Hooked on podcasts: evidence from a quasi-experimental approach

Abstract

The podcast is a tool that economic educators can employ in the classroom as an alternative approach to convey content. Using a unique quasi-experimental approach in four undergraduate economics classes, we measure student perceptions of podcast use and the effectiveness of the podcast as a pedagogical tool in the economics curriculum. This experiment controls for demographic characteristics, educational background, learning styles, instructor fixed effects, and class fixed effects while measuring the impact of podcast use on student learning outcomes. We find that exam scores of students assigned podcasts to supplement their learning were statistically no different than those of students assigned printed news articles to supplement their learning. However, in terms of perception, students believed they were learning more from podcasts than news articles.

Keywords:

• Learning styles
• pedagogical tool
• podcast
• quasi-experimental

JEL CODES:

• A20
• A22

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Podcasts for the point of view of this research were created by a third party (National Public Radio) and not by the students or by the instructors. The podcasts were made available as audio files through the university’s learning management system for the students to access as homework assignments outside of class time. The podcasts were all between 10 and 20 minutes in length. A representative podcast can be found at https://www.npr.org/sections/money/2012/07/31/157665908/episode-391-the-anti-addiction-pill-thats-big-business-for-drug-dealers.

2 Comparing podcast use or newspaper article use to no real-world homework assignments is unrealistic in today’s economics classroom and would cause unnecessary harm to students’ grades in a treatment group without access to real-world homework assignments. While such an experiment may yield interesting results, it would be difficult to find a university ethics committee willing to approve such an experiment.

3 We chose the third day of class because that day signaled the end of the drop/add period for students. We chose the second-to-last day of class because that was the last day of lecture; the final exam was held on the last day of class.

4 Similar time periods are defined as midday classes, all starting by 12:00 pm and ending before 3:00 pm, meeting every Monday, Wednesday, and Friday. IRB approval does not permit further description to preserve student, instructor, and university anonymity.

5 Even though printed news articles are not perfect substitutes for podcasts, the standard inclusion of news articles in the classroom makes them a valid control group as a close substitute. Comparing podcasts to a control group receiving no supplemental material would cause unnecessary harm to students and their learning objectives.

6 As pointed out by a discussant at the Conference on Teaching and Research in Economic Education, some podcasts had visual stimuli on the Web site from which they could be downloaded, while some news articles were accompanied by photos or graphs. The audio files were uploaded directly into the learning management system so that students did not have to take the extra step of searching for the Web sites. As such, the minimal visual components present in both are assumed to have negligible effects on the overall results.

7 This article displays results for key characteristics. Full results and sample materials are available on authors’ Web sites or upon request.

8 If students from the treatment groups interact with students from the control groups outside of class, then there is potential for bias in the results. If the students in the treatment group obtained and read the newspaper articles from the control groups or if the students in the control group obtained and listened to podcasts from the treatment group, then our treatment variable is contaminated. The likelihood of this happening is small and considered negligible in the research. There is also potential for bias if students from one group teach students from the other group what they have understood better given their different media assignments. This interaction was minimized by never having joint exam review sessions or office hours.

9 Survey questions are available upon request.

10 A class fixed effect was not included. A robustness check was completed to verify that the class fixed effect was not necessary. This was expected, given that the course sections used were held at similar times with similar class sizes and populations. The types of students who would register for one class in the experiment would likely be the same type to register for any of the classes in the experiment.

11 Future research on broad economic content applications using the pre-TUCE and post-TUCE scores could be completed using the method examined by Walstad and Wagner (2016), and Smith (2018, 2022).

12 Researchers used Huber-White robust standard errors in modeling the panel dataset.