Skip Headers
Oracle® Database SQL Reference
10g Release 2 (10.2)

Part Number B14200-02
Go to Documentation Home
Home
Go to Book List
Book List
Go to Table of Contents
Contents
Go to Index
Index
Go to Master Index
Master Index
Go to Feedback page
Contact Us

Go to previous page
Previous
Go to next page
Next
PDF · Mobi · ePub

STATS_BINOMIAL_TEST

Syntax

Description of stats_binomial_test.gif follows
Description of the illustration stats_binomial_test.gif

Purpose

STATS_BINOMIAL_TEST is an exact probability test used for dichotomous variables, where only two possible values exist. It tests the difference between a sample proportion and a given proportion. The sample size in such tests is usually small.

This function takes four arguments: expr1 is the sample being examined. expr2 contains the values for which the proportion is expected to be, and p is a proportion to test against. The fourth argument is a return value of type VARCHAR2. If you omit the fourth argument, the default is TWO_SIDED_PROB. The meaning of the return values is shown in Table 5-3.

Table 5-3 STATS_BINOMIAL Return Values

Return Value Meaning

TWO_SIDED_PROB

The probability that the given population proportion, p, could result in the observed proportion or a more extreme one.

EXACT_PROB

The probability that the given population proportion, p, could result in exactly the observed proportion.

ONE_SIDED_PROB_OR_MORE

The probability that the given population proportion, p, could result in the observed proportion or a larger one.

ONE_SIDED_PROB_OR_LESS

The probability that the given population proportion, p, could result in the observed proportion or a smaller one.


EXACT_PROB gives the probability of getting exactly proportion p. In cases where you want to test whether the proportion found in the sample is significantly different from a 50-50 split, p would normally be 0.50. If you want to test only whether the proportion is different, then use the return value TWO_SIDED_PROB. If your test is whether the proportion is more than the value of expr2, then use the return value ONE_SIDED_PROB_OR_MORE. If the test is to determine whether the proportion of expr2 is less, then use the return value ONE_SIDED_PROB_OR_LESS.

STATS_BINOMIAL_TEST Example The following example determines the probability that reality exactly matches the number of men observed under the assumption that 69% of the population is composed of men:

SELECT AVG(DECODE(cust_gender, 'M', 1, 0)) real_proportion,
       STATS_BINOMIAL_TEST
         (cust_gender, 'M', 0.68, 'EXACT_PROB') exact,
       STATS_BINOMIAL_TEST
         (cust_gender, 'M', 0.68, 'ONE_SIDED_PROB_OR_LESS') prob_or_less
  FROM sh.customers;