Significance testing - Overlap Correction

From mtab wikisupport
Jump to: navigation, search

If you are testing a product that represents a significant proportion of the total relative to the other products, then it can be argued that the significance testing is biased by including the tested product within the total.

Overlap correction removes each individual product from the total during a significance test of individual products relative to the total.

For example, suppose you are testing products A, B, C, and D against the total of A, B, C, D. If product A were to represent 40% of the total, then testing product A against the total will be biased by the relatively large contribution of product A to the total.

Applying overlap correction within mTAB will result in removing product A from the total; product A would be tested against (total - product A), product B tested against (total - product B) and so on.

Applying Overlap Correction:

From mTAB's spreadsheet view, select menu items Data, then Significance Indicators.

Click on the "Apply Overlap Correction" checkbox from the resultant Significance Indicator dialog.