What do economic education scholars study? Insights from machine learning

Abstract

The authors of this article use text mining techniques to uncover hidden or latent topics in economic education. The common use of JEL codes only identifies the academic setting for each paper but does not identify the underlying economic concept the paper addresses. An unsupervised machine learning algorithm called Latent Dirichlet Allocation is utilized to identify 15 hidden topics in economic education scholarly work. The text mining model identifies economic education topics by finding correlations in word usage across different documents. The authors show that these newly identified research topics explain more variation in citation counts than the commonly adopted JEL codes. Moreover, specific journals display preferences for certain topics within economic education research.

Keywords:

• Economic education
• Latent Dirichlet Allocation
• machine learning

JEL codes:

• A20
• C38
• C45
• C81

Notes

Notes

1 Authors’ calculations from a sample of 2,559 articles.

2 The American Economic Association assigned the JEL code G53 to financial literacy on May 5, 2019 (AEA n.d., online).

3 Miller and VanFossen (2008) ignored personal finance and entrepreneurship education as topics because their focus of precollege economic education is through the lens of social studies education.

4 Bholat et al. (2015) provide a literature review of the use and application of textual analysis when studying central banks.

5 The American Economic Association extends the following guidelines for the A2 classification, “covers studies about issues related to economics education and teaching of economics at all levels. Studies about economic issues related to education should be classified under the appropriate categories in I2, not here” (AEA n.d., online).

6 These same journals are identified by Lo et al. (2015) and Imazeki (2017).

7 Journal of Economics and Economic Education Research began in 2000; Journal of Economics and Finance Education began in 2002; International Review of Economics Education began in 2003; Perspectives on Economic Education Research began in 2005.

8 The tf-idf weight has a weakness when a single word is used frequently in a single document. For example, the word “chicken” is used by Zongker (2006) over 80 times in the abstract alone. We avoid this complication by removing any terms that appear is less than 1 percent of all documents.

9 More technical details of the estimation routine as well as how the number of topics are selected can be found in the Online Appendix. We use a density-based method as suggested by Cao et al (2009) to find the optimal number of latent topics.

10 Robust standard errors clustered by journal are estimated to capture unobserved heterogeneity and mitigate issues with over-dispersion (Wooldridge 1999).