Calculation of Two-Tailed Exact Probability in the Wald-Wolfowitz One-Sample Runs Test

The objectives of this paper are to demonstrate the algorithms employed by three statistical software programs (R, Real Statistics using Excel, and SPSS) for calculating the exact two-tailed probability of the Wald-Wolfowitz one-sample runs test for randomness, to present a novel approach for computing this probability, and to compare the four procedures by generating samples of 10 and 11 data points, varying the parameters n0 (number of zeros) and n1 (number of ones), as well as the number of runs. Fifty-nine samples are created to replicate the behavior of the distribution of the number of runs with 10 and 11 data points. The exact two-tailed probabilities for the four procedures were compared using Friedman’s test. Given the significant difference in central tendency, post-hoc comparisons were conducted using Conover’s test with Benjamini-Yekutielli correction. It is concluded that the procedures of Real Statistics using Excel and R exhibit some inadequacies in the calculation of the exact two-tailed probability, whereas the new proposal and the SPSS procedure are deemed more suitable. The proposed robust algorithm has a more transparent rationale than the SPSS one, albeit being somewhat more conservative. We recommend its implementation for this test and its application to others, such as the binomial and sign test.

Note on Rank-Biserial Correlation when There Are Ties

The objective of this article is to demonstrate with examples that the two-sided tie correction does not work well. This correction was developed by Cureton so that Kendall’s tau-type and Spearman’s rho-type formulas for rank-biserial correlation yield the same result when ties are present. However, a correction based on the bracket ties achieves the desired goal, which is demonstrated algebraically and checked with three examples. On the one hand, the 10-element random sample given by Cureton, in which the two-sided tie correction performs well, is taken up. On the other hand, two other examples are given, one with a 7-element random sample and the other with a clinical random sample of 31 participants, in which the two-sided tie correction does not work, but the new correction does. It is concluded that the new corrected formulas coincide with Goodman-Kruskal’s gamma as compared to Glass’ formula that matches Somers’ dY|X or asymmetric measure of association of Y ranking with respect to X dichotomy. The use of this underreported coefficient is suggested, which is very easy to calculate from its equivalence with Kruskal-Wallis’ gamma and Somers’ dY|X.

Measures of Variability for Qualitative Variables Using the R Software

Although there are many measures of variability for qualitative variables, they are little used in social research, nor are they included in statistical software. The aim of this article is to present six measures of variation for qualitative variables of simple calculation, as well as to facilitate their use by means of the R software. The measures considered are, on the one hand, Freemans variation ratio, Morals universal variation ratio, Kvalseths standard deviation from the mode, and Wilcoxs variation ratio which are most affected by proximity to a constant random variable, where the measures of variability for qualitative variables reach their minimum value of 0. On the other hand, the Gibbs-Poston index of qualitative variation and Shannons relative entropy are included, which are more affected by the proximity to a uniform distribution, where the measures of variability for qualitative variables reach their maximum value of 1. Point and interval estimation are addressed. Bootstrap by the percentile and bias-corrected and accelerated percentile methods are used to obtain confidence intervals. Two calculation situations are presented: with a sample mode and with two or more modes. The standard deviation from the mode among the six considered measures, and the universal variation ratio among the three variation ratios, are particularly recommended for use.

Rice University Rule to Determine the Number of Bins

This study aims to establish a rationale for the Rice University rule in determining the number of bins in a histogram. It is grounded in the Scott and Freedman-Diaconis rules. Additionally, the accuracy of the empirical histogram in reproducing the shape of the distribution is assessed with respect to three factors: the rule for determining the number of bins (square root, Sturges, Doane, Scott, Freedman-Diaconis, and Rice University), sample size, and distribution type. Three measures are utilized: the average distance between empirical and theoretical histograms, the level of recognition by an expert judge, and the accuracy index, which is composed of the two aforementioned measures. Mean comparisons are conducted with aligned rank transformation analysis of variance for three fixed-effects factors: sample size (20, 35, 50, 100, 200, 500, and 1000), distribution type (10 types), and empirical rule to determine the number of bins (6 rules). From the accuracy index, Rice’s rule improves with increasing sample size and is independent of distribution type. It outperforms the Friedman-Diaconis rule but falls short of Scott’s rule, except with the arcsine distribution. Its profile of means resembles the square root rule concerning distributions and Doane’s rule concerning sample sizes. These profiles differ from those of the Scott and Friedman-Diaconis rules, which resemble each other. Among the seven rules, Scott’s rule stands out in terms of accuracy, except for the arcsine distribution, and the square root rule is the least accurate.

Everything You Wanted to Know but Could Never Find from the Cochran-Mantel-Haenszel Test

The Cochran-Mantel-Haenszel (CMH) test, developed in the 1950s, is a classic in health research, especially in epidemiology and other fields in which dichotomous and polytomous variables are frequent. This nonparametric test makes it possible to measure and check the effect of an antecedent variable X on a health outcome Y, statistically controlling the effect of a third variable Z that acts as a confounding variable in the relationship between X and Y. Both X and Y are measured on a dichotomous qualitative scale and Z on a polytomous-qualitative or ordinal scale. It is assumed that the effect of X on Y is homogeneous between the k strata of Z, which is usually tested by the Breslow-Day test with the Tarone’s correction or the Woolf’s test. The main statistical programs have the CMH test together with a test to verify the assumption of a homogeneous effect across the strata, so that it is easy to apply. However, its fundamentals and details of calculations are a mystery to most researchers, and even difficult to find or understand. The aim of this article is to present these details in a clear and concise way, including the assumptions and alternatives to non-compliance. This technical knowledge is applied to a simulated realistic example of the area of epidemiology in health and, finally, an interpretive synthesis of the analyses is given. In addition, some suggestions for the test report are made.

Testing for Normality from the Parametric Seven-Number Summary

The objective of this study is to propose the Parametric Seven-Number Summary (PSNS) as a significance test for normality and to verify its accuracy and power in comparison with two well-known tests, such as Royston’s W test and D’Agostino-Belanger-D’Agostino K-squared test. An experiment with 384 conditions was simulated. The conditions were generated by crossing 24 sample sizes and 16 types of continuous distributions: one normal and 15 non-normal. The percentage of success in maintaining the null hypothesis of normality against normal samples and in rejecting the null hypothesis against non-normal samples (accuracy) was calculated. In addition, the type II error against normal samples and the statistical power against normal samples were computed. Comparisons of percentage and means were performed using Cochran’s Q-test, Friedman’s test, and repeated measures analysis of variance. With sample sizes of 150 or greater, high accuracy and mean power or type II error (≥0.70 and ≥0.80, respectively) were achieved. All three normality tests were similarly accurate;however, the PSNS-based test showed lower mean power than K-squared and W tests, especially against non-normal samples of symmetrical-platykurtic distributions, such as the uniform, semicircle, and arcsine distributions. It is concluded that the PSNS-based omnibus test is accurate and powerful for testing normality with samples of at least 150 observations.

Shape Measures for the Distribution of a Qualitative Variable

There are several shape measures for quantitative variables, some of which can also be applied to ordinal variables. In quantitative variables, symmetry, peakedness, and kurtosis are essential properties to evaluate the deviation from assumptions, particularly normality. They aid in selecting the most appropriate method for estimating parameters and testing hypotheses. Initially, these properties serve a descriptive role in qualitative variables. Once defined, they can be considered to check for non-compliance with assumptions and to propose modifications for testing procedures. The objective of this article is to present three measures of the shape of the distribution of a qualitative variable. The concepts of qualitative asymmetry and peakedness are defined. The measurement of the first concept involves calculating the average frequency difference between qualitative categories matched by frequency homogeneity or proximity. For the second concept, the peak-to-shoulder difference and the qualitative percentile kurtosis are taken into consideration. This last measurement is a less effective option than the peak-to-shoulder difference to measure peakedness. A simulated example of the application of these three measures is given and the paper closes with some conclusions and suggestions.

Shape Measures for the Distribution of a Qualitative Variable

There are several shape measures for quantitative variables, some of which can also be applied to ordinal variables. In quantitative variables, symmetry, peakedness, and kurtosis are essential properties to evaluate the deviation from assumptions, particularly normality. They aid in selecting the most appropriate method for estimating parameters and testing hypotheses. Initially, these properties serve a descriptive role in qualitative variables. Once defined, they can be considered to check for non-compliance with assumptions and to propose modifications for testing procedures. The objective of this article is to present three measures of the shape of the distribution of a qualitative variable. The concepts of qualitative asymmetry and peakedness are defined. The measurement of the first concept involves calculating the average frequency difference between qualitative categories matched by frequency homogeneity or proximity. For the second concept, the peak-to-shoulder difference and the qualitative percentile kurtosis are taken into consideration. This last measurement is a less effective option than the peak-to-shoulder difference to measure peakedness. A simulated example of the application of these three measures is given and the paper closes with some conclusions and suggestions.

Standardized Distance from the Mean to the Median as a Measure of Skewness

The normal distribution, which has a symmetric and middle-tailed profile, is one of the most important distributions in probability theory, parametric inference, and description of quantitative variables. However, there are many non-normal distributions and knowledge of a non-zero bias allows their identification and decision making regarding the use of techniques and corrections. Pearson’s skewness coefficient defined as the standardized signed distance from the arithmetic mean to the median is very simple to calculate and clear to interpret from the normal distribution model, making it an excellent measure to evaluate this assumption, complemented with the visual inspection by means of a histogram and a box-and-whisker plot. From its variant without tripling the numerator or Yule’s skewness coefficient, the objective of this methodological article is to facilitate the use of this latter measure, presenting how to obtain asymptotic and bootstrap confidence intervals for its interpretation. Not only are the formulas shown, but they are applied with an example using R program. A general rule of interpretation of ∓0.1 has been suggested, but this can only become relevant when contextualized in relation to sample size and a measure of skewness with a population or parametric value of zero. For this purpose, intervals with confidence levels of 90%, 95% and 99% were estimated with 10,000 draws at random with replacement from 57 normally distributed samples-population with different sample sizes. The article closes with suggestions for the use of this measure of skewness.