pretty routine eye-balling
this chapter discusses measures of association for two variables when at least one of the variables is measured at the nominal level. it will concentrate on two frequently used measures for such data: cramer's V and lambda
illustrate a measure of association for nominal data. it will also help us to demonstrate the difference between measures of association and tests of significance. in chapter 16 we conducted a chi-square test to assess whether there was a relationship between level of education and enjoyment of work, we used this example to illustrate the effect that sample size can have on the value of a test statistic such as chi-square
the value of chi-square is not large enough to warrant the rejection ofthe null hypothesis of independence:the two variables do not seem to be related. yet by multiplying the sample size by 10, the value of chi-square in table 20.1b is now very large and significant: the difference between observed and expected frequencies is large enough to allow us to reject the null hypothesis of independence. but if we focus on the column percentages,and ignore the chi-square test for a moment (table20.2), it is clear that the large value of chi-square in table 20.1b is due to the large sample size, rather than due to any strong pattern of dependence between the two variables