Literature DB >> 35389187

A comparison of methods to address item non-response when testing for differential item functioning in multidimensional patient-reported outcome measures.

Olawale F Ayilara1, Tolulope T Sajobi2, Ruth Barclay3, Eric Bohm4, Mohammad Jafari Jozani5, Lisa M Lix6.   

Abstract

PURPOSE: Item non-response (i.e., missing data) may mask the detection of differential item functioning (DIF) in patient-reported outcome measures or result in biased DIF estimates. Non-response can be challenging to address in ordinal data. We investigated an unsupervised machine-learning method for ordinal item-level imputation and compared it with commonly-used item non-response methods when testing for DIF.
METHODS: Computer simulation and real-world data were used to assess several item non-response methods using the item response theory likelihood ratio test for DIF. The methods included: (a) list-wise deletion (LD), (b) half-mean imputation (HMI), (c) full information maximum likelihood (FIML), and (d) non-negative matrix factorization (NNMF), which adopts a machine-learning approach to impute missing values. Control of Type I error rates were evaluated using a liberal robustness criterion for α = 0.05 (i.e., 0.025-0.075). Statistical power was assessed with and without adoption of an item non-response method; differences > 10% were considered substantial.
RESULTS: Type I error rates for detecting DIF using LD, FIML and NNMF methods were controlled within the bounds of the robustness criterion for > 95% of simulation conditions, although the NNMF occasionally resulted in inflated rates. The HMI method always resulted in inflated error rates with 50% missing data. Differences in power to detect moderate DIF effects for LD, FIML and NNMF methods were substantial with 50% missing data and otherwise insubstantial.
CONCLUSION: The NNMF method demonstrated comparable performance to commonly-used non-response methods. This computationally-efficient method represents a promising approach to address item-level non-response when testing for DIF.
© 2022. The Author(s), under exclusive licence to Springer Nature Switzerland AG.

Entities:  

Keywords:  Graded response model; Item response theory; Matrix factorization; Missing data; Unsupervised machine learning

Mesh:

Year:  2022        PMID: 35389187     DOI: 10.1007/s11136-022-03129-8

Source DB:  PubMed          Journal:  Qual Life Res        ISSN: 0962-9343            Impact factor:   3.440


  17 in total

1.  Learning the parts of objects by non-negative matrix factorization.

Authors:  D D Lee; H S Seung
Journal:  Nature       Date:  1999-10-21       Impact factor: 49.962

2.  Missing data: our view of the state of the art.

Authors:  Joseph L Schafer; John W Graham
Journal:  Psychol Methods       Date:  2002-06

3.  Differential item functioning and health assessment.

Authors:  Jeanne A Teresi; John A Fleishman
Journal:  Qual Life Res       Date:  2007-04-19       Impact factor: 4.147

4.  Missing data in a multi-item instrument were best handled by multiple imputation at the item score level.

Authors:  Iris Eekhout; Henrica C W de Vet; Jos W R Twisk; Jaap P L Brand; Michiel R de Boer; Martijn W Heymans
Journal:  J Clin Epidemiol       Date:  2013-12-02       Impact factor: 6.437

5.  Estimation of IRT graded response models: limited versus full information methods.

Authors:  Carlos G Forero; Alberto Maydeu-Olivares
Journal:  Psychol Methods       Date:  2009-09

6.  Evaluating methods for handling missing ordinal data in structural equation modeling.

Authors:  Fan Jia; Wei Wu
Journal:  Behav Res Methods       Date:  2019-10

7.  International use, application and performance of health-related quality of life instruments.

Authors:  R Berzon; R D Hays; S A Shumaker
Journal:  Qual Life Res       Date:  1993-12       Impact factor: 4.147

8.  Spectral Regularization Algorithms for Learning Large Incomplete Matrices.

Authors:  Rahul Mazumder; Trevor Hastie; Robert Tibshirani
Journal:  J Mach Learn Res       Date:  2010-03-01       Impact factor: 3.654

9.  Practical and statistical issues in missing data for longitudinal patient-reported outcomes.

Authors:  Melanie L Bell; Diane L Fairclough
Journal:  Stat Methods Med Res       Date:  2013-02-19       Impact factor: 3.021

10.  Impact of missing data on bias and precision when estimating change in patient-reported outcomes from a clinical registry.

Authors:  Olawale F Ayilara; Lisa Zhang; Tolulope T Sajobi; Richard Sawatzky; Eric Bohm; Lisa M Lix
Journal:  Health Qual Life Outcomes       Date:  2019-06-20       Impact factor: 3.186

View more

北京卡尤迪生物科技股份有限公司 © 2022-2023.