Poster Paper: The Effect of Equalization Reform on Elite and Disadvantaged Elementary Schools: Evidence from the Text Mining of Social Media

Saturday, November 9, 2019
Plaza Building: Concourse Level, Plaza Exhibits (Sheraton Denver Downtown)

*Names in bold indicate Presenter

Kelun Lu, Johns Hopkins University, Wei Ha, Peking University and Yiping BAI, University of Michigan

In recent years, China implemented the world’s largest equalization reforms in education to diminish inequity between elementary schools. In Beijing, the equalization reform requires elite and disadvantaged elementary schools to share and transfer teachers, students and resources. Its aim is to reduce polarization between elite and disadvantaged schools.

However, this reform also banned all kinds of inter-school standardized tests in order to blur the distinction between elite and disadvantaged schools. That may have helped to minimize gaps between schools, but it created obstacles to evaluate the outcomes of the reform. To date, much research has focused on conducting surveys, using financial input data or the price of housing in a given school district as variables to estimate the impact of reforms. But such approaches all have their respective drawbacks, according to the literature.

In order to evaluate whether Beijing’s equalization reform promotes education equality, this paper has devised a new way to circumvent these drawbacks by performing sentiment analysis. Social media enables parents to exchange opinions and information about schools. So we derived parents’ sentiments regarding schools from comments made in an Internet parents’ forum. We used these sentiments as an indicator of school quality, since parents are more likely to express their genuine feelings in the unrestricted context of an Internet forum. To start with, we crawled 360,000 comments from the forum. We used a pre-existing sentiment lexicon to assign positive and negative words from the comments and calculate each comment’s sentiment score. We then applied TF-IDF to filter out noise caused by some words that improperly appeared with sentiment scores. We used event study model to estimate parents’ emotions towards schools’ teaching and students’ quality before and after the equalization reforms began.

Results suggest that the equalization reform failed to minimize the gaps, according to parents’ opinions. More specifically, the inferior schools are under heated debate right after the implementation of the reform and the emotions in these comments are experiencing a sharp jump. However, the parents’ emotions decrease in the several quarters after the reform to the level similar to the status prior to the reform. Conversely, the elite schools seem not to be affected by the reform. Although more people also pay closer attention to those elite schools involved in the reform, the superior schools’ sentiment scores are still stable during the time period of 2010 to 2017. Therefore, the equalization reform raises parents’ hope to the inferior schools in a short time period, but the parents discovered the reality that the schools were not enhanced after the reform which weakened parents confidence to the inferior schools afterward. Along with the stability of superior schools’ sentiment scores, the reform didn’t diminish the gap between the elite and inferior schools.

The paper combines the methods of data mining and econometrics, providing researchers a new angle and a data source to evaluate the impact of the policies. These findings thoroughly reveal the outcomes of the equalization reform doesn't diminish the gap between elite and inferior schools.