Вы находитесь на странице: 1из 10

Chapter 19: The Binomial Test

The Binomial Test
• The binomial test provides a method for
testing hypotheses about population
proportions for populations consisting of
binomial data.
• Binomial data exist when the
measurement procedure classifies
individuals into exactly two distinct

The Binomial Test (cont.)
• Traditionally, the two categories are identified as
A and B, and the population proportions are
identified as p(A) = p and p(B) = q.
• The null hypothesis specifies the values of p and
q for the population.
• For example, when testing whether or not a coin
is balanced, the null hypothesis would state that
the coin is balanced or, p(Heads) = p = .50, and
p(Tails) = q = .50.

The Binomial Test (cont.)
• The sample data for the binomial test
consist of a sample of n individuals each
of whom is classified in category A or B.
• The sample statistic, X, is simply the
number of individuals classified in
category A.
• The logic underlying the binomial test is
identical to the logic for the original z-score
test or the t-statistic hypothesis tests.
The Binomial Test (cont.)
• The test statistic compares the sample
data with the hypothesized value for the
• If the data are consistent with the
hypothesis, we conclude that the
hypothesis is reasonable.
• However, if there is a large discrepancy
between the data and the hypothesis, we
reject the hypothesis.
The Binomial Test (cont.)
• When the values of pn and qn are both greater
than or equal to 10, the binomial distribution is
approximately normal with a mean of μ = pn and
a standard deviation of σ = npq.
• In this case, the binomial test can be conducted
by transforming the X value from the sample into
a z-score and then using the unit normal table to
determine critical values.
• If the z-score is only slightly into the critical
region, you should check both real limits for X to
ensure that the entire score is beyond the critical
The Sign Test
• The sign test is a special application of the
binomial test used to evaluate the results from a
repeated-measures research design comparing
two treatment conditions.
• The difference score for each individual is
classified as either an increase (+) or a decrease
(–) and the binomial test evaluates a null
hypothesis stating that increases and decreases
are equally likely: p(+) = p(–) = 1/2.