An experiment is said to be 'balanced' when alternative manipulations occur with equal frequency. For example:
- In MaxDiff experiments, balance means that all alternatives occur the same number of times.
- In Choice Modeling, balance means that all levels of all attributes appear the same number of times.
Balance is often assumed by the simplest statistical methods used to analyze experiments but it is not necessary when applying more advanced methods to analyze experiments (e.g., regression).