Ordinal Measures of Association

Kruskal, William

doi:10.2307/2281954

Cited by 292 publications

(198 citation statements)

References 54 publications

Supporting

Mentioning

193

Contrasting

Unclassified

Order By: Relevance

“…Values of Tau-B range from 21 to +1, where a value of 21 indicates perfect discordance of BAC and clinical assessment, a value of +1 indicates perfect concordance, and a value near zero indicates no association. 21 Cohen's k, the most commonly used index for measuring agreement between ordinal variables, has been found to depend heavily on the distribution of the variables analysed. 22 Kendall's Tau-B has been found to perform better than other measures such as Kendall's Tau-C, Goodman and Kruskal's a, and extensions of Spearman's r. 23 Values of Tau-B using Y90 and Y91 codes were estimated both for the overall sample as well as for those who reported drinking within six hours prior to injury.…”

Section: Discussionmentioning

confidence: 99%

Clinical assessment compared with breathalyser readings in the ER: concordance of ICD-10 Y90 and Y91 codes

Touquet¹

2005

Emergency Medicine Journal

View full text Add to dashboard Cite

Objectives: The purpose of this study was to analyse the validity of clinical assessment of alcohol intoxication (ICD-10 Y91) compared with estimated blood alcohol concentration (BAC) using a breath analyser (ICD-10 Y90) among patients in the emergency room (ER). Methods: Representative samples of ER patients reporting within six hours of injury (n = 4798) from 12 countries comprising the WHO Collaborative Study on Alcohol and Injuries were breath analysed and assessed blindly for alcohol intoxication at the time of ER admission. Data were analysed using Kendall's Tau-B to measure concordance of clinical assessment and BAC, and meta analysis to determine heterogeneity of effect size. Results: Raw agreement between the two measures was 86% (Tau-B 0.68), but was lower among those reporting drinking in the six hours prior to injury (raw agreement 39%; Tau-B 0.32). No difference was found by gender or for timing of clinical assessment in relation to breath analysis. Patients positive for tolerance or dependence were more likely to be assessed as intoxicated at low levels of BAC. Estimates were homogeneous across countries only for females and for those negative for alcohol dependence. Conclusions: Clinical assessment is moderately concordant with level of BAC, but in those patients who have actually been drinking within the last six hours the concordance was much less, possibly because, in part, of a tendency on the part of clinicians to assign some level of intoxication to anyone who appeared to have been drinking.

show abstract

Section: Discussionmentioning

confidence: 99%

Clinical assessment compared with breathalyser readings in the ER: concordance of ICD-10 Y90 and Y91 codes

Touquet¹

2005

Emergency Medicine Journal

View full text Add to dashboard Cite

show abstract

“…The Spearman coefficient ρ s (Kruskal, 1958) also relies on the concordance concept, but its population version has a more complex interpretation. From the copula point of view, it can be regarded as a scalar measure of the average distance between the underlying copula, that rules the random process, and the independence one.…”

Section: Association Measure Analysismentioning

confidence: 99%

Modelling the statistical dependence of rainfall event variables through copula functions

Balistrocchi

Bacchi

2011

Hydrol. Earth Syst. Sci.

View full text Add to dashboard Cite

Abstract. In many hydrological models, such as those derived by analytical probabilistic methods, the precipitation stochastic process is represented by means of individual storm random variables which are supposed to be independent of each other. However, several proposals were advanced to develop joint probability distributions able to account for the observed statistical dependence. The traditional technique of the multivariate statistics is nevertheless affected by several drawbacks, whose most evident issue is the unavoidable subordination of the dependence structure assessment to the marginal distribution fitting. Conversely, the copula approach can overcome this limitation, by dividing the problem in two distinct parts. Furthermore, goodnessof-fit tests were recently made available and a significant improvement in the function selection reliability has been achieved. Herein the dependence structure of the rainfall event volume, the wet weather duration and the interevent time is assessed and verified by test statistics with respect to three long time series recorded in different Italian climates. Paired analyses revealed a non negligible dependence between volume and duration, while the interevent period proved to be substantially independent of the other variables. A unique copula model seems to be suitable for representing this dependence structure, despite the sensitivity demonstrated by its parameter towards the threshold utilized in the procedure for extracting the independent events. The joint probability function was finally developed by adopting a Weibull model for the marginal distributions.

show abstract

“…When modeling joints through copula, correlation coefficient looses its meaning and a commonly used alternative for measuring dependence is the Kendall's τ (Kruskal 1958;Nelsen 2006). Kendall's τ is written in terms of the copula function as…”

Section: Copulas and Marginalsmentioning

confidence: 99%

“…Since Kendall's τ measures the dependence structure of the copula function, it can subsequently be used to elicit or tuning the copula parameter θ . Another important measure of dependence used in copula model is Spearman's ρ (Kruskal 1958;Nelsen 2006). However, for most copula function, Spearman's ρ does not have an analytical tractable form and are not entertained in this paper for simplicity.…”

Section: Copulas and Marginalsmentioning

confidence: 99%

Copula, marginal distributions and model selection: a Bayesian note

Silva

Lopes

2008

Stat Comput

View full text Add to dashboard Cite

Copula functions and marginal distributions are combined to produce multivariate distributions. We show advantages of estimating all parameters of these models using the Bayesian approach, which can be done with standard Markov chain Monte Carlo algorithms. Deviance-based model selection criteria are also discussed when applied to copula models since they are invariant under monotone increasing transformations of the marginals. We focus on the deviance information criterion. The joint estimation takes into account all dependence structure of the parameters' posterior distributions in our chosen model selection criteria. Two Monte Carlo studies are conducted to show that model identification improves when the model parameters are jointly estimated. We study the Bayesian estimation of all unknown quantities at once considering bivariate copula functions and three known marginal distributions.

show abstract

Ordinal Measures of Association

Cited by 292 publications

References 54 publications

Clinical assessment compared with breathalyser readings in the ER: concordance of ICD-10 Y90 and Y91 codes

Clinical assessment compared with breathalyser readings in the ER: concordance of ICD-10 Y90 and Y91 codes

Modelling the statistical dependence of rainfall event variables through copula functions

Copula, marginal distributions and model selection: a Bayesian note

Contact Info

Product

Resources

About