Q28 of 38 · Test design
What is orthogonal array testing and when is it more useful than standard pairwise?
Short answer
Short answer: Orthogonal array testing uses predefined combinatorial arrays (L4, L8, L9, L16) that guarantee balanced coverage of factor combinations with a known minimum number of tests. It is more useful than ad-hoc pairwise when the inputs have equal numbers of levels and you need a balanced, provably minimal test set.
Detail
An orthogonal array is a mathematical table where each row is a test case and each column is an input factor. The array is balanced: each value of each factor appears the same number of times across all tests, and every pair of values appears equally often.
The advantage over pairwise: the coverage guarantee is formally provable and the test count is fixed and minimal for a given number of factors and levels. The trade-off is rigidity — orthogonal arrays require inputs to have equal levels (e.g., all have 2 options, or all have 3), which rarely matches real features cleanly.
In practice, use orthogonal arrays when you have a hardware configuration or platform matrix (e.g., 3 OS × 3 browsers × 3 screen sizes), where the equal-levels assumption holds naturally. For software features with unequal input options, pairwise tools are more practical.