Noncentrality Interval Estimation and the Evaluation of Statistical Models

Select the Quick tab of the Single Proportion: Interval Estimation dialog box to access options to implement three procedures for interval estimation of the population proportion. For an extensive discussion of these and other related methods, consult Newcombe (1998).

Observed Proportion (p). Enter the observed value of the sample proportion in the Observed Proportion (p) field.

Sample Size (N). In the Sample Size (N) field, enter the sample size (N) for the experimental group used in the study.

Conf. Level. Enter the confidence level for the confidence interval calculation in the Conf. Level field. Note that confidence limits are presented for three quantities:

Pi (Exact). The upper and lower limits are for a confidence interval on the population proportion. These confidence intervals are the "exact, Clopper-Pearson" confidence intervals (See method 5, Newcombe, 1998, p. 859.)

Pi (Approximate). These upper and lower limits for employ a score method with continuity correction. (See method 4, Newcombe, 1998, p. 859.)

Pi (Crude). These upper and lower limits are the simple asymptotic method without a continuity correction, using the normal approximation to the binomial. (See method 1, Newcombe, 1998, p. 859.)

Automatic rounding. Notice that, in order to provide the most accurate exact confidence intervals, the sample proportion entered in this dialog should not be subject to rounding error. As an example, suppose N =7, the number of positive responses is 3, and the sample proportion is entered as .429, which is 3/7 rounded to three decimal places. STATISTICA automatically replaces .429 with the "possible" sample proportion that is closest to .429. This corrected value replaces the .429 in the Observed Proportion (p) field and in the results spreadsheet.