80 likes | 191 Vues
This guide explores key concepts in statistics, focusing on normal distribution and rank correlation. It discusses how the normal distribution serves as a model for natural phenomena measurements and introduces rank correlation methods, including Spearman's rank correlation coefficient. The document also compares Spearman's method with the product-moment correlation coefficient, providing essential formulas and resources for further understanding. Key resources are included for practical exploration of these concepts, enhancing clarity on correlation measures.
E N D
Uncertainties 5 16.6.11
Outline • Normal distribution • Rank correlation
ways to think about the normal distribution • A continuous version of the binomial distribution • when N (the number of trials) is big • http://onlinestatbook.com/stat_sim/normal_approx/index.html • A model for distribution of measurements of naturally occurring phenomena • Amazing that these coincide!!
Parameters of the Normal distribution • Mean • Variance = (standard deviation)squared • Try out at: • http://www.stattucino.com/berrie/dsl/index.html
Spearman’s rank correlation coefficient Correlation measured by ‘least squares’ line of best fit has gradient called the product-moment correlation coefficient. Spearman’s rank correlation coefficient is another measure of correlation which can be used when ranks rather than measures are given. Here is the formula: Where (d-sub-i)-squared is the difference between the two rankings squared and ‘n’ is the number of items.
MEI article on correlation • Please read (at least) the first four pages of this article: • http://www.mei.org.uk/files/pdf/Spearmanrcc.pdf • After you’ve read the pages you should be clearer about the differences between the two coefficients of rank with which you are familiar.