크론바흐 알파 신뢰계수에 대한 유의성검정에서 이상치가 검정력에 미치는 영향

엄용환 한국컴퓨터정보학회 2023 韓國컴퓨터情報學會論文誌 Vol.28 No.5

In this paper, we studied the impact of outliers on the power of the significance tests for Cronbach alpha reliability coefficient. Four variables were varied: sample size, the number of items, the number of outliers and population Cronbach Alpha levels. We simulated data using multivariate normal distribution and used outliers sampled from uniform distribution. To test the significance of Cronbach Alpha Reliability, parametric approach(F statistic) and permutation method were used. Consequently, we observed that the powers of permutation test are equal to or greater than those of F test under all conditions, and also both F test and permutation test lose the power as the number of outliers increases, and that these effects of outliers on the power are enhanced for increasing population alpha levels. 본 논문은 크론바흐 알파 신뢰계수의 유의성 검정에서 이상치가 검정력에 미치는 영향을 연구한 것이다. 표본 크기, 문항들의 수, 이상치의 수, 모집단의 크론바흐 알파 레벨의 네 개의 변수들에 변화를 주었다. 데이터 시물에이션을 위해 다변량 정규분포를 사용했고 균일분포로부터 이상치를 추출하여 사용했다. 크론바흐 알파 신뢰도의 유의성 검정을 위해 모수적 검정(F 검정)과 퍼뮤테이션 검정을 사용하였다. 결과적으로 퍼뮤테이션 검정의 검정력은 F검정의 검정력 보다 크거나 같았고, 두 검정의 검정력은 모두 이상치의 수가 많아질수록 감소하였으며 이러한 이상치의영향은 모집단의 알파 레벨이 증가할수록 크게 나타났다.

다중표본 위치검정 데이터와 관련된 독립적인 퍼뮤테이션 p값들의 통합

엄용환 한국컴퓨터정보학회 2020 韓國컴퓨터情報學會論文誌 Vol.25 No.7

Fisher’s classical method for combining independent p-values from continuous distributions is widely used but it is known to be inadequate for combining p-values from discrete probability distributions. Instead, the discrete analog of Fisher’s classical method is used as an alternative for combining p-values from discrete distributions. In this paper, firstly we obtain p-values from discrete probability distributions associated with multi-sample location test data (Fisher-Pitman test and Kruskall-Wallis test data) by permutation method, and secondly combine the permutaion p-values by the discrete analog of Fisher’s classical method. And we finally compare the combined p-values from both the discrete analog of Fisher’s classical method and Fisher’s classical method 연속형 분포로부터 얻은 독립적인 p값들을 통합하는 Fisher의 고전적인 방법은 널리 사용되고있지만 이산형 확률분포로부터 얻은 p값들을 통합하기에는 적절하지 않다. 대신에 유사 Fisher의통합방법이 이산형 확률분포의 p값들을 통합하는 대안으로 사용된다. 본 논문에서는 첫째, 여러표본들의 위치검정(Fisher-Pitman 검정과 Kruskal-Wallis 검정) 데이터와 관련된 이산형 확률분포로부터 퍼뮤테이션 방법에 의해 p값들을 구하고, 둘째로 이 p값들을 유사 Fisher의 통합방법을 이용하여 통합한다. 그리고 Fisher의 고전적인 방법과 유사 Fisher의 통합방법의 결과를 비교한다

A Simulation Study for Multivariate Independence Tests

엄용환 聖潔大學校 自然科學硏究所 1996 自然科學硏究 Vol.1 No.-

Many statistics have been proposed for testing the correlation between two random variables. It is natural to consider the multivariate test for this problem involving random vectors. A new multivariate test based on interdirections is proposed for this purpose. A comparison is made among the proposed statistic, Wilks likelihood ratio criterion and a component-wise quadrant statistic via a simulation study. Monte Carlo results demonstrate that the proposed statistic performs better than other when the underlying distributions are heavy-tailed. Also an example shows the robustness of the proposed statistic.

산포의 동질성 검정에서의 제1종 오류율과 검정력의 비교

엄용환 聖潔大學校 情報産業技術硏究所 1999 情報産業技術論叢 Vol.4 No.-

Parametric and Nonparametric procedures are compared for testing the equality of scale parameter between two groups. Estimated Type I error rates and power are reported for the Ansari-Bradley, Moses, and traditional F procedures. The effect of sample size, and the mean and the variance of the parent distribution are investigated for those tests. Normal and non-normal(exponential and t) distributions are used as the parent distributions. No test is robust and most powerful for all distributions. Ansari-Bradley test gives the most accurate estimate of Type I error rate for non-normal distributions. However, Moses test is more powerful than Ansari-Bradley test for non-normal distributions.

Quantile Estimates of Seafood Consumption Data

엄용환 聖潔大學校 情報産業技術硏究所 1997 情報産業技術論叢 Vol.2 No.-

distribution free 방법과 density smoothing 기법을 이용해서 플로리다 지역 주민의 어패류 소비 자료에 대한 quantile 추정치를 계산한다. 각 어패류에 대한 평균, 중앙값, 사분율과 같은 기술통계치를 얻는다. 또한 성별, 인종별(white, noniohite)에 따라 마찬가지의 추정치를 얻어낸다. 각각의 추정치에 대한 95% 신뢰구간을 계산한다.

특정 범주에 대한 평가자간 카파 일치도의 퍼뮤테이션 p값

엄용환,Um, Yonghwan 한국데이터정보과학회 2016 한국데이터정보과학회지 Vol.27 No.4

근사검정은 종종 표본이 작은 순서척도의 범주를 갖는 분할표를 분석할 때 그 p값이 과대추정 되거나 과소추정 되기 때문에 적절하지 못한 것으로 여겨진다. 본 논문에서는 순서화된 범주를 갖는 $k{\times}k$ 분할표에서 특정 범주에 대한 가중 일치도에 대해 정확한 p값과 재표본 기법에 의해 p값을 구하는 퍼뮤테이션 방법을 제시한다. 이를 위해 두 명의 평가자가 특정의 범주에서 얼마나 일치된 평가를 하는 지를 측정하기 위해 $Kv{\dot{a}}lseth$가 제안한 특정 범주에 대한 가중 일치도 (weighted specific-category kappa)를 사용한다. 사례 데이터로서 $3{\times}3$ 분할표 형태의 실제 데이터와 가상데이터 그리고 $4{\times}4$ 분할표 형태의 가상데이터를 이용하며, 정확한 퍼뮤테이션 p값과 재표본 퍼뮤테이션 p값 그리고 근사검정의 p값을 계산하여 비교한다. Asymptotic tests are often not suitable for the analysis of sparse ordered contingency tables as asymptotic p-values may either overestimate or underestimate the true pvalues. In this pater, we describe permutation procedures in which we compute exact or resampling p-values for a weighted specific-category agreement in ordered $k{\times}k$ contingency tables. We use the weighted specific-category kappa proposed by $Kv{\dot{a}}lseth$ to measure the extent to which two independent raters agree on the specific categories. We carried out comparison studies between exact p-values, resampling p-values and asymptotic p-values using $3{\times}3$ contingency data (real and artificial data sets) and $4{\times}4$ artificial contingency data.

Asymptotic Distribution of a Nonparametric Multivariate Test Statistic for Independence

엄용환,Um, Yong-Hwan The Korean Data and Information Science Society 2001 한국데이터정보과학회지 Vol.12 No.1

A multivariate statistic based on interdirection is proposed for detecting dependence among many vectors. The asymptotic distribution of the proposed statistic is derived under the null hypothesis of independence. Also we find the asymptotic distribution under the alternatives contiguous to the null hypothesis, which is needed for later use of computing relative efficiencies.

A Measure of Agreement for Multivariate Interval Observations by Different Sets of Raters

엄용환 한국데이터정보과학회 2004 한국데이터정보과학회지 Vol.15 No.4

A new agreement measure for multivariate interval data by different sets of raters is proposed. The proposed approach builds on Um's multivariate extension of Cohen's kappa. The proposed measure is compared with corresponding earlier measures based on Berry and Mielke's approach and Janson and Olsson approach, respectively. Application of the proposed measure is exemplified using hypothetical data set.

A new agreement measure for interval multivariate observations

엄용환 한국데이터정보과학회 2004 한국데이터정보과학회지 Vol.15 No.1

This article presents a new measure of chance-corrected interobserver agreement among multivariate ratings of many observers. Modifying an approach by Berry and Mielke, a new agreement measure is proposed. The important modificaton is to use the volume of simplex composed of data points as the disagreement masure. The proposed measure accounts agreement for multivariate interval observations among many observers. Hypothetical and real-life data sets are analyzed for illustrative purpose.

A Joint Agreement Measure Between Multiple Raters and One Standard

엄용환 한국데이터정보과학회 2005 한국데이터정보과학회지 Vol.16 No.3

This article addresses the problem of measuring a joint agreement between multiple raters and a standard set of responses. A new agreement measure based on Um's approach is proposed. The proposed agreement measure is used for multivariate interval responses. Comparison is made between the proposed measure and other corresponding agreement measures using hypothetical data set.