Increasing STEM Undergraduate Participation in Innovative Activities: Field Experimental Evidence v1

Summary of project and findings: Understanding how to increase innovative output is critical for effective science and technology policy design. We test whether the pool of innovators can be increased by providing direct financial transfers to those with the skills required to perform high impact innocation but are reluctant to apply their skills to innovative activities. We test this in the context of an innovation contest for STEM students at one of the top engineering and computer science schools in the world. We find that this financial inducement does increase the total innovative output in the contest. However, we find important quality differences between the innovative performance of participants who participate after receiving the inducement and those who self-select into participation without the inducement that highlight the potential importance of targeting inducement interventions towards specific subgroups to maximize the returns of these efforts. Methods: We designed and ran an innovation contest for undergraduate students registered in a computer science or engineering major at UC San Diego. The contest was promoted by the UCSD CSE department over a two-month period in early 2017. Submissions made by the contest deadline were evaluated by five technology industry participants who evaluated each submission across four categories; functionality, user-friendliness, novelty, and potential commercial value and provided a score of 1-5 on each category for a total score maximum of 20. The developers of the top three applications were awarded prize money. First place received $5,000, second place received $2,000, and third place received $1,000. We randomly selected 1,000 students who did not sign up for the contest by the deadline to receive our inducement treatment. Taking up the inducement offer did not require students to do anything other than put their name on the list of contest participants and agree to receive emails about the contest. In order to provide this incentive, the contest sign-up deadline was extended by one week. Students who had already signed up to participate were informed about the sign-up deadline extension and monetary incentive being offered to some students to increase the participant pool. They were also told they would receive the same amount of money being offered to the students who had not yet signed up. The innovation contest began the day after the extended contest sign-up deadline at which point the contest problem was revealed to students. We also implemented an encouragement treatment to test whether the self-selected and induced samples responded differently to confidence boosting messages. To implement this treatment, we randomly selected induced and self-selected students to receive 4 confidence boosting emails over the course of the contest. The text included in the emails differs from one email to the next, but they are all written to provide versions of messages that have been shown to correlate with employee satisfaction and productivity in organizational behavior research.