Regression Results
Bivariate Regression with Rural Interaction Variable
In order to achieve the goals set out by the project, we have to measure the impact of both rural and urban factors on academic scores for math, science, and reading. As such, we will utilize interaction variables in order to measure the difference in school resource quality between both areas. We identified a difference in both the level of significance and substance when converting the education resource quality variable into an interaction term. Before, the resqual variable, on it’s own, held very little significance or substance when individually interacting with the dependent variable of exam scores in math, science, and reading. Afterwards, the interaction of the education quality resources within an urban and rural setting helped to identify a difference in the quality of resources whereby, the poor quality of score resources in rural areas had a negative impact on exam scores as compared with urban areas, where the positive relationship indicates better quality school resources. Multivariate Regression: Rural - Urban in Latin America Results for Latin America overall show that the slope for rural resource quality is greater than that of urban resource quality in all three subject areas. This shows that resource quality has a greater impact on PISA scores in rural areas in comparison to urban areas. Regression results show that the slope is greatest in reading for rural areas and is smallest in science. Improving resource quality is therefore most effective for improving reading scores, and least effective for improving science scores. We believe this is due to the fact that reading does not require as many resources as science. Books are a relatively simple resource to acquire, and can greatly improve a child's reading scores. Science, on the other hand, requires many pieces of equipment. Improving science resources by 1 point will not impact a child's learning as much as improving reading resource by the same amount. Urban results, on the other hand, tell a different story. The more gentle slope of the urban line shows that improvements to resource quality do not have as great of an impact on PISA scores. We believe this is due to the many other factors that improve in urban areas. For example, urban areas often attract more qualified teachers. As many studies show, teacher quality can greatly impact a student's performance. This would thus reduce the impact of resource quality on a student's performance and reduce the beta coefficient, or the slope of the line. Multivariate Regression with Rural Interaction Variable:
Latin America The first regression for each subject is run without the interaction variable. The second regression incorporates the interaction variable of resource quality interacted with rural regions. This beta coefficient shows us the increase in slope that is created by this interaction. Reading shows the greatest slope increase, followed by science and then math. In the first regression, reading was still most affected by changes in resource quality, but was followed by math and then by science. Multivariate Regression with Rural Interaction Variable:
by Country We ran the same multivariate regression with the rural interaction, but separated each country. We notice two outliers in these regression results in regards to the rural interaction term, Argentina and Chile. Details of each country can be found here. |
Regression Tables:
|