The Use of Simplified or Misspecified Models: Linear Case

Wu, Shao-Hua; Harris, Thomas J.; McAuley, Kimberley B.

doi:10.1002/cjce.5450850401

Cited by 54 publications

(71 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As the aim of this study was to derive a predictive model, all items (domains) were included in the model despite their significance level, which is often considered best practice [33–35]. …”

Section: Methodsmentioning

confidence: 99%

A comparison of the responsiveness of EQ-5D-5L and the QOLIE-31P and mapping of QOLIE-31P to EQ-5D-5L in epilepsy

et al. 2017

View full text Add to dashboard Cite

ObjectiveTo investigate the responsiveness of and correlation between the EQ-5D-5L and the QOLIE-31P in patients with epilepsy, and develop a mapping function to predict EQ-5D-5L values based on the QOLIE-31P for use in economic evaluations.MethodsThe dataset was derived from two clinical trials, the ZMILE study in the Netherlands and the SMILE study in the UK. In both studies, patients’ quality of life using the EQ-5D-5L and QOLIE-31P was measured at baseline and 12 months follow-up. Spearman’s correlations, effect sizes (EF) and standardized response means (SRM) were calculated for both the EQ-5D-5L and QOLIE-31P domains and sub scores. Mapping functions were derived using ordinary least square (OLS) and censored least absolute deviations models.ResultsA total of 509 patients were included in this study. Low to moderately strong significant correlations were found between both instruments. The EQ-5D-5L showed high ceiling effects and small EFs and SRMs, whereas the QOLIE-31P did not show ceiling effects and also showed small to moderate EFs and SRMs. Results of the different mapping functions indicate that the highest adjusted R 2 we were able to regress was 0.265 using an OLS model with squared terms, leading to a mean absolute error of 0.103.ConclusionsResults presented in this study emphasize the shortcomings of the EQ-5D-5L in epilepsy and the importance of the development of condition-specific preference-based instruments which can be used within the QALY framework. In addition, the usefulness of the constructed mapping function in economic evaluations is questionable.Electronic supplementary materialThe online version of this article (doi:10.1007/s10198-017-0928-0) contains supplementary material, which is available to authorized users.

show abstract

Section: Methodsmentioning

confidence: 99%

A comparison of the responsiveness of EQ-5D-5L and the QOLIE-31P and mapping of QOLIE-31P to EQ-5D-5L in epilepsy

et al. 2017

View full text Add to dashboard Cite

show abstract

“…In fact, it is sometimes the case that removing predictors with small coefficients, even if they are statistically significant (and theoretically justified), results in improved prediction accuracy (see Wu et al 2007; for a simple example see Appendix A in Shmueli 2010).…”

Section: Assessing Predictive Power (Of Any Empirical Model)mentioning

confidence: 99%

“…Predictive models are based on association rather than causation between the predictors and the In fact, under-specified models can produce better predictions (Wu et al 2007). For instance, Montgomery et al (2005) showed that it is often beneficial to exclude the main effects in a model even if the interaction term between them is present.…”

Section: Choice Of Variablesmentioning

confidence: 99%

Predictive Analytics in Information Systems Research

2010

View full text Add to dashboard Cite

This research essay highlights the need to integrate predictive analytics into information systems (IS) research, and shows several concrete ways in which this can be accomplished. Predictive analytics include empirical methods (statistical and other) that generate data predictions as well as methods for assessing predictive power. Predictive analytics not only assist in creating practically useful models, they also play an important role alongside explanatory modeling in theory building and theory testing. We describe six roles for predictive analytics: new theory generation, measurement development, comparison of competing theories, improvement of existing models, relevance assessment, and assessment of the predictability of empirical phenomena. Despite the importance of predictive analytics, we find that they are rare in the empirical IS literature. The latter relies nearly exclusively on explanatory statistical modeling, where statistical inference is used to test and evaluate the explanatory power of underlying causal models. However, explanatory power does not imply predictive power and thus predictive analytics are necessary for assessing predictive power and for building empirical models that predict well. To show the distinction between predictive analytics and explanatory statistical modeling, we present differences that arise in the modeling process of each type. These differences translate into different final models, so that a pure explanatory statistical model is best tuned for testing causal hypotheses and a pure empirical predictive model is best in -known explanatory paper on TAM to a predictive context to illustrate these differences and show how predictive analytics can add theoretical and practical value to IS research.

show abstract

“…Unfortunately, it is often too difficult, or costly, to obtain enough good data to reliably estimate all of the unknown model parameters (e.g., Bagajewicz and Cabrera, 2003;Maria, 2004Maria, , 2006Mchaweh et al, 2004;Chang et al, 2005;Romdhane and Tizaoui, 2005;Wang et al, 2007). For complex models with many parameters, the resulting parameter estimates and model predictions may exhibit high variability, especially when the information content in the data available are limited, that is, the number of data points is small, measurements are noisy, the range of input-variable settings is small, and/or experimental designs are highly correlated (Wu et al, 2007). Consequently, decisions made using these models (or their parameter estimates) may be unreliable.…”

Section: Introductionmentioning

confidence: 99%

“…The practical advantages of a parsimonious model often overshadow concerns over the correctness of the model structure. When the available data are not informative, SMs can be expected to give better predictions with lower mean squared error (MSE) than the extended model (EM); (Rao, 1971;Hocking 1976;Wu et al, 2007). In the first article in this series (Wu et al, 2011), we compared the performance of nine commonly used Model-Selection Criteria (MSC) for selecting SMs when the number of data points is small and experimental designs are correlated.…”

Section: Introductionmentioning

confidence: 99%

Selection of simplified models: II. Development of a model selection criterion based on mean squared error

McAuley

Harris

2011

Can J Chem Eng

Self Cite

View full text Add to dashboard Cite

Simplified models (SMs) with a reduced set of parameters are used in many practical situations, especially when the available data for parameter estimation are limited. A variety of candidate models are often considered during the model formulation, simplification, and parameter estimation processes. We propose a new criterion to help modellers select the best SM, so that predictions with lowest expected mean squared error can be obtained. The effectiveness of the proposed criterion for selecting simplified nonlinear univariate and multivariate models is demonstrated using Monte-Carlo simulations and is compared with the effectiveness of the Bayesian Information Criterion (BIC).Des modèles simplifiés (MS) avec ensemble réduit de paramètres sont utilisés dans de nombreuses situations pratiques, particulièrement lorsque les données disponibles pour l'estimation des paramètres sont limitées. Divers modèles candidats sont souvent examinés durant la formulation du modèle, la simplification et l'estimation des paramètres. Nous proposons un nouveau critère pour aider les modélisateursà sélectionner le meilleur MS et obtenir des prédictions porteuses de l'erreur quadratique moyenne la plus faible. L'efficacité du critère proposé pour la sélection des modèles simplifiés non linéairesà variable uniques et multiples est démontrée en utilisant les simulations Monte-Carlo et comparéeà l'efficacité du critère bayesian des informations (BIC).

show abstract

The Use of Simplified or Misspecified Models: Linear Case

Cited by 54 publications

References 57 publications

A comparison of the responsiveness of EQ-5D-5L and the QOLIE-31P and mapping of QOLIE-31P to EQ-5D-5L in epilepsy

A comparison of the responsiveness of EQ-5D-5L and the QOLIE-31P and mapping of QOLIE-31P to EQ-5D-5L in epilepsy

Predictive Analytics in Information Systems Research

Selection of simplified models: II. Development of a model selection criterion based on mean squared error

Contact Info

Product

Resources

About