Test for one proportion
The Test for one proportion in the Tests menu can be used to test the hypothesis that an observed proportion is equal to a pre-specified proportion.
This test is not performed on data in the data table, but on statistics you enter in a dialog box.
- Observed proportion (%): the observed proportion, expressed as a percentage.
- Sample size: the sample size or total number of observations.
- Null Hypothesis value (%): the pre-specified proportion (the value to compare the observed proportion to), expressed as a percentage.
The program calculates:
- the 95% Confidence Interval (CI) for the observed proportion
- z statistic and associated P-value.
If the P-value is less than 0.05, the hypothesis that the observed proportion is equal to the pre-specified proportion value is rejected, and the alternative hypothesis that there is a significant difference between the two proportions can be accepted.
The significance level, or P-value, is calculated using a general z-test (Altman, 1991):
where p is the observed proportion; pexp is the Null hypothesis (or expected) proportion; and se(p) is the standard error of the expected proportion:
The P-value is the area of the normal distribution that falls outside ±z (see Values of the Normal distribution table).
SciStat.com calculates the "exact" Clopper-Pearson confidence interval for the observed proportion (Clopper & Pearson, 1934; Fleis et al., 2003).
- Altman DG (1991) Practical statistics for medical research. London: Chapman and Hall.
- Clopper C, Pearson ES (1934) The use of confidence or fiducial limits illustrated in the case of the binomial. Biometrika 26:404–413.
- Fleiss JL, Levin B, Paik MC (2003) Statistical methods for rates and proportions, 3rd ed. Hoboken: John Wiley & Sons. (p. 26)
- Binomial proportion confidence interval on Wikipedia.
Go to Test for one proportion.