• The Korean Journal of Applied Statistics
• /
• v.24 no.6
• /
• pp.1045-1054
• /
• 2011

During clinical trials a researcher is frequently able to observe a disease symptom in a subject as well as a severity score for those who experienced a symptom after a fixed length of treatment. The traditional method to evaluate a decreasing trend in proportion, when there is an intrinsic order in the treatment groups (for example control and two or more treatment groups) is a Cochran-Armitage test, while the method to evaluate a decreasing trend in continuous non-normal data is a Jonckheere-Tersptra test. The Cochran-Armitage test emphasizes the dichotomous data of symptom occurrence and the Jonckheere-Tersptra test emphasizes the continuous non-normal data of severity symptom scores. In this paper we propose new test statistics that consider the combined evidence from a symptom occurrence and disease severity score. We illustrate these methods with example data of schizophrenic inpatients that demonstrated antipsychotic-drug induced constipation. A small-scale simulation is conducted to compare the new trend tests with other trend tests.

• The Korean Journal of Applied Statistics
• /
• v.23 no.6
• /
• pp.1081-1091
• /
• 2010

Clinical trials are often carried out as multi-center studies because the patients enrolled for a trial study are very limited in one particular hospital. In these circumstances, the use of an ordinary Jonckheere (1954) and Terpstra (1952) test for testing trend among several independent treatment groups is invalid. We propose a the stratified Jonckheere-Terpstra test based on van Elteren (1960)'s stratified test of Wilcoxon (1945) statistics and an application of our method is demonstrated through example data. A simulation study compares the efficiency of stratified and unstratified Jonckheere-Terpstra trend tests.

• The Korean Journal of Applied Statistics
• /
• v.21 no.1
• /
• pp.19-33
• /
• 2008

Linear regression method, proposed by Haseman and Elston(1972), for detecting linkage to a quantitative trait of sib pairs is a linkage testing method for a single locus and a single trait. However, multivariate methods for detecting linkage are needed, when information from each of several traits that are affected by the same major gene are available on each individual. Amos et al. (1990) extended the regression method of Haseman and Elston(1972) to incorporate observations of two or more traits by estimating the principal component linear function that results in the strongest correlation between the squared pair differences in the trait measurements and identity by descent at a marker locus. But, it is impossible to control the probability of type I errors with this method at present, since the exact distribution of the statistic that they use is yet unknown. In this paper, we propose a multivariate nonparametric trend test for detecting linkage to multiple traits. We compared with a simulation study the efficiencies of multivariate nonparametric trend test with those of the method developed by Amos et al. (1990) for quantitative traits data. For multivariate nonparametric trend test, the results of the simulation study reveal that the Type I error rates are close to the predetermined significance levels, and have in general high powers.

• Convergence Security Journal
• /
• v.11 no.3
• /
• pp.19-24
• /
• 2011

Software failure time presented in the literature exhibit either constant monotonic increasing or monotonic decreasing, For data analysis of software reliability model, data scale tools of trend analysis are developed. The methods of trend analysis are arithmetic mean test and Laplace trend test. Trend analysis only offer information of outline content. In this paper, we discuss forecasting failure time case of failure time censoring. In this study, time series analys is used in the simple moving average and weighted moving averages, exponential smoothing method for predict the future failure times, Empirical analysis used interval failure time for the prediction of this model. Model selection using the mean square error was presented for effective comparison.

• The Korean Journal of Applied Statistics
• /
• v.21 no.6
• /
• pp.933-945
• /
• 2008

The genetic association test on age of onset trait aims to detect the putative gene by means of linear rank tests for a significant trend of onset distributions with genotypes. However, due to the selective sampling of recruiting subjects with ages less than a pre-specified limit, the genotype groups are subject to substantially different censored distributions and thus this is one reason for the low efficiencies in the linear rank tests. In testing the equality of two survival distributions, log-rank statistic is preferred to the Wilcoxon statistic, when censored observations are nonignorable. Therefore, for more then two groups, we propose a generalized log-rank test for trend as a genetic association test. Monte Carlo studies are conducted to investigate the performances of the test statistics examined in this paper.

• The Korean Journal of Applied Statistics
• /
• v.23 no.5
• /
• pp.845-855
• /
• 2010

Extensively used case-control designs in medical studies can also be powerful and efficient for family association studies as long as an analysis method is developed for the evaluation of association between candidate genes and disease. Traditional Cochran-Armitage trend test is devised for independent subjects data, and to apply this trend test to the biologically related siblings one has to take into account the covariance among related family members in order to maintain the correct type I error rate. We propose a more powerful trend test by introducing weights that reflect the number of affected siblings in families for the evaluation of the association of genetic markers related to the disease. An application of our method to a sample family data, in addition to a small-scale simulation, is presented to compare the weighted and unweighted trend tests.

• Communications for Statistical Applications and Methods
• /
• v.15 no.4
• /
• pp.531-542
• /
• 2008

The unit root test is the major tool for determining whether we use differencing or detrending to eliminate the trend from time series data. Dickey-Fuller test (Dickey and Fuller, 1979) has the low power of test when the sample size is small or the true coefficient of AR(1) process is almost unit root and the Bayesian unit root test has complicated testing procedure. We propose a new unit root testing procedure, which mixed Bayesian approach with the traditional testing procedure. Using simulation studies, our approach showed locally higher powers than Dickey-Fuller test when the sample size is small or the time series has almost unit root and simpler procedure than Bayesian unit root test procedure. Proposed testing procedure can be applied to the time series data that are not observed as process with unit root.

• The Korean Journal of Applied Statistics
• /
• v.23 no.1
• /
• pp.123-138
• /
• 2010

After selecting a group of women with premenstrual syndrome based on daily distress scores of 28 days, one needs to estimate threshold for the change of symptoms, which would be useful for the clinician's diagnosis in hospitals. However, a test of whether a change has occurred has to precede the estimation of the threshold. In this paper, we apply parametric and nonparametric testing methods to an example data obtained from a group of women. Nonparametric method does not assume any distributional form of distress scores and parametric testing method is based on the normal distributions of linear regression lines. Therefore, the optimal situation of both methods would be different and we will assess it with a simulation study.

• The Korean Journal of Applied Statistics
• /
• v.19 no.3
• /
• pp.565-578
• /
• 2006

Repeated measurement data between two group is often used in the field of medicine study. In this paper, we suggest a method for comparison of the trend between two groups based on repeated measurement data. First, we estimate regression coefficient of linear regression model from each subject and generate samples using the regression coefficient estimated previous. And then, we test the difference between two groups by unpaired t-test, Wilcoxon rank sum test and placement test using generated samples. Monte Carlo Simulation is adapted to examine the power and experimental significance levels of several methods in various combinations.

• Communications for Statistical Applications and Methods
• /
• v.16 no.6
• /
• pp.945-957
• /
• 2009

Repeated measurement data among several treatments with a control is often used in the field of medicine study. In this paper, we suggest a method for comparison of the linear trend of responds followed time among several treatments with a control based on repeated measurement data. First, we estimate slope from each subject and generate samples using the slope estimated previous. And then, we test the difference among treatment with a control by ANOVA F test, Jonckheere-Terpstra test, updated control group procedure using generated samples. Monte Carlo Simulation is adapted to compare the power and experimental significance levels in various configuration.