• Journal of Korean Society for Quality Management
• /
• v.34 no.1
• /
• pp.48-53
• /
• 2006

Applying Bayesian optimal design principles is not easy when a prior distribution is not certain. We present a optimal design criterion which possibly yield a reasonably good design and also robust with respect to misspecification of the prior distributions. The criterion is applied to the problem of estimating the turning point of a quadratic regression. Exact mathematical results are presented under certain conditions on prior distributions. Computational results are given for some cases not satisfying our conditions.

• The Korean Journal of Financial Management
• /
• v.12 no.2
• /
• pp.207-219
• /
• 1995

This paper develops an equilibrium asset pricing model with taxation in the economy. The expected excess rate of return on a risky asset is shown to be an increasing function of the covariance of asset return with aggregate consumption rate changes and the covariance of asset return with the tax rates as well. Thus, the expected execss rate of return can be decomposed as the consumption risk premium and the tax premium. The capital asset pricing model derived in the absence of taxes is shown to understate the expected excess rate of return and to have a misspecification error in the economy with taxation.

• Journal of the Korean Data and Information Science Society
• /
• v.10 no.1
• /
• pp.193-198
• /
• 1999

The estimation of parameters in regression models with multiplicative errors is usually based on the gamma or log-normal likelihoods. Under reciprocal misspecification, we compare the small sample efficiencies of two sets of estimators via a Monte Carlo study. We further consider the case where the errors are a random sample from a Weibull distribution. We compute the asymptotic relative efficiency of quasi-likelihood estimators on the original scale to least squares estimators on the log-transformed scale and perform a Monte Carlo study to compare the small sample performances of quasi-likelihood and least squares estimators.

• Journal of Korea Society of Industrial Information Systems
• /
• v.22 no.6
• /
• pp.71-84
• /
• 2017

Over the past decade, we have witnessed Serious Debates in MISQ and Other Journals Between Two Camps that have Differing Views on the use of Causal Indicators to Measure Constructs. There is the Camp that advocates Causal Indicators (ADVOCATE) and the Camp that opposes Their Usage (OPPONENT). The Debates have been primarily centered on the OPPONENT's Argument that the Meaning of a Latent Variable is determined by its Outcome Variables. However, Little Effort has been made to Validate the ADVOCATE's Dispute (Against the OPPONENT's Arguments) that the Meaning of a Latent Variable is decided by its Causal Indicators if there is no Misspecification. Our Study precisely examines the Integrity of the Argument. For this, we empirically examine how the two Primary Psychometric Properties-Comprehensiveness and Interrelationship-of Causal Indicators Influence Theory Testing between Latent Variables through Three Different Tests (i.e., Comprehensive Test, Interrelationship Test, and Mixed Test). Conducted on Two Different Datasets, Our Analysis Consistently Reveals that Structural Path Coefficients are Hardly Sensitive to the Changes (i.e., Misspecification) in the Properties of Causal Indicators. The Discovery offers Important Evidence that the Sound Theoretical Logic of a Causal Model is not in Sync with the Empirical Mechanism of Parameter Estimation. This Underscores that a Latent Variable Formed by Causal Indicators is empirically an elusive notion that is Difficult to Operationalize. As Our Results have Significant Implications on the Integrity of Numerous IS studies which have conducted Theory or Hypothesis Testing Using Causal Indicators, we strongly advocate Open Discussions among Methodologists regarding Our Findings and Their Implications for Both Published IS Research and Future Practices.

• Journal of the Korean Data and Information Science Society
• /
• v.24 no.5
• /
• pp.1089-1099
• /
• 2013

The Bradley-Terry model is widely used for analysis of pairwise preference data. We explain that the popularity of Bradley-Terry model is gained due to not only easy computation but also some nice asymptotic properties when the model is misspecified. For information retrieval required to analyze big ranking data, we propose to use a pseudo likelihood based on the Bradley-Terry model even when the true model is different from the Bradley-Terry model. We justify using the Bradley-Terry model by proving that the estimated ranking based on the proposed pseudo likelihood is consistent when the true model belongs to the class of Thurstone models, which is much bigger than the Bradley-Terry model.

• The Korean Journal of Financial Studies
• /
• v.14 no.1
• /
• pp.41-56
• /
• 2008

주식 수익률이 정상적 과정이 아니라 비정상적 과정에 의해서 생성되고 있다는 사실이 여러 실증 분석에서 제시되고 있다. 시계열의 평균이 시간의 흐름에 따라 변하면 이 시계열은 비정상적 과정에 의하여 생성된다. 시간의 흐름에 따라 평균이 변하는 비정상 시계열은 단위근과 공적분에 의하여 시계열의 운동을 모형화하고 있다. 한편 시계열의 비정상성은 분산이 시간의 흐름에 따라 변할 때에도 발생한다. 시간의 흐름에 따라 무조건부 분산은 변하지 않고 있지만 이용 가능한 정보 집합을 조건으로 하는 조건부 분산이 변하는 경우도 있다. 이 같은 성질을 가진 주가 시계열은 자기회귀 조건부 이분산(ARCH) 계통의 과정으로 모형화하고 있다. 그러나 무조건부 분산이 시간의 흐름에 따라 변하면 ARCH 계통은 중대한 모형정립과오(misspecification)에 직면하게 된다. 따라서 본 논문은 무조건부 분산이 시간의 흐름에 따라 변할 때 자기 회귀 과정의 모수를 추정하는 방법을 검토하고, 이 방법을 한국 종합주가 지수에 적용하여 자기회귀 과정의 모수를 추정하였다. 이 방법에 의하여 추정된 2계 자기회귀 과정의 모수값 중 상수항과 제1계 항의 계수는 통상 최소자승법에 의한 값과 유사하다. 그러나 제2계 항 모수의 값은 양자가 상당히 다르다. 최소자승에 의한 제2계 값이 과대 추정되고 있다.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.12 no.1
• /
• pp.111-111
• /
• 1987

With the increasing interest of reducing process variation, statistical process control has served the pivotal tool in most industrial quality programs. In this study, system analyses have been performed associated with a cost incorporated version of a process control, a quadratic loss-based X over bar control chart model. Specifically, two issues, the capital/research investments for improvement of a system and the precision of a parameter estimation, have been addressed and discussed. Through the analysis of experimental results, we show that process variability is seen to be one of the most important sources of loss and quality improvement efforts should be directed to reduce this variability. We further derive the results that, even if the optimal designs may be sensitive, the model appears to be robust with regard to misspecification of parameters. The approach and discussion taken in this study provide a meaningful guide for proper process control. We conclude this study with providing general comments.

• Korean Management Science Review
• /
• v.12 no.1
• /
• pp.111-124
• /
• 1995

With the increasing interest of reducing process variation, statistical process control has served the pivotal tool in most industrial quality programs. In this study, system analyses have been performed associated with a cost incorporated version of a process control, a quadratic loss-based X over bar control chart model. Specifically, two issues, the capital/research investments for improvement of a system and the precision of a parameter estimation, have been addressed and discussed. Through the analysis of experimental results, we show that process variability is seen to be one of the most important sources of loss and quality improvement efforts should be directed to reduce this variability. We further derive the results that, even if the optimal designs may be sensitive, the model appears to be robust with regard to misspecification of parameters. The approach and discussion taken in this study provide a meaningful guide for proper process control. We conclude this study with providing general comments.

• Journal of the Korean Data and Information Science Society
• /
• v.28 no.3
• /
• pp.685-695
• /
• 2017

The traditional mixture of experts (ME) modeled the gate network using a certain parametric function. However, if the assumed parametric function does not properly reflect the true nature, the prediction strength of ME would become weak. For example, the parametric ME often uses logistic or multinomial logistic models for the network model. However, this could be very misleading if the true nature of the data is quite different from those models. Although, in this case, we may develop more flexible parametric models by extending the model at hand, we will never be free from such misspecification problems. In order to alleviate such weakness of the parametric ME, we propose to use the semi-parametric mixture of experts (SME) in which the gate network is estimated in a non-parametrical way. Based on this, we compared the performance of the SME with those of ME and neural networks via several simulation experiments and real data examples.

• Journal of the Korean Statistical Society
• /
• v.26 no.3
• /
• pp.365-382
• /
• 1997

In this paper we study the behaviors of the generalized Durbin-Watson (DW) statistics when the nonstationary seasonal time series regression model is misspecified. It is observed that when the series is seasonally integrated the generalized DW statistic for the seasonal period order autocorrelation converges in probability to zero while teh generalized DW statistic for the first order autocorrelation has nondegenerate asymptotic distribution. When the series is regularly and seasonally integrated the generalized DW for the first order autocorrelation still converges in probability to zero.