In a population of adolescents, I have to compare distribution of Family Affluence Scale (FAS) according to school level (level). I use the FREQ procedure (with chisq option) to do so, and I wonder why SAS does not warn me in the output of expected values under 5.
I know that the chi-square may be valid (because 5 is an empirical choice and is conservative, especially when degrees of freedom and sample size are high, see Agresti "Categorical Data Analysis" for example), but I don't understand why SAS does not warn me. I thought SAS always printed a warning message when expected values were under 5.
I do get the Warning message, but only when there more than 20 % cells with expected counts under 5.
See Freq procedure documentation, on the NOWARN option in the TABLES statement :
suppresses the log warning message that the asymptotic chi-square test may not be valid. By default, PROC FREQ displays this log message when more than 20 percent of the table cells have expected frequencies less than five.