The software calculates exact, mid-p values and asymptotic p-values in 2x2 contingency tables. Both unconditional and conditional test statistics are available.

Download and installation instructions can be found here.

Start the program by running the "main.py" file. A window similar to the snapshot below should appear. The interface consists of five different pages and a row of buttons. The pages are accessed by clicking on the tabs labeled "P-value", "Power", etc. A description of the use and purpose of each page can be found under their respective headers.

These buttons are available for all pages and do the following:

• Stop - Stop current calculations.
• Save - Save options (calculation options, startup values and user options) to the file "options.txt". If this file should be lost, corrupted or otherwise unreadable, a new file will be created the next time the program is started.
• Restore - Restore all options to default values and save the "options.txt" file.
• Help - Launch the user manual (this page).
• Exit - Quit the application.

• Conditional - Calculate p- and mid p-values using a conditional exact test (such as Fisher's exact test).
• Unconditional - Calculate p- and mid p-values using an unconditional exact test (such as Barnard's exact test).
• Asymptotic - Calculate p-values using an asymptotic test.

During and after the calculations a report will be written in the "Report" window. The report includes which options were used, some intermediary values, the elapsed computing time, and the resulting p-value(s). The options controlling the calculations (like the hypothesis and the test statistic) may be set in the options page. The contents of the report window may be saved as a text file by using the "Save report" button. (Some of the text may be selected and copied into the copy-buffer or saved directly to file. Use the right mouse button to access this feature.)

Details

Calculation details from the last test will be displayed by pressing the "Details" button. This feature is only available for conditional and unconditional tests. Some properties of the figures (like the colours of the acceptance region and the rejection region) may be changed by using the buttons in the top of the windows. The windows are also resizable and the contents may be saved to file (more on this here).

• Conditional test details A window showing the distribution of the test statistic values will appear. In general, the test statistic, T(n), is a function of all possible tables. But in a conditional test, the row and column sums are fixed, so that the table is given as a function of x₁₁ only. The observed table is the one tagged with an "X". Position the mouse above a column to see the actual T-value of this table.
  
• Unconditional test details Three windows will appear: one showing the distribution of the test statistic values and two showing the p-value and mid p-value as functions of the nuisance parameter (π). The test statistic window will now show a two-dimensional function T(n) as no conditions are set on the n₊₁ or n₊₂ values. (If a rather large table was computed, the test statistic window may take some time being created.)
  Below is an example of the window showing the p-value as a function of π. This is a visualisation of how the maximum p(π)-value is found: First the π-interval, (0,1) if no confidence interval is used, is partitioned into a number of discreet values and their p-value is calculated. The greatest of these values are then picked. A finer dicretization is then made around each of these π-values and new p-values are evaluated. The final p-value is chosen as the greatest of all calculated p-values. The number of discreet values in the first partition ("First sweep"), the number of top values selected ("Tops"), and the number of values in the second partition ("Second sweep") may be set in the options page. Clicking the "Create curve" button will produce a window with a curve representation of the column chart shown in the "First Sweep" page. The coordinates (pi, p) may be saved to an ordinary text file by using the "Save data" button.
  
  

• Conditional p - Calculate power for a conditional test, the p-value version.
• Conditional mid p - Calculate power for a conditional test, the mid p-value version.
• Conditional both - Calculate power for a conditional test, both for the p-value version and the mid p-value version.
• Unconditional p - Calculate power for an unconditional test, the p-value version.
• Unconditional mid p - Calculate power for an unconditional test, the mid p-value version.
• Unconditional both - Calculate power for an unconditional test, both for the p-value version and the mid p-value version.
• Asymptotic p - Calculate p-value-power using an asymptotic test

A report will be written in the "Report" window during and after calculations. The report includes which options were used, the elapsed computing time, and the resulting power-value(s). The options controlling the calculations (like the hypothesis and the test statistic) may be set in the options page. The contents of the report window may be saved as a text file by using the "Save report" button. (Some of the text may be selected and copied into the copy-buffer or saved directly to file. Use the right mouse button to access this feature.)

Two checkbuttons influence the way the power-values are calculated (but not the results):

• "Calculate border only" - If this button is on, only the stricktly necessary p-values (and/or mid p-values) will be calculated, i.e. the p-values defining the border of the rejection region. Otherwise all p-values (as defined by the "Pop.1 total" and "Pop.2 total" parameters) will be calculated. This is, of course, more computer time demanding. If a conditional test or asymptotic test is selected, all p-values will be calculated regardless of the checkbutton's state. Note that this option is unavailable if the Odds ratio test statistic is selected.
• "Visualise while comp." - If on, the p-values will be visualised (more on this below) in a separate window during calculations.

Details

Pressing the "Details" button will display a window showing the p-values (and/or mid p-values) used in the power calculation algorithm. Only the elements with p-values less than the significance level (α) contribute to the power value. Position the mouse above an element to see the actual p-value of this element. The colours distinguishing the acceptance region and the rejection region may be changed by using the buttons in the top of the window. The window is resizable and the contents may be saved to file (more on this here).

Once calculations have started, a window displaying the results will appear. If the checkbutton "Show remaining computing time [...]" in the user options page is on, a second popup window will appear to estimate the remaining computer time. The calculations can be stopped by pressing the "Stop" button. The two buttons labeled "Save table" and "Save full report" will create a file containing a text representation of the results. The "Save table" button will save only the actual calculated values and the headers (more on this below), while the "Save full report" button will print information on the parameters and display the table in a more readable format.

How to export table to Excel

Press the button "Save table" and select a file. The table will be saved as plain text with one line for each row (headers included). The entries in each row will be separated by a semicolon (;). In Excel, find the menu "Data" and select the "Get External Data -> Import Text File..." command. Open the text file saved above. (You may also use the command "Open..." found in the "File" menu.) In the "Text Import Wizard - Step 1", select "Delimeted" and press "Next >". In step 2, select the checkbutton labeled "Semicolon" and deselect all other checkbuttons. Press the "Next >" button. Apply data formatting (if any) in step 3 and press "Finished". Finally, choose where you want to put the data and press "OK".

• Hypothesis - Set the alternative hypothesis (and thereby the null hypothesis). This will affect all conditional, unconditional and asymptotic tests performed in the "P-value", "Power" and "Multipower" pages.
• Test statistic - Decide which test statistic to use. This will affect all conditional, unconditional and asymptotic tests performed in the "P-value" and "Power" pages (but not the "Multipower" page).
• How to maximize p(pi) - These values are used in all unconditional tests to find the maximum value of the function p(π). More on this here.
• Use confidence interval - A confidence interval for Berger and Boos correction may be applied instead of the interval (0, 1). The "from", "to" and "gamma" values have to be inserted manually, but an Excel-program ("Clopper-Pearson.xls") calculating the values can be found in the main folder (more on this here).
• Table values at startup - The startup values of the table in the "P-value" page.
• Power (multipower) values at startup - The startup values of the parameters used to calculate power in the "Power" and "Multipower" pages.

• Escape - Stop current calculations
• Control + c - Stop current calculations
• Control + s - Save options to file
• Control + r - Restore options to default values and save to file
• Control + h - Display the user manual (this page)
• Control + q - Quit application
• F1 - F5 - Go to page 1 - 5