CRIM 320 Lecture 9: Bivariate Statistics for Ordinal Data
16 Apr 2018
School
Department
Course
Professor
Document Summary
Crim320 lecture 9: bivariate statistics for ordinal data. Ordinal variables have more desirable statistical properties. Bc values of variables are not ordered. To understand the relationship between ordinal variables, we look at contingency tables. Easier to identify patterns by looking at %s than raw numbers. Pay attn. to how data is clustered in certain cells. We need measures of association for this. It is possible to determine the direction of a relationship. There are many commonly used measures of association: kruskal"s gamma ( , somers" d, kendall"s tabu-b and tau-c, spearman"s rho ( s) Values range between -1. 00 and +1. 00, bc the measure is directional. Determines what extent our ability to predict the score of y is improved by taking into account the. # of case pairs that are ranked in the same order on both variables. # of case pairs that are ranked in diff order on the variables. Alternative to gamma adjusting for tied ranks on the dv.
