Showing papers on "Intraclass correlation published in 1987"

How Many Subjects?: Statistical Power Analysis in Research

Abstract: CHAPTER 1 Introduction CHAPTER 2 General Concepts CHAPTER 3 The Pivotal Case Intraclass Correlation CHAPTER 4 Equality of News Z- and T- tests, Balanced ANOVA CHAPTER 5 Correlation Coefficients CHAPTER 6 Linear Regression CHAPTER 7 Homogeneity of Variance Tests CHAPTER 8 Binomial Tests CHAPTER 9 Contingency Table Analysis CHAPTER 10 Conclusions

Reliability of computerized surface electromyography for determining the onset of muscle activity.

Abstract: The reliability of a computerized analysis system (CAS) for determining muscle burst onset was compared with the subjective assessments of three trained examiners. A sample of 154 randomly selected, full-wave rectified and filtered electromyographic recordings was evaluated using a test-retest paradigm. Percentages of agreement, Pearson product-moment correlations, analyses of variance (ANOVAs), and intraclass correlation coefficients (ICCs) were used to measure the reliability. The between-rater agreement, which included the computerized EMG assessments, was only 23%. Within-rater agreement and Pearson correlation coefficients were perfect for CAS. The trained examiners' within-rater assessments averaged only 51% agreement, but test-retest correlations were high ( r = .78 - r = .82). All ICCs were statistically significant, ranged from .46 to .60, and tended to be higher when the CAS onset determinations were deleted from the analysis. The ANOVAs revealed that trained examiners were more consistent among each other than when their assessments were compared with CAS assessments of EMG recordings. This finding, however, may be facility-specific in that any generalization to other examiners was limited. In contrast to trained examiners, the CAS was free of variations in judgment, ensured perfect reproducibility of trial assessments, and was highly useful for analyzing multichannel EMG recordings. Although the CAS ensures perfect reliability, validity determinations require visual inspection of trial data.

A generalized model of logistic regression for clustered data

Abstract: This paper proposes a generalized logistic regression model that can account for the correlation among responses on subunits. The subunits may arise as data on multiple observations within an individual. This method generalizes earlier work by Rosner (1984 a,b) and others. Methodological generalizations include: (1) the use of the more general Polya-Eggenberger distribution instead of the beta-binomial distribution to model the correlation structure, so that cases with negative, positive, or zero intraclass correlation can be handled; (2) a stepwise approach; (3) linear and non-linear regression; and, (4) the inclusion of the case of a truncated distribution. The model can accommodate missing data and covariates on the unit and subunit level. The derivative-free simplex algorithm is used to estimate the parameters. The model is applied to data describing the progression of obstruction in coronary disease where multiple arterial segments are studied for each patient. The correlation in response that may ex...

A nonparametric correlation coefficient and a two-sample test for random vectors or directions

Abstract: SUMMARY For multivariate or directional data a nonparametric correlation coefficient and a two-sample statistic are proposed. Both of these statistics are coordinate-free, resistant and unchanged under almost all small perturbations of the data. The correlation coefficient generalizes Kendall's tau and the angular-angular correlation coefficient of Fisher & Lee (1982). The two-sample statistic is related to the Cramer-von Mises statistic and to Watson's (1962) U2. A numerical example is given.

A comparison of parent-offspring correlation estimators in terms of large-sample mean squared error and bias

Abstract: In the analysis of familial data there has been some confusion regarding the best estimator of the parent-offspring correlation and/or regression. Rather than rely on simulation evidence, we examine explicit formulae for the bias and mean squared error of four (asymptotically unbiased) estimators: the pairwise correlation and regression coefficients, the sib-mean regression coefficient, and the ensemble estimator. In terms of bias, it is shown that the ensemble estimator is a better estimator of the interclass correlation than the pairwise. In terms of mean squared error, the pairwise correlation is, on balance, the best estimator, certainly when the intraclass correlation is less than about .3: this "watershed" value has been observed by other workers in simulation studies, and varies from about .3 for families with a low mean number of progeny to about .1 for those with a high mean number. An explicit expression for the "watershed" value is obtained in terms of the arithmetic mean, the harmonic mean, and the variance of the family size distribution.

Nonnull distribution of the likelihood ratio criterion for testing equality of covariance matrices under intraclass correlation model

Abstract: The nonnull moments of the likelihood ratio statistic for testing equality of covariance matrices under intraclass correlation structure are obtained in terms of the Lauricella's hypergeometric functions and also in terms of zonal polynomials. Then the nonnull asymptotic distribution of the statistic is derived for certain alternatives.

Inferences about Interclass and Intraclass Correlations from Familial Data

Abstract: A brief review of estimation and tests of hypotheses concerning interclass and intraclass correlations from familial data is given. A modified likelihood ratio test is proposed for testing the equality of intraclass correlations in two multivariate normal populations. The asymptotic null distribution of the proposed test statistic is obtained. A test procedure based on Fisher’s z-transformation is also discussed.

Testing Hypotheses on the Mean Vector Under an Intraclass Correlation Structure

Abstract: A statistic is proposed for testing hypotheses on the mean vector of a p-variate normal distribution when the variance covariance matrix has an intraclass correlation structure. The powers of the test are computed and compared with those obtained from the statistics of Cox and Han (1982) and Clement et al. (1981). The power comparisons support the use of the proposed test.

A Note on Heritability of a Dichotomous Trait

Abstract: In a random effects model for a dichotomous trait, we prove that the heritability of the dichotomous trait given in Dempster & Lerner (1950) is strictly smaller than the intraclass correlation. A numerical comparison is given.