Two process adjustments that each appear beneficial on their own may behave differently when combined. The effect of one factor can depend on the level of another. Before adopting a combined setting, design the comparison so that this interaction can be observed rather than assumed away.
The NIST statistical glossary defines an interaction as a factor’s effect depending on other factor values. In a PCB study, that might mean a cleaning adjustment behaves differently for two assembly constructions. The example is a study question, not a claim that any particular process combination will interact.
Define factors and responses precisely
Specify the controllable settings and the outcome to measure. Avoid vague factors such as “better process” that combine several changes at once. Record the actual setting values, material identities, and measurement method for every experimental unit, together with its assigned treatment combination.
Decide which combinations are meaningful and permitted by the process. Do not test arbitrary extremes merely to fill a statistical table. The study should remain within an appropriate engineering domain and include conditions relevant to the proposed production decision.
Compare effects across conditions
| Comparison | Question |
|---|---|
| Factor A changes at one B setting | What is the effect in that condition? |
| Factor A changes at another B setting | Is the effect similar or different? |
| Replicated combinations | How much variation surrounds the apparent effects? |
| Confirmation at the proposed setting | Does the combined choice perform as expected? |
Plot responses with grouping that preserves the combinations. A single overall average for each factor can hide opposite effects in different conditions. Keep board, lot, and run-order information so that a production drift is not misidentified as interaction.
Plan the experiment before collecting data
- State the decision and candidate factors.
- Choose a design that can estimate the effects of interest.
- Plan replication and control relevant nuisance variation.
- Analyze interaction alongside practical magnitude.
- Confirm the selected combination in representative conditions.
A one-factor-at-a-time sequence generally provides limited evidence about untested combinations. If interactions matter to the decision, use a design that observes them directly. For reduced experimental designs, understand which effects cannot be separated before assigning a physical explanation.
Report where the conclusion applies. An adjustment that works for one material or product construction should not automatically become a universal standard. The useful outcome is a justified combination of settings and an explicit statement of dependencies, supported by observations that distinguish the combined behavior from the separate averages.
Move from reading to a build-ready project
Use the fabrication and assembly checklists to prepare drawings, production files and test requirements.
Open the DFM checklist