Comparing squared multiple correlation coefficients: Examination of a confidence interval and a test significance.

I. Olkin and J. D. Finn (1995) presented 2 methods for comparing squared multiple correlation coefficients for 2 independent samples. In 1 method, the researcher constructs a confidence interval for the difference between 2 population squared coefficients; in the 2nd method, a Fisher-type transformation of the sample squared correlation coefficient is used to obtain a test statistic. Both methods are based on asymptotic theory and use approximations to the sampling variance. The approximations are incorrect when the population multiple correlation coefficient is zero. The 2 procedures were examined for equal and unequal population multiple correlation coefficients in combination with equal and unequal sample sizes. As expected, the procedures were inaccurate when the population multiple correlation coefficients were zero or very small and, in some conditions, were inaccurate when sample sizes and coefficients were unequal. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Performance of four computer-based diagnostic systems

BACKGROUND: Computer-based diagnostic systems are available commercially, but there has been limited evaluation of their performance. We assessed the diagnostic capabilities of four internal medicine diagnostic systems: Dxplain, Iliad, Meditel, and QMR. METHODS: Ten expert clinicians created a set of 105 diagnostically challenging clinical case summaries involving actual patients. Clinical data were entered into each program with the vocabulary provided by the program's developer. Each of the systems produced a ranked list of possible diagnoses for each patient, as did the group of experts. We calculated scores on several performance measures for each computer program. RESULTS: No single computer program scored better than the others on all performance measures. Among all cases and all programs, the proportion of correct diagnoses ranged from 0.52 to 0.71, and the mean proportion of relevant diagnoses ranged from 0.19 to 0.37. On average, less than half the diagnoses on the experts' original list of reasonable diagnoses were suggested by any of the programs. However, each program suggested an average of approximately two additional diagnoses per case that the experts found relevant but had not originally considered. CONCLUSIONS: The results provide a profile of the strengths and limitations of these computer programs. The programs should be used by physicians who can identify and use the relevant information and ignore the irrelevant information that can be produced.     

Detecting repeated measures effects with univariate and multivariate statistics.

Power differences between multivariate and -adjusted univariate tests are presented for various configurations and sizes of population means, degrees of nonsphericity, and sample sizes for small and large repeated measures designs with 1 within-subjects factor. The results are applicable to various designs (e.g., longitudinal). Power differences were calculated by adapting procedures presented in K. E. Muller and C. N. Barton (1989, 1991). The results demonstrate that, for parametric conditions likely to be encountered by psychological researchers, the differences between the 2 approaches can be considerable. The authors recommend that sample size be chosen according to the procedures enumerated by Muller and Barton but provide simple guidelines for use when the information required by the Muller-Barton approach is not available. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

A generally robust approach for testing hypotheses and setting confidence intervals for effect sizes.

Standard least squares analysis of variance methods suffer from poor power under arbitrarily small departures from normality and fail to control the probability of a Type I error when standard assumptions are violated. This article describes a framework for robust estimation and testing that uses trimmed means with an approximate degrees of freedom heteroscedastic statistic for independent and correlated groups designs in order to achieve robustness to the biasing effects of nonnormality and variance heterogeneity. The authors describe a nonparametric bootstrap methodology that can provide improved Type I error control. In addition, the authors indicate how researchers can set robust confidence intervals around a robust effect size parameter estimate. In an online supplement, the authors use several examples to illustrate the application of an SAS program to implement these statistical methods. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

The assessment of unidimensionality of normal and lognormal data: a look at two nonparametric procedures

In this Monte Carlo study, the Type I error rate and the power of the Stout T procedure (DIMTEST) and the Holland-Rosenbaum procedure (HR) were examined for normal and lognormal data sets. Both procedures are based on a nonparametric item response model, where the key assumption is the item response function is monotonically nondecreasing. The two procedures performed adequately under certain conditions for both normal and lognormal data sets. Of the two, however, the Stout T procedure showed adequate power under more conditions than the Holland-Rosenbaum procedure.     

Alternatives to Simonton's analyses of the interrupted and multiple-group time-series designs.

D. K. Simonton (see record 1978-00178-001) discussed the use of several linear models for the analysis of data arising in the interrupted time-series design and the multiple-group time-series design. The present authors contend that the objectives of Simonton's analyses can be realized using profile analysis. Statistical procedures for analyzing the interrupted time-series and the multiple-group time-series designs are outlined. The procedures are applicable when several Ss are observed on several pre- and posttreatment occasions and when the number of Ss is greater than the number of occasions. (11 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Generalized Eta and Omega Squared Statistics: Measures of Effect Size for Some Common Research Designs.

The editorial policies of several prominent educational and psychological journals require that researchers report some measure of effect size along with tests for statistical significance. In analysis of variance contexts, this requirement might be met by using eta squared or omega squared statistics. Current procedures for computing these measures of effect often do not consider the effect that design features of the study have on the size of these statistics. Because research-design features can have a large effect on the estimated proportion of explained variance, the use of partial eta or omega squared can be misleading. The present article provides formulas for computing generalized eta and omega squared statistics, which provide estimates of effect size that are comparable across a variety of research designs. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

An Alternative to Cohen's Standardized Mean Difference Effect Size: A Robust Parameter and Confidence Interval in the Two Independent Groups Case.

The authors argue that a robust version of Cohen's effect size constructed by replacing population means with 20% trimmed means and the population standard deviation with the square root of a 20% Winsorized variance is a better measure of population separation than is Cohen's effect size. The authors investigated coverage probability for confidence intervals for the new effect size measure. The confidence intervals were constructed by using the noncentral t distribution and the percentile bootstrap. Over the range of distributions and effect sizes investigated in the study, coverage probability was better for the percentile bootstrap confidence interval. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Robustness of the independent samples Hotelling's T2 to variance-covariance heteroscedasticity when sample sizes are unequal and in small ratios.

Type I error rates for the independent samples Hotelling's T–2 were estimated using simulated data for conditions in which the variance–covariance matrices were heteroscedastic and the ratios of unequal sample sizes were small (n?:n??=?1:1.25, 1.25:1, 1:1.1, and 1.1:1). Type I error rates were estimated for various combinations of the number of variables, sample-size-to-variables ratio, and degree and type of heteroscedasticity. Even with a sample size ratio as small as 1.1:1, the T–2 procedure can be seriously nonrobust. As expected, the procedure became less robust as the number of variables and degree of heteroscedasticity increased. (PsycINFO Database Record (c) 2010 APA, all rights reserved)