The project repo for Group 7 of CSPB 4502: Data Mining
A Study of Financing versus Student Performance
Jarryd Allison
This project analyzes the academic performance of public school students from 2019 - 2023 in order to assess how performance has changed, what might have reasonably affected performance, and how public funds have been spent in relation to performance. It's intention was to better understand school performance and financing, and derive interesting and novel insights into how public policy might change to better Colorado schools.
This project attempted to answer the following questions:
- Does a correlation exist between the spend per student and the mathematical academic performance of a student body?
- If a correlation exists, is it positive or negative?
- How has funding and performance changed over the time period from 2019 to 2023?
The following answers, after mining the existing data, revealed some surprising results:
- Per the contingency analysis conducted on the existing data set, a correlation, albeit slightly weak, does exist for every grade and year from 2019 to 2023. This was calculated using the x^2 correlation analysis, as well as lift.
- The correlation in every case was negative, though, as the report suggests, the correlation was weak. This implies that more funding is not a guarantor of academic performance and, per the dataset, more funding often is an indicator of decreased performance.
- From the data set, performance saw an enormous dropoff during 2020, which was expected. However, it has improved slightly since the 2020 school year, though has not yet recovered to pre-COVID levels.
- A few surprises revealed themsselves through the data mining process:
- One school reported an average spend of over $400,000 USD during the 2020-2021 school year. These outliers either at best destroy trust in the transparency of the data provided to the taxpayers, or at worst give rise to questions about how this type of spend could possibly be justified.
- While performance is improving, it is at a lower rate at which the budget for education is growing. For example, the Grade 8 performance in 2023 fell by 2%, while the budget increased by nearly 7%.
- It was discovered that Colorado does not require performance reporting if a school can report 16 or fewer valid scores for the academic year. This data is collected at the school level, and on average would represent nearly 9% of the total reported values, but is instead permitted to be omitted. This undoubtedly would have a drastic impact on the reporting, and should be reviewed to improve the understanding of academic performance state-wide.
This data is surprising, seemingly counterintuitive, and above all depressing considering the performance of hundreds of thousands of children in the state education system of Colorado. Despite over $7B (USD) spent in 2023, performance remains low, with no solid plan in site to either appropriate funds more accurately to where it may improve scores, or indeed any cohesive plan in the legislature to address this shocking shortfall in academic performance. This report, then, is applicable in a number of different venues, all within the public sphere.
- Articles and awareness It is the intention of the author to publish these findings in a series of substack articles or twitter articles. The spending represented in the above findings comes directly from state taxpayers, many of whom know or have participants currently in the education system. Increasing public awareness of the performance of public education in the state can result in real change, by prompting legislatures to review expenditures, correct erroneous reporting, conduct investigations, and acting on this data to help improve student outcomes.
- Further analysis This report may spur additional researchers, and even this author, to continue to conduct research, find more information, and better understand the underlying patterns in academic performance. No doubt additional overlaying of census, location, average income, ethnicity, and other such data sources would more richly identify the various underlying variables contributing to the educational decline in the state.
- Legislative action Should the data be collected into a convincing format, it could also drive legislative plans to improve the educational outcomes. The example in Figure 12 is but one instance where a close eye must be kept on how funds are appropriated and spent. No doubt underfunded schools would benefit immensely from a second review of how money is allocated.
- School initiatives PTA organizations and individual schools should review the top performers who are able to spend less per student, but achieve above average performances from their students. Specifically, Figure 5 identifies a massive gap that must be closed to ensure students are both learning and granted more opportunities as an aggregate.