Levin and Fox Chapter 12 Statistics for Nonparametric Measures
Parametric measures 1.All of the statistical tests we have examined thus far require: a)Interval level data b)Normality in the population (or at least large samples so that the sampling distribution of means is normal)
Nonparametric tests What about social researchers who cannot employ a parametric test, that is, who either cannot…. a.Assume normality? b.Does not work with large numbers of cases? c.Not using measures that are interval?
Nonparametric tests Nonparametric tests make less stringent demands, but are less powerful tests than their parametric counterparts. An researchers is more likely to reject the null hypothesis when using a parametric test than when using a nonparametric test.
Examples of nonparametric tests One-Way Chi-Square (nominal data) Two-Way Chi-Square (nominal data – 2 variables) The Median Test (ordinal data) Spearman’s Rank-Order Correlation (ordinal) Goodman’s and Kruskal’s Gamma (ordinal) Phi coefficient (nominal – 2 variables) Contingency coefficient (nominal – more than 2)
Extra Credit Homework Two-Way Chi-Square (chapter 9) Convert Chi-Square into a Phi Coefficient (ch.12) Check your Chi-Square statistic using the chi- quare1.sav file on the webpage to check your work. Replaces a low scoring homework