From my regression, I found that if women were paid as much as men across the board, the real GDP of the US would increase about eight trillion dollars. WOW! My results are biased but I got really strong p and t values.
Enjoyed seeing some good presentations again today. Unfortunately I missed some because I had to speak with my professor after a class right before this, but I also heard they were all good. Enjoyed this class all semester and it was good to see all your research in action
This course has given me great insight about the economic research process. A future research paper will be a far less daunting task. Having taken Econometrics prior was definitely an advantage when running my regressions. On the other hand, after taking econometrics, I did not fully understand the process of writing a research paper. I really only understood how to analyze data. This class taught me how to make sense of my data which is important when attempting to promote policy or further research.
As a senior economics major, before I took this class I had little experience in doing original economic research, let alone writing a research paper about and presenting it. The majority of the work I’ve done and the classes I’ve taken so far in the economics department have been mainly focused on the theoretical and mathematical side of econ, with only a few short research papers dispersed therein. Before I took this class I didn’t feel comfortable doing my own research, because frankly I didn’t know where to begin. The step by step approach we took to writing and presenting the papers proved immensely helpful in confronting my uneasiness about the research process, and the skills I’ve learned this semester will undoubtedly prove helpful in the future.
The presentations have been great so far. I’ve been really impressed with everyone’s level of professionalism, whether their hypotheses were proven correct or not. Watching classmates present was a great wrap up to the semester. The diversity of the topics we chose to explore is amazing.
Writing my rough draft was a daunting proposal at first. I had never written a paper that long, or spent as much time on one assignment, let alone an entire semester. It wasn’t until I sat down to write my rough draft that I realized just how valuable the research assignments we had completed up to that point were. Putting each part of the paper together would have been exponentially more difficult had I not spent time doing research abstracts, getting feed back on my theoretical model, and putting together data.
Chapter 12 was extremely helpful in structuring my paper. Reading about how to logically lead each section of the paper, from the introduction to the theoretical model to the empirical analysis to the conclusion, into the next really helped me with the writing process. At the beginning of this class we learned how “writing can be used as a research tool” At first I was confused about this statement, but after 21 pages of economic research and analysis I can now fully grasp the concept. As my ideas developed from the introduction to the conclusion I found myself more clearly understanding and explaining the concepts, theories and ideas contained within my research.
Looking over classmates papers, and giving them feedback was a really helpful and interesting part of the research process. It’s like the old saying- “the best way to learn something is to teach it”, offering advice on how to format and improve people’s papers ultimately gave me a good idea of how to structure my own, and what pitfalls to avoid.
I can now finally say that my final paper is done and ready to be analyzed. I think the most difficult thing to do with this was to relate how I did not get the results I wanted and how I can improve. I know I messed up my theory and at least I know how I could possibly continue my research in the future.
When I first put the data for my variables into E-views and ran an ordinary least squares regression, the results were interesting. GDP per capita proved to be an irrelevant variable in explaining the growth rate of the Gini coefficient between 1995 and 2013, so I chose to discard it from my estimating equation going forward. Median income would have possibly produced more significant results. Second, the observed algebraic sign of the explanatory variable unemployment rate was negative, the opposite of what I predicted in my theoretical equation. I was surprised to see that unemployment rates didn’t increase as inequality increased. The only variables with the expected p levels and algebraic signs in the first regression I ran were % of population that was African American and % of population that was hispanic. The p value for lottery expenditure per capita (1995), the most important variable in my regression, was .39, so unfortunately it is not a significant predictor of the growth of income inequality.
I re ran my regression and replaced the variable lottery expenditure per capita (1995) with a dummy variable coded 1 if the state had a lottery and 0 if it didn’t, and used data from all 50 states. While the p value for this variable remained at 0.39, the p values for all my other variables improved, however the value of r-squared decreased, meaning my previous regression had more explanatory power. As professor Greenlaw noted in the textbook for the class “a failed hypothesis does not mean a failed research project” Based on the results of my regression my hypothesis was not proven to be true, however, this opens the door for further analysis and research.
Aside from actually writing my research paper, finding and compiling data was the most laborious and difficult part of the research process. I began putting my data together by looking at the sources of data used in my sources, namely the scholarly article “The Lottery and Income Inequality in the States” I found the sources those authors used to find data on the explanatory variables in their regression equation. I found data for state population estimates, lottery expenditures by state, and the percent of state populations that were African-American and Hispanic on the census website. I found data on state unemployment rates, and percent of workers employed in manufacturing at the state level at the Bureau of Labor Statistics website. I found data on state GDP per capita at the Bureau of Economic Analysis website. And lastly I found data online for Gini coefficients over the past 20 years in a study conducted by researchers at Sam Houston State University. Some of the variables in my theoretical estimating equation such as “lottery expenditure per capita” required combining two different data sets, namely total lottery expenditure, and state population estimates. Excel was helpful for this. Some of my data, such as % of population that is African American and Hispanic was only available in pdf format, requiring me to enter the data by hand- definitely a labor of love. After compiling my data, and manipulating some of it to fit the format of my explanatory variables, I ran multiple regressions in E-views to find which equation provided the best fit. In the end my explanatory variables ended up being lottery income per capita (1995), % of workers in manufacturing (1995), % of population that is African America, % of population that is Hispanic, and unemployment rate, and my dependent variable was gini coefficient growth rate- all at the state level.