My contribution to the Ongoing Discussion
My purpose for creating this site was to build upon the current discussion of factors that influence academic achievement. I attempted to do this through the development of a regression model that would be able to predict a college student's GPA by utilizing factors that are known to impact achievement. Through my research of the subject, I was able to determine 9 individual independent variables to be utilized in my regression model in an attempt to quantify their effect on GPA, which was being used as a measure of academic success. I then intended on using these results to try to find any potential pattern between the explanatory variables, GPA, and the major they chose. However, after an analysis of my regression model, I came to the conclusion that it was not able to provide statistically significant results.
This is not the worst thing in the world, since my results were meant to be suggestive as opposed to concrete. Despite the fact that my results are not statistically significant, I do think that there are positives that can be taken away from it. I have reason to believe that my results came back the way they did as a result of the sample I drew as opposed to the variables I included or the overall regression itself. First of all, the population from which I drew my sample was the Emory University student population. However, the Emory student population is in no way representative of the overall population of American college students. On the contrary, since Emory is an elite university, the students represented in the sample are all extremely strong students with very high quality educational backgrounds and (for the most part) good study habits. This is definitely not the case for the overall population of college students in America. On top of this, the sample that I drew was not a true random sample, as is required for this sort of analysis, but rather was closer to a convenience sample that exhibited at least one form of bias. Despite my efforts to try to keep the sample as random as possible, I was unable to actually do this. Furthermore, my sample size was a mere 45 students. A sample this small could in no way represent the entire population of college students, especially considering the fact that the sample was drawn with sampling biases from an unrepresentative population.
As I said previously, I have no reason to believe there is any sort of issue with the variables I chose to utilize or the way the regression model was constructed. Each of the variables included was chosen after careful research on variables that affect the achievement gap. Each study that I read provided significant results of the described factor having an impact on academic attainment, which provided the basis for their inclusion. Furthermore, the regression model was constructed in compliance with Econometric theory and fulfilled the requirements for an OLS estimator. Because of these things, I feel that someone with the time and resources to gather a unbiased sample from a more representative population would fare better, as their data would be much more likely to provide statistically significant results. These results would be capable of predicting GPA (or another pre-determined measure of academic success), but would not necessarily be able to determine a relationship with the college majors that students chose. However, it may be possible for the researcher to determine such a relationship. Furthermore, this exact regression model can be used to test the effect of these factors on economic success, as measured by things such as wages, if the data is collected from adults (perhaps in their 40s) as opposed to college students. Overall, I feel that my research and the development of this regression model was successful, and I hope that my work has furthered the extremely important discussion surrounding the achievement gap in America.
This is not the worst thing in the world, since my results were meant to be suggestive as opposed to concrete. Despite the fact that my results are not statistically significant, I do think that there are positives that can be taken away from it. I have reason to believe that my results came back the way they did as a result of the sample I drew as opposed to the variables I included or the overall regression itself. First of all, the population from which I drew my sample was the Emory University student population. However, the Emory student population is in no way representative of the overall population of American college students. On the contrary, since Emory is an elite university, the students represented in the sample are all extremely strong students with very high quality educational backgrounds and (for the most part) good study habits. This is definitely not the case for the overall population of college students in America. On top of this, the sample that I drew was not a true random sample, as is required for this sort of analysis, but rather was closer to a convenience sample that exhibited at least one form of bias. Despite my efforts to try to keep the sample as random as possible, I was unable to actually do this. Furthermore, my sample size was a mere 45 students. A sample this small could in no way represent the entire population of college students, especially considering the fact that the sample was drawn with sampling biases from an unrepresentative population.
As I said previously, I have no reason to believe there is any sort of issue with the variables I chose to utilize or the way the regression model was constructed. Each of the variables included was chosen after careful research on variables that affect the achievement gap. Each study that I read provided significant results of the described factor having an impact on academic attainment, which provided the basis for their inclusion. Furthermore, the regression model was constructed in compliance with Econometric theory and fulfilled the requirements for an OLS estimator. Because of these things, I feel that someone with the time and resources to gather a unbiased sample from a more representative population would fare better, as their data would be much more likely to provide statistically significant results. These results would be capable of predicting GPA (or another pre-determined measure of academic success), but would not necessarily be able to determine a relationship with the college majors that students chose. However, it may be possible for the researcher to determine such a relationship. Furthermore, this exact regression model can be used to test the effect of these factors on economic success, as measured by things such as wages, if the data is collected from adults (perhaps in their 40s) as opposed to college students. Overall, I feel that my research and the development of this regression model was successful, and I hope that my work has furthered the extremely important discussion surrounding the achievement gap in America.