PURM 10 Week Research Internship: CHIBE
My summer research experience at the Center for Health Incentives and Behavioral Economics (CHIBE) has not been focused only on one specific study, but rather multiple research projects under the guidance of Professor Harsha Thirumurthy. One project that I focused on was based around estimating lost GDP due to infectious and parasitic diseases in various regions around the world. When somebody gets sick and dies prematurely, (or if the disease forces them to leave the labor force) their potential labor productivity decreases which translates into lost GDP. By compiling a large data set of various disease and GDP statistics from each country and performing some calculations using macroeconomic concepts, such as purchasing power parity, I was able to actually calculate this lost GDP. I analyzed the data by creating graphics that broke down the effect of these diseases by country income level and region. What I ended up finding was that, surprisingly, it wasn’t the poorest countries that had the highest proportion of lost GDP as one would initially presume, but rather those categorized as middle income.
One other study that I focused on during my ten weeks was the “Jikinge” study, taking place in Kenya. The premise of the study is to analyze the effects of secondary HIV self-test distribution on HIV transmission and sexual behavior. “Secondary” signifies that it is not doctors or medical centers distributing HIV self-test kits, but women of high HIV risk who are given multiple kits to give to their sexual partners. Because this study enrolls over two thousand women, and the data was being collected and electronically stored in Kenya, it was essential to maintain accurate backups of the main data set and scan through for errors and inconsistencies. Professor Thirumurthy’s analyst team introduced me to data quality control, and the massive effort that goes into making sure a large, international study runs smoothly. Throughout the summer I was assigned tasks such as designing technology instruction sheets for the data collection team in Kenya, creating and updating summaries of missing data to pinpoint errors between the main server and backup, as well as making sure data was being logged at their appropriate times throughout the multiple year data collection period. The Jikinge project also gave me my first introduction to Stata, a statistical software package used in economic research. I analyzed scatterplot matrices, performed regressions, and created histograms to try and identify behavioral trends in the large data set to test the hypothesis of fatalism, which predicts that those who believe their risk of contracting HIV is greatest actually partake in riskier behavior.
In my final weeks at CHIBE, I worked with Stata once more in the analyses of a household consumption survey in Africa with the end goal being a calculation of average household spending. Having little coding experience prior to research this summer, I definitely had a large learning curve to overcome but learned a lot about coding in Stata and how coding processes such as loops work. Overall, I learned a lot this summer not just about the technical skills needed to perform research, but also the massive effort that goes into data collection and organization for large research studies.
PURM 10 Week Research Internship: CHIBE
My summer research experience at the Center for Health Incentives and Behavioral Economics (CHIBE) has not been focused only on one specific study, but rather multiple research projects under the guidance of Professor Harsha Thirumurthy. One project that I focused on was based around estimating lost GDP due to infectious and parasitic diseases in various regions around the world. When somebody gets sick and dies prematurely, (or if the disease forces them to leave the labor force) their potential labor productivity decreases which translates into lost GDP. By compiling a large data set of various disease and GDP statistics from each country and performing some calculations using macroeconomic concepts, such as purchasing power parity, I was able to actually calculate this lost GDP. I analyzed the data by creating graphics that broke down the effect of these diseases by country income level and region. What I ended up finding was that, surprisingly, it wasn’t the poorest countries that had the highest proportion of lost GDP as one would initially presume, but rather those categorized as middle income.
One other study that I focused on during my ten weeks was the “Jikinge” study, taking place in Kenya. The premise of the study is to analyze the effects of secondary HIV self-test distribution on HIV transmission and sexual behavior. “Secondary” signifies that it is not doctors or medical centers distributing HIV self-test kits, but women of high HIV risk who are given multiple kits to give to their sexual partners. Because this study enrolls over two thousand women, and the data was being collected and electronically stored in Kenya, it was essential to maintain accurate backups of the main data set and scan through for errors and inconsistencies. Professor Thirumurthy’s analyst team introduced me to data quality control, and the massive effort that goes into making sure a large, international study runs smoothly. Throughout the summer I was assigned tasks such as designing technology instruction sheets for the data collection team in Kenya, creating and updating summaries of missing data to pinpoint errors between the main server and backup, as well as making sure data was being logged at their appropriate times throughout the multiple year data collection period. The Jikinge project also gave me my first introduction to Stata, a statistical software package used in economic research. I analyzed scatterplot matrices, performed regressions, and created histograms to try and identify behavioral trends in the large data set to test the hypothesis of fatalism, which predicts that those who believe their risk of contracting HIV is greatest actually partake in riskier behavior.
In my final weeks at CHIBE, I worked with Stata once more in the analyses of a household consumption survey in Africa with the end goal being a calculation of average household spending. Having little coding experience prior to research this summer, I definitely had a large learning curve to overcome but learned a lot about coding in Stata and how coding processes such as loops work. Overall, I learned a lot this summer not just about the technical skills needed to perform research, but also the massive effort that goes into data collection and organization for large research studies.