Impact of Incentives on Physician Participation in Research Surveys: Randomized Experiment

Background Web-based surveys can be effective data collection instruments; however, participation is notoriously low, particularly among professionals such as physicians. Few studies have explored the impact of varying amounts of monetary incentives on survey completion. Objective This study aims to conduct a randomized study to assess how different incentive amounts influenced survey participation among neurologists in the United States. Methods We distributed a web-based survey using standardized email text to 21, 753 individuals randomly divided into 5 equal groups (≈4351 per group). In phase 1, each group was assigned to receive either nothing or a gift card for US 10, 20, 50, or 75, which was noted in the email subject and text. After 4 reminders, phase 2 began and each remaining individual was offered a US 75 gift card to complete the survey. We calculated and compared the proportions who completed the survey by phase 1 arm, both before and after the incentive change, using a chi-square test. As a secondary outcome, we also looked at survey participation as opposed to completion. Results For the 20, 820 emails delivered, 879 (4. 2%) recipients completed the survey; of the 879 recipients, 622 (70. 8%) were neurologists. Among the neurologists, most were male (412/622, 66. 2%), White (430/622, 69. 1%), non-Hispanic (592/622, 95. 2%), graduates of American medical schools (465/622, 74. 8%), and board certified (598/622, 96. 1%). A total of 39. 7% (247/622) completed their neurology residency more than 20 years ago, and 62. 4% (388/622) practiced in an urban setting. For phase 1, the proportions of respondents completing the survey increased as the incentive amount increased (46/4185, 1. 1%; 76/4165, 1. 8%; 86/4160, 2. 1%; 104/4162, 2. 5%; and 119/4148, 2. 9%, for US 0, 10, 20, 50, and 75, respectively; P<. 001). In phase 2, the survey completion rate for the former US 0 arm increased to 3% (116/3928). Those originally offered US 10, 20, 50, and 75 who had not yet participated were less likely to participate compared with the former US 0 arm (116/3928, 3%; 90/3936, 2. 3%; 80/3902, 2. 1%; 88/3845, 2. 3%; and 74/3878, 1. 9%, for US 0, 10, 20, 50, and 75, respectively; P=. 03). For our secondary outcome of survey participation, a trend similar to that of survey completion was observed in phase 1 (55/4185, 1. 3%; 85/4165, 2%; 96/4160, 2. 3%; 118/4162, 2. 8%; and 135/4148, 3. 3%, for US 0, 10, 20, 50, and 75, respectively; P<. 001) and phase 2 (116/3928, 3%; 90/3936, 2. 3%; 80/3902, 2. 1%; 88/3845, 2. 3%; and 86/3845, 2. 2%, for US 0, 10, 20, 50, and 75, respectively; P=. 10). Conclusions As expected, monetary incentives can boost physician survey participation and completion, with a positive correlation between the amount offered and participation.