Firms which decide to conduct employee attitude surveys should make sure surveys cover a random, representative sample of a target audience. Once questionnaires are designed, audiences must be identified. A decision must then be made whether to send questionnaires to an entire audience or to a representative sample. A survey is random when everyone in an audience has an equal opportunity to receive and complete a questionnaire. Surveys will be representative if response rates are acceptable, audience population estimates are precise, and confidence levels are high enough. Firms which want to increase response rates can: make participation voluntary and confidential; include personalized cover letters, and provide incentives.