Can Innovation Be Taught? Evidence from a School-Based RCT in India

Saloni Gupta

Published:
ERCT Check Date:
DOI:
  • mathematics
  • science
  • K12
  • Asia
  • project-based learning
0
  • C

    The study utilized a clustered randomized controlled trial design randomizing at the school level, which satisfies the requirement for class-level or higher randomization.

    "The study was conducted as a clustered randomized controlled trial across 80 schools, with 40 assigned to treatment and 40 to control." (p. 3)

  • E

    The primary innovation outcomes rely on custom-developed measures rather than standardized exams, although standardized tests were used for secondary academic outcomes.

    "Because existing measures of innovation are scarce and rarely validated in low-income school settings, I develop a novel, multi-method toolkit..." (p. 2)

  • T

    The intervention spanned two full academic years, significantly exceeding the one-term duration requirement.

    "The curriculum delivered 60 hours of training through weekly 90-minute sessions over two academic years." (p. 4)

  • D

    The control group's condition (self-directed preparation) and baseline characteristics are clearly documented and compared to the treatment group.

    "Control schools received only information about an annual innovation challenge with detailed criteria but no structured training..." (p. 4)

  • S

    The study randomized 80 schools to treatment or control conditions, satisfying the school-level randomization requirement.

    "The study was conducted as a clustered randomized controlled trial across 80 schools, with 40 assigned to treatment and 40 to control." (p. 3)

  • I

    The intervention was implemented by an independent NGO (Inqui-Lab), while the evaluation was conducted by an academic researcher from Stanford.

    "The intervention was implemented by Inqui-Lab Foundation... I conduct a large-scale clustered randomized controlled trial..." (pp. 2-4)

  • Y

    The study tracked students over two full academic years, exceeding the one-year duration requirement.

    "I follow approximately 5,000 students from Grade 8 through Grade 9 during academic years 2022-2024." (p. 4)

  • B

    The intervention provided additional resources (kits, training) that were integral to the treatment being tested, while educational time was balanced across groups.

    "This created a research design that tests whether structured training improves innovation outcomes compared to self-directed learning." (p. 4)

  • R

    No independent replications of this specific intervention were found in peer-reviewed literature.

    "This paper provides the first experimental evidence that students can learn core innovation-related skills." (p. 2)

  • A

    The study assesses Math and Science but does not assess other core subjects like Language Arts or Social Studies using standardized exams.

    "I assess math and science performance to examine potential spillovers to academic domains... I measure math and science achievement using standardized assessments..." (p. 16)

  • G

    The study tracked students through Grade 9 but did not track them until graduation.

    "I follow approximately 5,000 students from Grade 8 through Grade 9 during academic years 2022-2024." (p. 4)

  • P

    The study was pre-registered with the AEA Trial Registry prior to the start of data collection.

    "The RCT is registered with the AEA Trial Registry (RCT ID: AEARCTR-0009696)." (p. 2)

Abstract

Innovation fuels long-run economic growth, yet education systems in developing countries often overlook the skills required for innovation. This paper provides the first experimental evidence that students can learn core innovation-related skills. I conduct a large-scale clustered randomized controlled trial with 4,800 eighth-grade students in India, comparing a two-year school-based innovation training program to a control group with access to self-directed project opportunities. Because existing measures of innovation are scarce and rarely validated in low-income school settings, I develop a novel, multi-method toolkit that includes a strategic exploration task, expert-rated idea quality assessments, and a live funding competition. Treated students significantly outperformed controls across all three domains (0.10-0.24 SD) and secured 24 percentage points more external funding.

Full Article

ERCT Criteria Breakdown

  • Level 1 Criteria

    • C

      Class-level RCT

      • The study utilized a clustered randomized controlled trial design randomizing at the school level, which satisfies the requirement for class-level or higher randomization.
      • "The study was conducted as a clustered randomized controlled trial across 80 schools, with 40 assigned to treatment and 40 to control." (p. 3)
      • Relevant Quotes: 1) "I conduct a large-scale clustered randomized controlled trial with 4,800 eighth-grade students in India..." (p. 2) 2) "The study was conducted as a clustered randomized controlled trial across 80 schools, with 40 assigned to treatment and 40 to control." (p. 3) Detailed Analysis: The ERCT standard requires randomization at the class level or stronger (school level). The paper explicitly states that the randomization was conducted at the school level ("across 80 schools"). This design avoids contamination within schools and satisfies the requirement for a robust randomization unit. Since school-level randomization is stronger than class-level, this criterion is met. Final sentence explaining if criterion C is met: Criterion C is met because the study randomized participants at the school level, exceeding the class-level requirement.
    • E

      Exam-based Assessment

      • The primary innovation outcomes rely on custom-developed measures rather than standardized exams, although standardized tests were used for secondary academic outcomes.
      • "Because existing measures of innovation are scarce and rarely validated in low-income school settings, I develop a novel, multi-method toolkit..." (p. 2)
      • Relevant Quotes: 1) "Because existing measures of innovation are scarce and rarely validated in low-income school settings, I develop a novel, multi-method toolkit that includes a strategic exploration task, expert-rated idea quality assessments, and a live funding competition." (p. 2) 2) "I assess math and science performance to examine potential spillovers... using standardized assessments adapted from school textbooks and the Mindspark study (Muralidharan et al., 2019)." (p. 16) 3) "I created the Innovation-S scale, a novel psychometric tool designed to measure the quality of innovative ideas... developed through a process involving in-depth interviews..." (p. 14) Detailed Analysis: The ERCT standard requires the use of standardized exam-based assessments that are not specially designed for the study. While the secondary academic outcomes use "adapted" standardized assessments (Mindspark/textbooks), the primary outcomes related to the study's main research question (innovation) are explicitly described as a "novel... toolkit" and a "novel psychometric tool" (Innovation-S scale) developed by the author. The "multi-armed bandit task" is also an adaptation designed for the field. As the core assessments were created specifically for this research due to a lack of existing measures, this criterion is not met. Final sentence explaining if criterion E is not met: Criterion E is not met because the study relies on custom-developed measures (Innovation-S scale, adapted bandit task) for its primary outcomes rather than widely recognized standardized exams.
    • T

      Term Duration

      • The intervention spanned two full academic years, significantly exceeding the one-term duration requirement.
      • "The curriculum delivered 60 hours of training through weekly 90-minute sessions over two academic years." (p. 4)
      • Relevant Quotes: 1) "The curriculum delivered 60 hours of training through weekly 90-minute sessions over two academic years." (p. 4) 2) "I follow approximately 5,000 students from Grade 8 through Grade 9 during academic years 2022-2024." (p. 4) Detailed Analysis: The standard requires outcomes to be measured at least one full academic term after the intervention begins. This study implemented the intervention over "two academic years" (2022-2024) with data collection at baseline, midline (end of Year 1), and endline (end of Year 2). This duration far exceeds the minimum requirement of one term. Final sentence explaining if criterion T is met: Criterion T is met because the intervention and measurement period covered two full academic years.
    • D

      Documented Control Group

      • The control group's condition (self-directed preparation) and baseline characteristics are clearly documented and compared to the treatment group.
      • "Control schools received only information about an annual innovation challenge with detailed criteria but no structured training..." (p. 4)
      • Relevant Quotes: 1) "Control schools received only information about an annual innovation challenge with detailed criteria but no structured training, similar to statewide challenges previously conducted by the Telangana government..." (p. 4) 2) "Comparison group teams were encouraged to use this same time to develop their innovation ideas on their own." (p. 11) 3) "Table 2 presents baseline characteristics and balance checks across treatment and control groups." (p. 17) Detailed Analysis: The paper provides a detailed description of the control group, specifying that they participated in "self-directed preparation" and received information about the challenge but no formal training. Table 2 exhaustively lists demographic information, baseline academic scores, and other characteristics for the control group alongside the treatment group. This satisfies the requirement for detailed documentation. Final sentence explaining if criterion D is met: Criterion D is met because the study clearly describes the control group's activities and provides detailed baseline demographic and performance data.
  • Level 2 Criteria

    • S

      School-level RCT

      • The study randomized 80 schools to treatment or control conditions, satisfying the school-level randomization requirement.
      • "The study was conducted as a clustered randomized controlled trial across 80 schools, with 40 assigned to treatment and 40 to control." (p. 3)
      • Relevant Quotes: 1) "The study was conducted as a clustered randomized controlled trial across 80 schools, with 40 assigned to treatment and 40 to control." (p. 3) 2) "I employed a stratified randomization design... I randomly assigned 40 schools to the treatment group and 40 to the comparison group within each stratum." (p. 10) Detailed Analysis: The criterion requires randomization to occur at the school level. The author explicitly states that 80 schools were selected and randomly assigned (40 treatment, 40 control) using stratified randomization. This perfectly matches the requirement for School-level RCT. Final sentence explaining if criterion S is met: Criterion S is met because the randomization was performed at the school level across 80 different institutions.
    • I

      Independent Conduct

      • The intervention was implemented by an independent NGO (Inqui-Lab), while the evaluation was conducted by an academic researcher from Stanford.
      • "The intervention was implemented by Inqui-Lab Foundation... I conduct a large-scale clustered randomized controlled trial..." (pp. 2-4)
      • Relevant Quotes: 1) "The intervention was implemented by Inqui-Lab Foundation, an NGO founded by engineers-turned-teachers who developed a structured pedagogy..." (p. 4) 2) "I conduct a large-scale clustered randomized controlled trial... Saloni Gupta... Stanford King Center on Global Development" (p. 2) 3) "I am extremely grateful to the... Inqui-Lab Foundation... for their leadership and collaboration in this study." (p. 2) Detailed Analysis: The standard requires the study to be conducted independently from the intervention designers. The text identifies "Inqui-Lab Foundation" as the entity that developed and implemented the program. The author (Saloni Gupta) is a researcher at Stanford University evaluating the program. This separation between the NGO delivering the program and the academic researcher evaluating it satisfies the independence criterion. Final sentence explaining if criterion I is met: Criterion I is met because the intervention was designed and implemented by Inqui-Lab Foundation, while the evaluation was conducted by an external academic researcher.
    • Y

      Year Duration

      • The study tracked students over two full academic years, exceeding the one-year duration requirement.
      • "I follow approximately 5,000 students from Grade 8 through Grade 9 during academic years 2022-2024." (p. 4)
      • Relevant Quotes: 1) "I follow approximately 5,000 students from Grade 8 through Grade 9 during academic years 2022-2024." (p. 4) 2) "The curriculum delivered 60 hours of training through weekly 90-minute sessions over two academic years." (p. 4) 3) "The experiment spans two academic years, with data collection at three primary points: baseline in August 2022, midline in March 2023... and endline in March 2024..." (p. 11) Detailed Analysis: The criterion requires outcomes to be measured at least one full academic year after the intervention begins. This study tracked students and delivered the intervention over a period of two academic years (August 2022 to March 2024), clearly satisfying the year-long requirement. Final sentence explaining if criterion Y is met: Criterion Y is met because the study spans two academic years with start and end dates covering roughly 19 months.
    • B

      Balanced Control Group

      • The intervention provided additional resources (kits, training) that were integral to the treatment being tested, while educational time was balanced across groups.
      • "This created a research design that tests whether structured training improves innovation outcomes compared to self-directed learning." (p. 4)
      • Relevant Quotes: 1) "This created a research design that tests whether structured training improves innovation outcomes compared to self-directed learning." (p. 4) 2) "Treatment schools scheduled their weekly 90-minute sessions during independent study periods... Comparison group teams were encouraged to use this same time to develop their innovation ideas on their own. Thus, both groups utilized the same independent study time allocation..." (p. 11) 3) "Each classroom also received an 'innovation kit' valued at Rs. 15,000 ($180 USD)..." (p. 9) Detailed Analysis: The study tests "structured training" (which includes curriculum and kits) against "self-directed learning." The treatment group received extra resources ($180 kits), which were not provided to the control group. However, the study explicitly frames the testing of this "structured training" (including the resources) as the treatment variable compared to self-directed efforts. The authors took specific steps to balance the *time* variable: both groups used the same "independent study periods" (90 mins/week), ensuring that the control group had comparable dedicated time for the task. Since the additional resources were an integral part of the treatment being tested and time was balanced, the criterion is met. Final sentence explaining if criterion B is met: Criterion B is met because the additional resources were integral to the treatment variable being tested, and educational time was strictly balanced between groups.
  • Level 3 Criteria

    • R

      Reproduced

      • No independent replications of this specific intervention were found in peer-reviewed literature.
      • "This paper provides the first experimental evidence that students can learn core innovation-related skills." (p. 2)
      • Relevant Quotes: 1) "This paper provides the first experimental evidence that students can learn core innovation-related skills." (p. 2) Detailed Analysis: The paper claims to provide the "first experimental evidence" on this topic. As of the check date, there are no available citations or references pointing to independent replications of the Inqui-Lab Foundation innovation curriculum by other research teams in different contexts published in peer-reviewed journals. The study is a working paper from 2025 and has not yet been reproduced externally. Final sentence explaining if criterion R is met: Criterion R is not met because this is an original study without referenced independent replications.
    • A

      All-subject Exams

      • The study assesses Math and Science but does not assess other core subjects like Language Arts or Social Studies using standardized exams.
      • "I assess math and science performance to examine potential spillovers to academic domains... I measure math and science achievement using standardized assessments..." (p. 16)
      • Relevant Quotes: 1) "I assess math and science performance to examine potential spillovers to academic domains... I measure math and science achievement using standardized assessments adapted from school textbooks..." (p. 16) 2) "Table 6: Treatment Effects on Academic Outcomes... Math achievement... Science achievement" (p. 25) 3) "I also gathered data on subject preferences across eight school subjects..." (p. 17) Detailed Analysis: The criterion requires measuring impact on "all main subjects taught in the school." While the author collected data on "subject preferences" for eight subjects, the actual *exam-based assessments* were strictly limited to Math and Science. There is no evidence of standardized testing for languages or social studies, which are core subjects in Grade 8/9 curricula. Furthermore, since criterion E (Exam-based Assessment) for the primary outcome was not met, criterion A also fails. Final sentence explaining if criterion A is met: Criterion A is not met because exams were limited to math and science only, excluding other core subjects.
    • G

      Graduation Tracking

      • The study tracked students through Grade 9 but did not track them until graduation.
      • "I follow approximately 5,000 students from Grade 8 through Grade 9 during academic years 2022-2024." (p. 4)
      • Relevant Quotes: 1) "I follow approximately 5,000 students from Grade 8 through Grade 9 during academic years 2022-2024." (p. 4) 2) "Endline Mar-2024... Funding Round Nov-2024" (p. 11) Detailed Analysis: The study concludes at the end of Grade 9 (with a competition in the following academic year). Graduation from secondary school in India typically occurs at Grade 10 or 12. The study explicitly states tracking occurred "from Grade 8 through Grade 9." No subsequent papers or follow-up studies were found tracking this specific cohort to graduation. Final sentence explaining if criterion G is met: Criterion G is not met because tracking ended at Grade 9 rather than continuing through graduation.
    • P

      Pre-Registered

      • The study was pre-registered with the AEA Trial Registry prior to the start of data collection.
      • "The RCT is registered with the AEA Trial Registry (RCT ID: AEARCTR-0009696)." (p. 2)
      • Relevant Quotes: 1) "The RCT is registered with the AEA Trial Registry (RCT ID: AEARCTR-0009696)." (p. 2) 2) "Baseline Aug-2022" (p. 11) Detailed Analysis: The paper identifies the registry ID AEARCTR-0009696. A check of the AEA Social Science Registry confirms that this trial ("Can Innovation be Taught? Evidence from a Randomized Control Trial in India") was registered. The registration date corresponds to July 2022, which is prior to the baseline data collection date of August 2022 stated in the paper. Final sentence explaining if criterion P is met: Criterion P is met because the study protocol was pre-registered before data collection began.

Request an Update or Contact Us

Are you the author of this study? Let us know if you have any questions or updates.

Have Questions
or Suggestions?

Get in Touch

Have a study you'd like to submit for ERCT evaluation? Found something that could be improved? If you're an author and need to update or correct information about your study, let us know.

  • Submit a Study for Evaluation

    Share your research with us for review

  • Suggest Improvements

    Provide feedback to help us make things better.

  • Update Your Study

    If you're the author, let us know about necessary updates or corrections.