Researchers can choose from several survey mode and contact method options when designing survey fielding plans. Each of these options is ass
ociated with known costs, yet the expected success of each fielding option, which often depends on the circumstances and preferences of a specific population of respondents, is much more ambiguous. Surveys with multiple rounds present an opportunity for researchers to identify the most successful and cost-efficient survey fielding plan in one round, to guide survey fielding in subsequent rounds. During the fielding of the 2015 Comprehensive Primary Care (CPC)Practice Survey, sponsored by the Centers for Medicare & Medicaid Services, we implemented a two-by-two survey experiment with a subset of our sample of practice managers for primary care practice sitesfor whom we did not have valid email addresses, only practice mailing addresses. We randomly assigned 269 practice managers to receive a letter that directed the practice to either (1) complete the surveyonline with a personalized web login, or (2) complete an enclosed hard copy questionnaire. We also randomly assigned the practice managers in each of those groups to be sent all outreach material using either (1) regular USPS mail or (2) FedEx next-day shipping. Though practice managers in the online and FedEx groups responded to the survey more quickly initially, over time the difference in cumulative response rates between groups disappeared. In further analyses of this experiment, we will examine the impact of survey mode (online versus hard copy) and mail method (USPS versus FedEx) on data quality, representativeness, and cost. We will recommend the best mode and procedures for future rounds of the CPC Practice Survey, as well as reflect on the general lessons learned on conducting fielding experiments in multi-round surveys.