T-Test

A t-test is a test of statistical significance used by statisticians to identifying patterns in a particular set of data that are unlikely to be the result of chance. MarketSight is a very easy-to-use web-based software application that enables you to run t-tests and other stat tests. MarketSight automatically selects the right stat test for you, based on your data, so you don’t have to choose between a t-test, an f-test, or a z-test in any given situation. MarketSight does the work for you.

Stat tests are critical in determining significance within the data, but many of us have not had the benefit of extensive statistical training. There are actually many more tests, other than the t-test, that can be appropriate, depending on whether you’re comparing means or column proportions.

The t-test, in the form of Fisher’s Least Significant Difference (LSD), for example, is used to test whether or not two independent Means are equal to one another. In MarketSight, a modified version of the t-test is run for multiple response variables (MRVs) - those questions where multiple answers may be selected, for both contrast and pairwise tests.

Based on the t-test or other tests that are run, MarketSight can highlight items of statistical significance in your crosstab using color coding, helping you to focus on these important items within the data. The name of the test and the p-value can optionally be displayed in a pop-up.

MarketSight an attractive solution for data analysis because it selects the right test for you, a tremendous benefit for users who don’t have an in-depth statistical background. It’s also an excellent tool for stat experts, because they can set advanced options for their particular needs.

MarketSight automatically runs t-tests and other Stat tests

t test

MarketSight charts can include statistical significance notation on charts

t test

Common Questions about T-Test and MarketSight

How does the t-test apply to multiple response variables?

MRV’s are first tagged within MarketSight. The solution will then run a modified version of the t-test for both Contrast and Pairwise tests.

Can MarketSight run corrected t-tests?

Yes. A corrected contrast t-test and a corrected pairwise t-test are run when the option to correct for Type I errors in all mean comparisons is applied.

If I am a stat expert, do I have control over MarketSight stat settings?

Yes. MarketSight automates all of the backend settings, but stat experts can set advanced options, such as adding a second confidence level, treating each pair of columns as an independent sample, and changing the base for statistical calculations.