Local response dependence and the Rasch factor model

Similar documents
Conditional maximum likelihood estimation in polytomous Rasch models using SAS

Using modern statistical methodology for validating and reporti. Outcomes

A Note on Item Restscore Association in Rasch Models

Studies on the effect of violations of local independence on scale in Rasch models: The Dichotomous Rasch model

Monte Carlo Simulations for Rasch Model Tests

Rater agreement - ordinal ratings. Karl Bang Christensen Dept. of Biostatistics, Univ. of Copenhagen NORDSTAT,

Comparison between conditional and marginal maximum likelihood for a class of item response models

What is an Ordinal Latent Trait Model?

PIRLS 2016 Achievement Scaling Methodology 1

Lesson 7: Item response theory models (part 2)

Pairwise Parameter Estimation in Rasch Models

A Marginal Maximum Likelihood Procedure for an IRT Model with Single-Peaked Response Functions

Estimating the Hausman test for multilevel Rasch model. Kingsley E. Agho School of Public health Faculty of Medicine The University of Sydney

Overview. Multidimensional Item Response Theory. Lecture #12 ICPSR Item Response Theory Workshop. Basics of MIRT Assumptions Models Applications

Basic IRT Concepts, Models, and Assumptions

Fitting Multidimensional Latent Variable Models using an Efficient Laplace Approximation

Log-linear multidimensional Rasch model for capture-recapture

Contributions to latent variable modeling in educational measurement Zwitser, R.J.

On the Construction of Adjacent Categories Latent Trait Models from Binary Variables, Motivating Processes and the Interpretation of Parameters

A Comparison of Item-Fit Statistics for the Three-Parameter Logistic Model

Links Between Binary and Multi-Category Logit Item Response Models and Quasi-Symmetric Loglinear Models

An Equivalency Test for Model Fit. Craig S. Wells. University of Massachusetts Amherst. James. A. Wollack. Ronald C. Serlin

Doctor of Philosophy

Whats beyond Concerto: An introduction to the R package catr. Session 4: Overview of polytomous IRT models

Confirmatory Factor Analysis: Model comparison, respecification, and more. Psychology 588: Covariance structure and factor models

The Covariate-Adjusted Frequency Plot for the Rasch Poisson Counts Model

Anders Skrondal. Norwegian Institute of Public Health London School of Hygiene and Tropical Medicine. Based on joint work with Sophia Rabe-Hesketh

Computationally Efficient Estimation of Multilevel High-Dimensional Latent Variable Models

ABSTRACT. Yunyun Dai, Doctor of Philosophy, Mixtures of item response theory models have been proposed as a technique to explore

Meiser et. al.: Latent Change in Discrete Data 76 as new perspectives in the application of test models to social science issues (e.g., Fischer & Mole

Item-Focussed Trees for the Detection of Differential Item Functioning in Partial Credit Models

Walkthrough for Illustrations. Illustration 1

Creating and Interpreting the TIMSS Advanced 2015 Context Questionnaire Scales

Testing Algebraic Hypotheses

Summer School in Applied Psychometric Principles. Peterhouse College 13 th to 17 th September 2010

Introduction to Structural Equation Modeling

Item Response Theory (IRT) Analysis of Item Sets

Item Response Theory (IRT) an introduction. Norman Verhelst Eurometrics Tiel, The Netherlands

Sequential Analysis of Quality of Life Measurements Using Mixed Rasch Models

Parametric Modelling of Over-dispersed Count Data. Part III / MMath (Applied Statistics) 1

A Very Brief Summary of Statistical Inference, and Examples

New Developments for Extended Rasch Modeling in R

Mixtures of Rasch Models

An Overview of Item Response Theory. Michael C. Edwards, PhD

Dimensionality Assessment: Additional Methods

Correlations with Categorical Data

The application and empirical comparison of item. parameters of Classical Test Theory and Partial Credit. Model of Rasch in performance assessments

36-720: The Rasch Model

LSAC RESEARCH REPORT SERIES. Law School Admission Council Research Report March 2008

A Note on a Tucker-Lewis-Index for Item Response Theory Models. Taehun Lee, Li Cai University of California, Los Angeles

MARGINAL HOMOGENEITY MODEL FOR ORDERED CATEGORIES WITH OPEN ENDS IN SQUARE CONTINGENCY TABLES

Contents. 3 Evaluating Manifest Monotonicity Using Bayes Factors Introduction... 44

Ensemble Rasch Models

Model comparison and selection

Measurement Invariance (MI) in CFA and Differential Item Functioning (DIF) in IRT/IFA

Preliminary Manual of the software program Multidimensional Item Response Theory (MIRT)

On the Use of Nonparametric ICC Estimation Techniques For Checking Parametric Model Fit

Nonparametric tests for the Rasch model: explanation, development, and application of quasi-exact tests for small samples

Joint Modeling of Longitudinal Item Response Data and Survival

ANALYSIS OF ORDINAL SURVEY RESPONSES WITH DON T KNOW

Center for Advanced Studies in Measurement and Assessment. CASMA Research Report

Polytomous Item Explanatory IRT Models with Random Item Effects: An Application to Carbon Cycle Assessment Data

Comparison of parametric and nonparametric item response techniques in determining differential item functioning in polytomous scale

Comparing Multi-dimensional and Uni-dimensional Computer Adaptive Strategies in Psychological and Health Assessment. Jingyu Liu

A Goodness-of-Fit Measure for the Mokken Double Monotonicity Model that Takes into Account the Size of Deviations

Growth Mixture Model

Structural Equation Modeling and Confirmatory Factor Analysis. Types of Variables

Applied Psychological Measurement 2001; 25; 283

Item Response Theory and Computerized Adaptive Testing

Standard Error of Technical Cost Incorporating Parameter Uncertainty

Likelihood-Based Methods

1 if response to item i is in category j 0 otherwise (1)

GENERALIZED LATENT TRAIT MODELS. 1. Introduction

LOG-MULTIPLICATIVE ASSOCIATION MODELS AS LATENT VARIABLE MODELS FOR NOMINAL AND0OR ORDINAL DATA. Carolyn J. Anderson* Jeroen K.

Seminar über Statistik FS2008: Model Selection

For more information about how to cite these materials visit

Comparing IRT with Other Models

Title: Testing for Measurement Invariance with Latent Class Analysis. Abstract

Extensions and Applications of Item Explanatory Models to Polytomous Data in Item Response Theory. Jinho Kim

STAT 461/561- Assignments, Year 2015

RANDOM INTERCEPT ITEM FACTOR ANALYSIS. IE Working Paper MK8-102-I 02 / 04 / Alberto Maydeu Olivares

The Mixture Approach for Simulating New Families of Bivariate Distributions with Specified Correlations

Local Dependence Diagnostics in IRT Modeling of Binary Data

A DISSERTATION SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF THE UNIVERSITY OF MINNESOTA BY. Yu-Feng Chang

IRT Model Selection Methods for Polytomous Items

A Nonlinear Mixed Model Framework for Item Response Theory

An Introduction to Causal Mediation Analysis. Xu Qin University of Chicago Presented at the Central Iowa R User Group Meetup Aug 10, 2016

Analysis of Multinomial Response Data: a Measure for Evaluating Knowledge Structures

Diagnostic Classification Models: Psychometric Issues and Statistical Challenges

SELECTION OF ITEMS FITTING A RASCH MODEL

WU Weiterbildung. Linear Mixed Models

A class of latent marginal models for capture-recapture data with continuous covariates

NESTED LOGIT MODELS FOR MULTIPLE-CHOICE ITEM RESPONSE DATA UNIVERSITY OF TEXAS AT AUSTIN UNIVERSITY OF WISCONSIN-MADISON

Definition 3.1 A statistical hypothesis is a statement about the unknown values of the parameters of the population distribution.

FACTOR ANALYSIS AND MULTIDIMENSIONAL SCALING

DETECTION OF DIFFERENTIAL ITEM FUNCTIONING USING LAGRANGE MULTIPLIER TESTS

The SAS Macro-Program %AnaQol to Estimate the Parameters of Item Responses Theory Models

1. Fisher Information

Multiple Linear Regression for the Supervisor Data

Generalized Linear Latent and Mixed Models with Composite Links and Exploded

Transcription:

Local response dependence and the Rasch factor model Dept. of Biostatistics, Univ. of Copenhagen Rasch6 Cape Town

Uni-dimensional latent variable model X 1 TREATMENT δ T X 2 AGE δ A Θ X 3 X 4 Latent variable Θ, X 1,..., X 4 items. Monotone relationships. [Holland and Rosenbaum, 1986]

Uni-dimensional latent variable model - Rasch model X 1 TREATMENT δ T X 2 AGE δ A Θ X 3 X 4 Latent variable Θ, X 1,..., X 4 items. Monotone relationships. Rasch: invariance, sufficiency

Rasch model for item response X i Item parameters β i = (β i0, β i1, β i2,...) P(X i = x Θ = θ) = exp(xθ + β ix )K i (θ) 1 (1) for x = 0, 1,..., m i. β i0 = 0 for convenience and K i (θ) = m i h=0 exp(hθ + β ih)

Assumption of local independence Response vector X = (X i ) i I P( X = x Θ = θ) = i I P(X i = x Θ = θ) yielding P( X = x Θ = θ) = exp ( Rθ + i I β ixi ) K(θ) 1 (2) where K(θ) = i I K i(θ) R = i I X i sufficient.

Marginal probability P( X = x) = P( X = x Θ = θ)ϕ(θ)dθ ( ) = exp β ixi exp(rθ)k(θ) 1 ϕ σ (θ)dθ i I

Conditional probability P( X = x R = r) = exp ( i I β ix i ) γ R ( β) (3) β = ( β i ) i I, (3) independent of θ

Likelihood functions, inference MML l MML ( β, σ) = v i I β ixi + log exp(rθ)k(θ) 1 ϕ σ (θ)dθ (4) CML (complete cases) l CML ( β) = v Pairwise conditional l PW ( β) = v β ixvi log γ Rv ( β) (5) i I β ixvi log γ R (i,i ) i i v ( β) (6) Implementation: ConQuest [Wu et al., 2007], DIGRAM [Kreiner, 2003], RUMM [Andrich et al., 2010]. Standard software [Christensen, 2006].

Uni-dimensional latent variable model X 1 TREATMENT δ T X 2 AGE δ A Θ X 3 X 4 Latent variable Θ, X 1,..., X 4 items. Monotone relationships. [Holland and Rosenbaum, 1986]

Uni-dimensional latent variable model, local dependence X 1 TREATMENT δ T X 2 AGE δ A Θ X 3 X 4 Latent variable Θ, X 1,..., X 4 items. Monotone relationships. [Andrich, 1991]

Uni-dimensional latent variable model, local dependence X 1 TREATMENT δ T X 2 AGE δ A Θ X 3 X 4 Latent variable Θ, X 1,..., X 4 items. Monotone relationships. Rasch: invariance, sufficiency. [Kreiner and Christensen, 2007]

Testing LD 1 Generalized Tjur test [Tjur, 1982] 2 Include interaction term in (9), i.e. loglinear Rasch model [Kelderman, 1984] 3 Item splitting [for dichotomous items] 4 Correlation structure in residuals

Testing LD 1 Generalized Tjur test Result for three dichotomous Rasch items [Tjur, 1982]: X 1 X 2 X 1 + X 3 generalization [Kreiner and Christensen, 2004]: X 1 X 2 R (1) and X 1 X 2 R (2) where R (1) = i 1 X i and R (2) = i 2 X i are rest scores. (Note: two p-values, need to control type I error rate). 1 Implementation: Standard software or the item screening in DIGRAM

Testing LD 2 Loglinear Rasch model Rasch model (9): ( P( X = x Θ = θ) = exp Rθ + i I β ixi ) K(θ) 1 include interaction term [Kelderman, 1984]: ( ) P( X = x Θ = θ) = exp Rθ + β ixi + δ(x 1, x 2 ) i I K(θ) 1 (7) compare (9) and (7) using likelihood ratio test. 2 Implementation: DIGRAM

Testing LD 3 Item splitting Splitting dependent item X 2 into Xi2 [Andrich and Kreiner, 2010] Illustration (dichotomous items) (0) and X i2 (1). ID X 1 X 2 1 1 1 2 0 1 3 1 0 4 1 1 5 0 0... ID X 1 X2 (0) X 2 (1) 1 1. 1 2 0 1. 3 1. 0 4 1. 1 5 0 0..... Fit model for items X 2 (0), X 2 (1), X 3,... compare item parameters ( DIF test). Need asymptotic (co)variance for formal test. 3 Implementation: RUMM, DIGRAM

Item splitting X 1 TREATMENT δ T X 2 AGE δ A Θ X 3 X 4

Item splitting X 1 TREATMENT δ T X 2 AGE δ A Θ X 3 X 4

Testing LD 4 Residuals Local dependence for (X 1, X 2 ). [Yen, 1984, Q3] Observe (X 1v, X 2v ) v=1,...,n Estimate (θ v ) v=1,...,n Compute standardized residuals R iv = X iv E(X i ˆθ v ) V (X i ˆθ v ) and ρ OBS = corr(r 1, R 2 ) Test LD: Is observed value of ρ OBS due to random variation 4 Implementation: RUMM

Testing LD using residuals Local dependence for (X 1, X 2 ). [Yen, 1984, Q3] Observe (X 1v, X 2v ) v=1,...,n Estimate (θ v ) v=1,...,n Compute residuals R iv = X iv E(X i ˆθ v ) and ρ OBS = corr(r 1, R 2 ) Test LD: Is observed value of ρ OBS due to random variation But we don t know the null distribution of ρ...

Testing LD using residuals Local dependence for (X 1, X 2 ). [Yen, 1984, Q3] Observe (X 1v, X 2v ) v=1,...,n Estimate (θ v ) v=1,...,n Compute residuals R iv = X iv E(X i ˆθ v ) and ρ OBS = corr(r 1, R 2 ) Test LD: Is observed value of ρ OBS due to random variation But we don t know the null distribution of ρ... No formal test, rule-of-thumb ρ OBS > ρ 0, ρ 0 = 0.2, 0.3, 0.6,...

The problem with residuals we don t know the distribution of ˆθ ˆθ is a biased estimate of θ R iv can only take m i + 1 different values under the Rasch model corr(r 1, R 2 ) < 0 0 1 0. 1. [Kreiner and Christensen, 2011a]

Data example (9 items from the SF36, acute leukemia) How much of the time during the past 4 weeks Did you feel full of pep? Have you been a very nervous person? Have you felt so down in the dumps that nothing could cheer you up? Have you felt calm and peaceful? Did you have a lot of energy? Have you felt downhearted and blue? Did you feel worn out? Have you been a happy person? Did you feel tired? Response options: All of the Time, Most of the Time, A Good Bit of the Time, Some of the Time, A Little of the Time, None of the Time

Generalized Tjur test X 9 X 8 X 7 X 6 Θ X 5 X 4 X 1 X 2 X 3 Latent variable Θ, X 1,..., X 9 items. Results from item screening [Kreiner and Christensen, 2011b]

Log linear Rasch model X 9 X 8 X 7 X 6 Θ X 5 X 4 X 1 X 2 X 3 Latent variable Θ, X 1,..., X 9 items. Same as Tjur + five additional (neg. LD also found, not shown)

Correlation matrix of observed residuals 1.00 0.32 0.06 0.12 0.16 0.21 0.16 0.14 0.13 1.00 0.05 0.01 0.31 0.39 0.04 0.12 0.13 1.00 0.19 0.10 0.02 0.02 0.01 0.02 1.00 0.11 0.08 0.22 0.12 0.31 1.00 0.25 0.03 0.05 0.12 1.00 0.00 0.09 0.19 1.00 0.39 0.16 1.00 0.25 1.00

Correlation matrix of observed residuals 1.00 0.32 0.06 0.12 0.16 0.21 0.16 0.14 0.13 1.00 0.05 0.01 0.31 0.39 0.04 0.12 0.13 1.00 0.19 0.10 0.02 0.02 0.01 0.02 1.00 0.11 0.08 0.22 0.12 0.31 1.00 0.25 0.03 0.05 0.12 1.00 0.00 0.09 0.19 1.00 0.39 0.16 1.00 0.25 1.00

LD summary Generalized Tjur tests and residuals disagree x x x x x x x x x x x All models incorrect. Agreement about the most highly significant pair type I error known for Tjur tests. need to know the (multivariate) distribution of the correlation matrix (or the distribution of residuals) under the Rasch model.

Rasch measurement X 1 TREATMENT δ T X 2 AGE δ A Θ X 3 X 4 Latent variable Θ, X 1,..., X 4 items.

Rasch measurement, multidimensionality X 1 TREATMENT δ 1,A δ 1,T Θ 1 X 2 AGE δ 2,T δ 2,A Θ 2 X 3 X 4 Latent variable Θ, X 1,..., X 4 items.

Two-dimensional model Set of items split into d disjoint sets. For d = 2: θ = [ θ1 θ 2 I = I 1 I 2, (8) with items in I 1 and I 2 measuring latent variables θ 1 and θ 2 respectively. P(X = x Θ = θ) = exp ( R 1 θ 1 + R 2 θ 2 + i I η ) ix i K( θ) = exp ( d R dθ d + i I η ) ix i where R d = i I d x i and K( θ) = d K( θ) i I d K i (θ d ) ]

Marginal probability ( ) ( ) P( X = x) = exp β ixi exp R d θ d K(θ) 1 ϕ Σ ( θ)d θ i I can write d dim. marginal likelihood function. Unidim. likelihood function (4) nested within, but singular. can t compare nested models using LRT [Christensen et al., 2002, Harrell-Williams and Wolfe, 2014] d

Conditional probability P( X = x R = r) = exp ( i I β ix i ) d γ(d) r d (( β i ) i Id ) d dim. conditional likelihood function is the sum l CML ( β (d) ). Unidim. likelihood function (4) nested within? d

Rasch measurement, multidimensionality Patterns in residuals t-test observed vs. expected sub score correlationn Rasch factor model Summary of tests and diagnostics by Horton et al [Horton et al., 2013]

Patterns in residuals 1-0.61 0.12-0.21 2 0.76 0.14 0.05 3-0.05 0.14 0.80 4 0.02 0.59 0.52 5-0.56-0.27 0.24 6 0.67 0.17-0.15 7 0.27-0.64 0.12 8-0.32 0.63-0.26 9-0.11-0.69 0.11

Patterns in residuals 2 0.76 0.14 0.05 6 0.67 0.17-0.15 7 0.27-0.64 0.12 8-0.32 0.63-0.26 4 0.02 0.59 0.52 1-0.61 0.12-0.21 5-0.56-0.27 0.24 9-0.11-0.69 0.11 3-0.05 0.14 0.80

Tests currently avaliable Use patterns in residuals or Tjur tests to provide hypothetises beware of circular logic I 1 I 2 t-test, observed vs. expected sub score correlation and other tests are significant for many hypothetical I = I 1 I 2 Lack of overview, many tests, risk of type I error.. (still) need to know the (multivariate) distribution of the matrix of residuals under the Rasch model.

Rasch factor model Use P( X = x θ) = P( X = x R = r) P( R = r θ) }{{} ν to write extended likelihood l EML ( β, ν) = d l CML ( β (d) ) + v log ν v with the probabilities of the marginal sub score vector distribution as unrestricted parameters ν. Compare nested models using LRT [Christensen et al., 2002]

Factor models (1, 5, 8) + (2, 6) + (3, 4) + (7, 9) (1, 2, 5, 6, 8) + (3, 4) + (7, 9) (1, 3, 4, 5, 8) + (2, 6) + (7, 9) (1, 5, 8) + (2, 6) + (3, 4, 7, 9) (1, 2, 5, 6, 8) + (3, 4, 7, 9) (1, 3, 4, 5, 8) + (2, 6, 7, 9) (1, 2,..., 9) Compare nested models using LRT (or use BIC [Schwarz, 1978])

Factor models l =..., NPAR = 205 l =..., NPAR = 184 l =..., NPAR = 194 l =..., NPAR = 190 l =..., NPAR = 141 l =..., NPAR = 159 l =..., NPAR = (45 1) + (45 1) = 88 Compare nested models using LRT (or use BIC [Schwarz, 1978])

Andrich, D. (1991). Book review: Langeheine and Rost Latent Trait and Latent Class Models. Psychometrika, 56(1):155 168. Andrich, D. and Kreiner, S. (2010). Quantifying response dependence between two dichotomous items using the rasch model. Applied Psychological Measurement, 34(3):181 192. Andrich, D., Sheridan, B., and Luo, G. (2010). RUMM2030 [Computer software and manual]. RUMM Laboratory, Perth, Australia. Christensen, K. B. (2006). Fitting polytomous Rasch models in SAS. Journal of applied measurement, 7(4):407 17.

Christensen, K. B., Bjorner, J. B., S., K., and Petersen, J. H. (2002). Testing unidimensionality in polytomous Rasch models. Psychometrika, 67:563 574. Harrell-Williams, L. and Wolfe, E. (2014). Performance of the likelihood ratio difference (g2 diff) test for detecting unidimensionality in applications of the multidimensional rasch model. Journal of applied measurement, 15(3):267 275. Holland, P. W. and Rosenbaum, P. R. (1986). Conditional association and unidimensionality in monotone latent variable models. The Annals of Statistics, 14(4):1523 1543. Horton, M., Marais, I., and Christensen, K. B. (2013). Dimensionality, pages 137 158.

John Wiley & Sons, Inc. Kelderman, H. (1984). Loglinear Rasch model tests. Psychometrika, 49:223 245. Kreiner, S. (2003). Introduction to digram. Research Report 10, Department of Statistics, University of Copenhagen. Kreiner, S. and Christensen, K. B. (2004). Analysis of local dependence and multidimensionality in graphical loglinear rasch models. Communications in Statistics - Theory and Methods, 33:1239 1276. Kreiner, S. and Christensen, K. B. (2007). Validity and objectivity in health-related scales: Analysis by graphical loglinear rasch models.

In Multivariate and Mixture Distribution Rasch Models, Statistics for Social and Behavioral Sciences, pages 329 346. Springer New York. Kreiner, S. and Christensen, K. B. (2011a). Exact evaluation of bias in Rasch model residuals. In Baswell, editor, Advances in Mathematics Research vol.12, pages 19 40. Nova publishers. Kreiner, S. and Christensen, K. B. (2011b). Item screening in graphical loglinear rasch models. Psychometrika, 76(2):228 256. Schwarz, G. E. (1978). Estimating the dimension of a model. Annals of Statistics, 6:461 464. Tjur, T. (1982). A connection between Rasch s item analysis model and a multiplicative poisson model.

Scandinavian Journal of Statistics, 9:23 30. Wu, M. L., Adams, R. J., Wilson, M. R., and Haldane, S. A. (2007). ACER ConQuest Version 2: Generalised item response modelling software. Australian Council for Educational Research, Camberwell. Yen, W. M. (1984). Effects of local item dependence on the fit and equating performance of the three-parameter logistic model. Applied Psychological Measurement, 8(2):125 145.