SOC222H5 Lecture Notes - Lecture 8: Test Statistic, Null Hypothesis, Categorical Variable
Document Summary
Measures of association for categorical data (lambda and gamma) chi-square. Lambda and gamma structured in same way, ratio of explained and unexplained ratio of proportion. Chi-square is a hypothesis test, test of dependence and independence. The value is not represented same as lambda (usually reserved for relationship characterized for nominal level data) and gamma (mostly used for relationships where independent and dependant variable are ordinal) Chi-square will tell whether there is a relationship or not but not strong or weak. Chi square does not have a negative value. Lambda and gamma interpreted in the same way. Pre: gives you an amount of error. Cannot go higher than 1, has to go between 1 and -1. Errors are the opposite of what is explained, they are unexplained data. It is not a good thing as we are trying to make an explanation. 84 is least occuring value, not the mode, only looking at dependant, always error is a small number.