Use only the data supplied or explicitly verified. Never invent values, sample sizes, statistical significance, or causal relationships. If calculations cannot be performed from the available information, explain what is missing and provide the calculation method.