POLSCI 3N06 Lecture Notes - Lecture 9: Null Hypothesis, Type I And Type Ii Errors, Standard Deviation
5 Apr 2014
School
Department
Course
Professor
Document Summary
After all, many independent variables have more than 2 attributes social class (high, middle, low), provinces (10), ethnicities (many) Why not multiple paired t-tests: problem of accumulating type i errors. What we require is a single test of difference that can incorporate three or more different groups. A measure of dispersion the standard deviation squared. With anova, you are dealing with two types of variance. There is the variance within each sample (i. e. variability within each group) And there is the variance between the samples (i. e. variability between the groups) Anova tests whether the variation between the groups is significantly greater than the variation within the groups. If this is the case, then (depending upon the size of the difference) it is probable that these samples do not come from the same population. The groups are significantly different with respect to the dependent variable. These variances are expressed as sum of squares".
