UNISTAT - the ultimate Excel statistics add-in

6.7.6. Two Proportions

Sample Size and Power-Two Proportions

6.7.6.1. Sample Size for Two Proportions

One sample size (n1) can be estimated assuming that the ratio of the two sample sizes (r = n2/n1) is known. The algorithm for power of the test for two proportions (see next section) is also used here, employing a half-interval search method.

The user is asked to enter:

·      Proportion 1

·      Proportion 2

·      Power of the test

·      Confidence Level

·      1 or 2 tailed test

·      n2/n1

and the program will output the estimated sample size (n1).

Example

Example 24.26 on p. 561 from Zar, J. H. (1999). Select Statistics 1Sample Size and Power EstimationTwo Proportions and the Sample Size option. Enter the following data at the next dialogue:

Sample Size and Power Estimation: Two Proportions

Sample Size

Proportion 1 =

 0.4500

Proportion 2 =

 0.2500

Power of the Test =

 0.9000

Confidence Level =

 0.9500

1 or 2 Tailed Test =

 2.0000

N2 / N1 =

 1.0000

Estimated Sample Size =

 127.2343

 

Next, click [Last Procedure Dialogue], enter n2 / n1 = 3 and click [Finish] again:

Sample Size and Power Estimation: Two Proportions

Sample Size

Proportion 1 =

 0.4500

Proportion 2 =

 0.2500

Power of the Test =

 0.9000

Confidence Level =

 0.9500

1 or 2 Tailed Test =

 2.0000

N2 / N1 =

 3.0000

Estimated Sample Size =

 83.1527

n2 is found as 3 x 83.1527 = 249.4581.

6.7.6.2. Power of the Test for Two Proportions

Power of the test in comparing two proportions is given by the following formula:

   Sample Size and Power-Two Proportions

where:

   Sample Size and Power-Two Proportions

   Sample Size and Power-Two Proportions

   Sample Size and Power-Two Proportions

   Sample Size and Power-Two Proportions

   Sample Size and Power-Two Proportions

   Sample Size and Power-Two Proportions

·      Sample Size and Power-Two Proportionsis the (one or) two-tailed critical value from the standard normal distribution for Type I error probability α.

The user is expected to enter:

·      Sample Size 1

·      Sample Size 2

·      Proportion 1

·      Proportion 2

·      Confidence Level

·      1 or 2 tailed test

and the program will output the two Z-statistics A and B and the power of the test computed from the above equation. For a one-tailed test the power of the test is computed from:

   Sample Size and Power-Two Proportions

Example

Example 24.16 on p. 554 from Zar, J. H. (2010). Select Statistics 1Sample Size and Power EstimationTwo Proportions and the Power of the Test option. Enter the following data at the next dialogue:

Sample Size and Power Estimation: Two Proportions

Power of the Test

Sample Size 1 =

 50.0000

Sample Size 2 =

 45.0000

Proportion 1 =

 0.7500

Proportion 2 =

 0.5000

Confidence Level =

 0.9500

1 or 2 Tailed Test =

 2.0000

Power of the Test:

 

2-Tail Probability =

 0.6402