",
author = "Hui Xie and Weihua Gao and Baodong Xing and Heitjan, {Daniel F.} and Donald Hedeker and Chengbo Yuan",
year = "2018",
month = "10",
day = "1",
doi = "10.1016/j.cmpb.2018.06.014",
language = "English (US)",
volume = "164",
pages = "207--220",
journal = "Computer Methods and Programs in Biomedicine",
issn = "0169-2607",
publisher = "Elsevier Ireland Ltd",

}

TY - JOUR

T1 - Measuring the Impact of Nonignorable Missingness Using the R Package isni

AU - Xie, Hui

AU - Gao, Weihua

AU - Xing, Baodong

AU - Heitjan, Daniel F.

AU - Hedeker, Donald

AU - Yuan, Chengbo

PY - 2018/10/1

Y1 - 2018/10/1

N2 - Background and Objective: The popular assumption of ignorability simplifies analyses with incomplete data, but if it is not satisfied, results may be incorrect. Therefore it is necessary to assess the sensitivity of empirical findings to this assumption. We have created a user-friendly and freely available software program to conduct such analyses. Method: One can evaluate the dependence of inferences on the assumption of ignorability by measuring their sensitivity to its violation. One tool for such an analysis is the index of local sensitivity to nonignorability (ISNI), which evaluates the rate of change of parameter estimates to the assumed degree of nonignorability in the neighborhood of an ignorable model. Computation of ISNI avoids the need to estimate a nonignorable model or to posit a specific magnitude of nonignorability. Our new R package, named isni, implements ISNI analysis for some common data structures and corresponding statistical models. Result: The isni package computes ISNI in the generalized linear model for independent data, and in the marginal multivariate Gaussian model and the linear mixed model for longitudinal/clustered data. It allows for arbitrary patterns of missingness caused by dropout and/or intermittent missingness. Examples illustrate its use and features. Conclusions: The R package isni enables a systematic and efficient sensitivity analysis that informs evaluations of reliability and validity of empirical findings from incomplete data.

AB - Background and Objective: The popular assumption of ignorability simplifies analyses with incomplete data, but if it is not satisfied, results may be incorrect. Therefore it is necessary to assess the sensitivity of empirical findings to this assumption. We have created a user-friendly and freely available software program to conduct such analyses. Method: One can evaluate the dependence of inferences on the assumption of ignorability by measuring their sensitivity to its violation. One tool for such an analysis is the index of local sensitivity to nonignorability (ISNI), which evaluates the rate of change of parameter estimates to the assumed degree of nonignorability in the neighborhood of an ignorable model. Computation of ISNI avoids the need to estimate a nonignorable model or to posit a specific magnitude of nonignorability. Our new R package, named isni, implements ISNI analysis for some common data structures and corresponding statistical models. Result: The isni package computes ISNI in the generalized linear model for independent data, and in the marginal multivariate Gaussian model and the linear mixed model for longitudinal/clustered data. It allows for arbitrary patterns of missingness caused by dropout and/or intermittent missingness. Examples illustrate its use and features. Conclusions: The R package isni enables a systematic and efficient sensitivity analysis that informs evaluations of reliability and validity of empirical findings from incomplete data.

KW - Analytical reliability

KW - Data quality

KW - Missing data

KW - Missing not at random

KW - Multivariate normal

KW - Selection model

UR - http://www.scopus.com/inward/record.url?scp=85051661988&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85051661988&partnerID=8YFLogxK

U2 - 10.1016/j.cmpb.2018.06.014

DO - 10.1016/j.cmpb.2018.06.014

M3 - Article

C2 - 30195428

AN - SCOPUS:85051661988

VL - 164

SP - 207

EP - 220

JO - Computer Methods and Programs in Biomedicine

JF - Computer Methods and Programs in Biomedicine

SN - 0169-2607

ER -