This chapter will discuss common tests for nominal data (binomial, chi square [I], chi square [II], McNemar, and the Cochran Q), ordinal data (Kolmogorov-Smirnov, Mann-Whitney U, sign, Wilcoxon, Kruskal-Wallis, and Friedman), and interval data ( t test [I], t test [II], t test [III], and Scheffe's test).
Some of the statistical tests you will be studying are used to analyze data with many categories or outcomes . However, when only two categories are present, the binomial test (sometimes called a test of proportion) is applicable . The two categories could be, for example: grades in a pass/fail course, party affiliation as broken down into either Republican or Democrat, evaluation of a product as good or bad, a decision about a process as go or no go, outcome in tossing a coin heads or tails , or outcome in rolling a six or not a six on a die.
The proportion of cases in one category is referred to as P and in the other category as Q. The value of P + Q always equals one. If you know the value of P, you find Q by subtracting P from one. The requirements for the binomial test are:
Nominal data
One- group test
Two categories only
Sample size can be less than five
Independent observations
Simple random sample
Data in frequency form
The general formula for the binomial is:
The terms are binomial coefficients and are computed by the following formula:
The meanings of the symbols are as follows : ! = factorial, N = number of trials or sample size, X = number of favorable outcomes for a series of trials, P = probability of favorable outcome in a single trial, Q = (1 - P) = probability of unfavorable outcome in a single trial.