Statistical Testing

 

You can run a variety of significance tests including Chi-Square. Chi-Square is used to test the independence of two variables. Statistical testing validates claims about differences in your results. Means and percents show various differences in your results, whereas statistical testing can measure the importance of those differences.

 

WinCross offers a variety of significance tests that can be applied across any number of columns in a banner. For each row of the table that includes means or percents, the tests are conducted to report the difference between selected columns. You indicate one or two confidence levels to measure.

 

Statistical testing between percents applies to the percent displayed. If more than one percent is displayed (for example, vertical percent and horizontal percent), the statistical testing applies to the vertical percent. If vertical percent is not displayed and more than one percent is displayed (for example, horizontal percent and constant percent), the statistical testing applies to the horizontal percent.

 

Statistical testing between percents only applies to constant percents if constant percent is the only percent displayed and if the statistical test being done is an independent test (Independent (using unpooled proportions) or Independent (using pooled proportions)).

 

WinCross significance tests are two-tailed tests.

 

IMPORTANT: Statistical testing must be selected for a banner AND selected for a table.

 

To specify statistical testing for a table, use any of the following:

Use Setup|Banners to specify testing for a banner:

Means tests:

WinCross selects T-Test

Independent T-Tests

Dependent Paired/Overlap T-Tests

One-way ANOVA

Percents tests:

WinCross selects Z-Test

Independent Z-Tests

Dependent Paired/Overlap Z-Tests

Chi-Square

Related topics:

Comparison Groups

Statistical Test To Use

Show Statistical Footnote Below Table

Advanced processing features