The organisation should identify, early in the project, cultural and linguistic features unrelated to what the test measures and reduce their influence on the intended populations. Such features can sit in item format, materials, time limits and similar design choices, and an assessment of linguistic and cultural distance can weigh differences in religion, lifestyle, family structure, values and language. Translators should be native to the target culture and language, since fluency in the language by itself will not reveal sources of method bias. Where participant research is hard to obtain, information should be gathered from the translators, and nuisance variables can later be controlled in analysis.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.