My experiences in IS-301 were on the whole good, though not entirely without mistakes. At times I struggled with time management and proper planning, but in the end I grasped the concepts and successfully applied them to the analysis of our data for our final project.

I think I did a number of things poorly, including recording the information I learned from the textbook, leaving some gaps in my blog posts, and planning the final paper with my group. When doing the worksheets, I developed a bad habit of learning what I needed to do for the worksheet, and then not coming back to the material. I would have done better to keep a notebook in which I recorded the steps necessary to perform different types of analyses, along with what types of variables they were designed to analyze. Furthermore, I think I ought to have worked along with the professor in class, instead of working on other projects for other classes as I did so often.

I began the semester with the intent of writing a blog post every week, and I stuck to this for the first month or so. However, as I began to get bogged down in other responsibilities, I let my regularity slip. I think I remember creating a reminder in my e-mail calendar to prompt me to post weekly, but I’m not quite sure if I did in fact do so, or what became of it if I did. I ought to have created more reminders, possibly physical ones and contented myself with less-than-perfect posts for the sake of posting something.

Finally, I think I ought to have established a firmer timeline with my group. We did make efforts to do this, and had planning sessions for that purpose, but we were all so involved with other responsibilities that we ended up leaving a lot of tasks vaguely defined and treating the deadlines that we set for ourselves extremely loosely. I think that we ought to have laid out a long-term schedule for completing the project on a semester level, and then met in person every so often to discuss our findings and assess our progress.

On the other hand, I think I did some things well, such as getting the big concepts down at the beginning of the semester, writing quality blog posts, and conducting the analysis for the final project. I scored very well on the concept-based quizzes at the beginning of the semester, simply because I took a little bit of time to read the material and made a few flashcards. Those quizzes were quite easy for those who had read the material but mystifying to those who had not; I think I exercised good time management in devoting just enough time to studying for them to do well, but not long enough that it detracted from other projects.

My blog posts were generally on the longer side, and I made a conscious effort to relate my content to the course material and the actual practice of computer analysis. For instance, when I was discovering the implications of a particular method of analysis, or when my group discovered an error in our data, I found it a fruitful topic on which to reflect. Thus, although I may not have posted as often as I should have, my posts were generally of good quality.

I think perhaps my greatest success came when I performed all of the analysis for my group. I was able to apply the correct type of analysis to the various types of data that we had collected. For instance, I used a bivariate correlation to analyze the relationship between Chinese FDI and exports (two interval variables) and used two independent samples t-tests to compare the mean FDI of countries that recognized Taiwan with those that didn’t, and the mean FDI of countries that exported petroleum to China with those that didn’t (an interval dependent variable and two nominal independent variables).

In sum, I ought to have organized my study of the statistical methods more systematically, and also planned both my own blog postings and my group’s work more carefully. However, my self-reflections were of good quality and I successfully grasped and applied the concepts being taught. At the end of the day, I firmly believe that I am leaving the class with a much firmer grasp on how to conduct quantitative analyses.

My group learned an important lesson about how easy it is to make mistakes when dealing with large amounts of data. We wanted to examine the effects of FDI on Chinese exports to African countries out to five years in the future. So, at Col. Sanborn’s suggestion, we simply shifted the Export column down one, thus lining up Export values with FDI values from the previous year (or so we thought). Then, we shifted it down two rows for two years in the future, three for three years, and so on. We were excited to discover that, when we ran bivariate correlations using this shifted data, the Pearson’s coefficient increased from .210 the year the FDI was received to .365 five years after the year of receipt. However, our excitement was short-lived, as we discovered that we had ignored the fact that our data came from multiple countries. Simply shifting the Export column down worked within a country’s data, but once it crossed over the line into another country’s data it became misaligned. So, for instance, SPSS was running correlations on the relationship between Chinese FDI to Angola in 2015 and Chinese exports to Benin in 2003, which didn’t provide any insight into whether or not there was a causal link.

The good news is, once we fixed the problem and only compared FDI and Exports for the same country, we reran the program and found that there was an even stronger correlation with the correct data than with the flawed data. The Pearson’s Coefficient decreased slightly from 0.210 the year of receipt to .183 one year after, but then increased steadily after that to .250 three years after, .384 four years after, and finally .423 five years after. This data shows that, not only is Chinese FDI a good predictor of Chinese exports to a given African country the same year it is received, but it becomes a better predictor as time passes. Indeed, FDI is twice as good at predicting Exports five years after it is received than in the original year of receipt. This supports our hypothesis that Chinese FDI to African countries has a positive impact on Chinese exports to those countries, and thus further suggests that Chinese FDI is indeed serving China’s strategic interests in Africa. The fact that both we and Col. Sanborn failed to realize the error demonstrates just how easy it is to make mistakes when manipulating large amounts of data, and how those mistakes can effect research results.

We have discovered that the student version of SPSS won’t let us analyze the volume of data that we have. So, we are relying on Col. Sanborn to help us use department computers to analyze our data with the full version of SPSS. We have collected data for 50 African countries over a 13 year period (2003-2015), producing 650 total data points for each area of analysis. We have struggled to find good sources for much of our data. Data on Chinese FDI to African countries was extremely difficult to locate, until we finally stumbled upon a data set that had been compiled by the Johns Hopkins University School of Advanced International Studies’s China-Africa Research Initiative. Finding data on yearly natural resource exports to China from African countries was also hard, as we were able to find data organized by country of origin and by year, but not by both simultaneously. We eventually had to go through and manually copy and paste the results from each year one at a time, which was a tedious and time-consuming process. Having professional sources of macro-data on which to draw would have made our project much easier.