17.3.9.1 The Two-Sample Proportion Test Dialog Box


Contents

Supporting Information

To open the Two-Sample Proportion Test dialog box:

  1. With the worksheet active, click Statistics: Hypothesis Testing: Two-Sample Proportion Test...

See Also:

Recalculate

Recalculate

Controls recalculation of analysis results

  • None
  • Auto
  • Manual

For more information, see: Recalculating Analysis Results

Input

Summarized Perform the test using summary statistics that you enter into the dialog box manually.
Number of Events of Sample1
The number of observed "events" in sample 1
Sample Size of Sample1
The total number of observations in sample 1
Number of Events of Sample2
The number of observed "events" in sample 2
Sample Size of Sample2
The total number of observations in sample 2
Raw

Perform the test using two datasets composed of "success" or "failure" values. The datasets must contain no more than two distinct values.

Input Data Form
Specify the data ranges of sample 1 and sample 2
1st Data Range
2nd Data Range
Event
The dataset value that denotes event.
Index

Perform the test on input data consisting of one sample data column and one "grouping" column. The sample data column should be in the form of "success" or "failure" and must contain no more than two distinct values. The grouping column should contain information on the two populations of interest, indexed (by row number) to the sample data column.

Input Data Form
Specify one sample data column and one grouping column.
Group Range
Data Range
Event
The dataset value that denotes even.

Proportion Test

Test Difference The hypothetical proportion difference to be tested (generally "0").
Null Hypothesis The null hypothesis to be tested.
Alternative Hypothesis Options:
  • Proportion 1 - Proportion2<> Test Difference
Perform a two-tailed test of the hypothesis that Proportion1 - Proportion2 = Test Difference.
  • Proportion 1 - Proportion2 > Test Difference
Perform an upper-tailed test of the hypothesis that Proportion1 - Proportion2 is greater than Test Difference.
  • Proportion 1 - Proportion2 < Test Difference
Perform a lower-tailed test of the hypothesis that Proportion1 - Proportion2 is less than Test Difference.
Significance Level The significance level of the test.
Confidence Interval(s) Compute confidence intervals for proportion differences.
Confidence Level(s) in % Specifies the confidence levels for which the confidence intervals will be computed. This option is only available when Confidence Interval(s) is selected.
Test Methods To learn more about test methods, see the algorithms page.
  • Normal Approximation
Compute the p value and confidence interval in result with normal approximation normal distribution
  • Use pooled p for test
Specify to use a pooled estimate of p for the test when Test Difference is 0
  • Fisher's Exact Test
Compute the p value and confidence interval with the Fisher's exact test, which is based on the Binomial Test .

Power Analysis

Unset Minimal Difference of Interest

Difference(s) for Specified Power When Minimal Difference of Interest is not set, this option is available.

Output the differences for the specified power levels. When this option is selected, the specified power levels setting will affect the power level in power curve.

Power Level(s) When check the Difference(s) for Specified Power option, you can enter the power levels separated by space in this edit box. Default power levels are 60%, 70%, 80% and 90%.

Set Minimal Difference of Interest

Minimal Difference of Interest (MDI) Specify the minimum difference you are interested in. Once you enter a value in this box, the Power Analysis graph will show the minimum sample size to detect this difference in the different powers.
Note:

For Two Sample Proportion test, there is a limitation for the Minimal Difference of Interest (MDI) setting. That is related to the difference bound which will be output in the PlotData worksheet.

  • When Proportion 1 - Proportion2<> Test Difference , MDI value should be a positive value less than the minimum value between the difference upper bound and absolute value of difference lower bound.
  • When Proportion 1 - Proportion2 > Test Difference , MDI value should be a positive value less than the difference upper bound
  • When Proportion 1 - Proportion2 < Test Difference , MDI value should be a positive value less than the absolute value of difference lower bound.
Sample Size(s) for Specified Power When check the Minimal Difference of Interest option, this option is available.

Output the Sample Sizes for the specified power levels. When this option is selected, the specified power levels setting will affect the power level in power curve. If it is selected, it will use the default power levels.

Power Level(s) When check the Sample Size(s) for Specified Power option, you can enter the target power levels separated by space in this edit box. Default power levels are 60%, 70%, 80% and 90%.
Power Based on Test Data When specify the Minimal Difference of Interest, this option is available.

Compute the power for the sample size of the test data. (the probability of rejecting the null hypothesis when it is false).

Hypothetical Power When specify the Minimal Difference of Interest, this option is available.

Compute the power of the test at various sample sizes. When check this option, it will add corresponding curve to power curve.

Hypothetical Sample Size(s) When specify the Minimal Difference of Interest and Hypothetical Power is checked, this option is available.

Specify hypothetical sample size(s) for power computation.

Output

For help with the range controls, see: Output Results

Results Log Output Select to output results to the Results Log.
Graphical Summary Report Specify whether to display the graphical summary report
Graphical Power Analysis Report Specify whether to display the graphical Power Analysis report. The power curve shows how the statistical power of the test changes as the difference in proportions varies. Power is the probability of rejecting the null hypothesis when the true difference differs from 0.
Output Results Specify the worksheet for output of test results
Output Plot Data Specify the worksheet for output the plot datasets.
Note:
  • When input data form is Raw or Indexed, Masked Data and Bad Data(missing value) table will be in the report sheet.