Introduction to Categorical Data Analysis

46 CONTINGENCY TABLES
n11 equals
P(n11) =
� ��
n1+ n2+
n11
� n
n+1 − n11
n+1
�
� (2.11)
The binomial coefficients equal � � a
b = a!/b!(a − b)!.
To test H0: independence, the P -value is the sum of hypergeometric probabilities
for outcomes at least as favorable to Ha as the observed outcome. We illustrate for
Ha: θ>1. Given the marginal totals, tables having larger n11 values also have larger
sample odds ratios ˆθ = (n11n22)/(n12n21); hence, they provide stronger evidence in
favor of this alternative. The P -value equals the right-tail hypergeometric probability
that n11 is at least as large as the observed value. This test, proposed by the eminent
British statistician R. A. Fisher in 1934, is called Fisher’s exact test.
2.6.2 Example: Fisher’s Tea Taster
To illustrate this test in his 1935 book, The Design of Experiments, Fisher described
the following experiment: When drinking tea, a colleague of Fisher’s at Rothamsted
Experiment Station near London claimed she could distinguish whether milk or tea
was added to the cup first. To test her claim, Fisher designed an experiment in which
she tasted eight cups of tea. Four cups had milk added first, and the other four had tea
added first. She was told there were four cups of each type and she should try to select
the four that had milk added first. The cups were presented to her in random order.
Table 2.8 shows a potential result of the experiment. The null hypothesis H0: θ = 1
for Fisher’s exact test states that her guess was independent of the actual order of pouring.
The alternative hypothesis that reflects her claim, predicting a positive association
between true order of pouring and her guess, is Ha: θ>1. For this experimental
design, the column margins are identical to the row margins (4, 4), because she knew
that four cups had milk added first. Both marginal distributions are naturally fixed.
Table 2.8. Fisher’s Tea Tasting Experiment
Guess Poured First
Poured First Milk Tea Total
Milk 3 1 4
Tea 1 3 4
Total 4 4
The null distribution of n11 is the hypergeometric distribution defined for all 2 × 2
tables having row and column margins (4, 4). The potential values for n11 are (0, 1,