This past semester, the IPD Health Team has been working on a Senegalese aid analysis project designed to encourage other countries to make similar types of information—such as birthrates, pregnancy rates, and aid distribution to various population groups—available to AidData. Using ArcGIS maps, data visualizations, and written analyses, our goal was to demonstrate how data can be used for more effective aid distribution and potentially improve the standard of living within a country. One of the most important findings, however, was that the direction our analysis took was driven by our lack of, rather than abundance of, data. Accordingly, we didn’t merely demonstrate the importance of data, but also the impact that missing data can have on attempts to analyze trends in aid.
Our original plan was to create five one-page documents, each with a different health indicator mapped alongside World Bank or other health aid projects. These indicators included pregnancies, births, water stress, age structures, and ethnic group distribution; however, inaccessibility to age structure and water stress data limited our analyses in these areas. Most of the data that we used, provided by AidData, was easily displayed on maps and thorough enough for statistical and spatial analysis. The health aid projects, for example, each included at least ten pieces of information including project start date, donor, location, and total funding committed. Some of the data that we pulled from other sources—such as WorldPop’s pregnancy and birth data—integrated seamlessly with the information provided by AidData. Other data, however—such as the water stress data from RiverThreat.net—made it much more difficult to create detailed maps.
Our original plan was to create five one-page documents, each with a different health indicator mapped alongside World Bank or other health aid projects. These indicators included pregnancies, births, water stress, age structures, and ethnic group distribution; however, inaccessibility to age structure and water stress data limited our analyses in these areas. Most of the data that we used, provided by AidData, was easily displayed on maps and thorough enough for statistical and spatial analysis. The health aid projects, for example, each included at least ten pieces of information including project start date, donor, location, and total funding committed. Some of the data that we pulled from other sources—such as WorldPop’s pregnancy and birth data—integrated seamlessly with the information provided by AidData. Other data, however—such as the water stress data from RiverThreat.net—made it much more difficult to create detailed maps.
Figure 1: Water Stressed Areas of Senegal, source: RiverThreat.org
The only water stress data available to us was a color-coded map of the entire world. As you can see, a zoomed-in view of Senegal on this map provides low-resolution, unitless data with little value for analysis. Similarly, WorldPop claimed to provide age structure data for Senegal, but in its place we only found a map containing population density information. The lack of data for these two indicators prevented us from creating two of our five desired maps and left many of our questions unanswered. These gaps in data also reduced the impact and power of the maps that we were able to create, leaving us with maps and analyses depicting fairly intuitive patterns.
FIGURE 2: Aid Distribution and Pregnancies in Senegal, Source: IPD Global Health Team
For example, one of the more obvious takeaways from our pregnancy distribution analysis was that pregnancies are concentrated around Dakar, the capital of the country. One could reach the same conclusion by doing a quick search of Senegal on Google or Wikipedia, where the area surrounding Dakar is shown to be the most densely populated region in the country. While data from the rest of the country provided valuable insight on the more rural regions, where population information is not as widely available, we may have been able to provide more illuminating analyses if the water stress or age structure information had been accessible. Some examples of questions that we were unable to answer due to data limitations include:
We created three one-page documents using the information that was available to us, but we knew that more complete, though currently unavailable, data would have enhanced our ability to answer more interesting questions.
Access to every possible statistic about a country would obviously be ideal, but this is an impractical wish. So how do we deal with this lack of data? Do we submit to it, or do we stand up to it? For this project, we submitted; we did our very best with the data that was available to us. But it doesn’t always have to be this way! Promoting accessibility and transparency of foreign aid data is a primary goal of IPD, and the necessity of such improvements grew increasingly apparent throughout our project.
By the end of this analysis, I had developed a newfound understanding of the types of problems that any group working with development data must deal with. When is the available data adequate? When do we search for more available data? When do we collect new data? Each research project demands a different answer. However, increased accessibility and transparency of data would address this shortcoming in every case.
Jonathan Walsh is an undergraduate studying biomedical engineering at the University of Texas at Austin.
- Which areas in Senegal have both high numbers of pregnancies and poor access to water?
- In which regions of Senegal are there a large amount of health projects relative to the amount of people and access to water?
- Which densely populated regions have the highest percentage of young children and elder citizens (two of the populations most susceptible to disease and illness)?
We created three one-page documents using the information that was available to us, but we knew that more complete, though currently unavailable, data would have enhanced our ability to answer more interesting questions.
Access to every possible statistic about a country would obviously be ideal, but this is an impractical wish. So how do we deal with this lack of data? Do we submit to it, or do we stand up to it? For this project, we submitted; we did our very best with the data that was available to us. But it doesn’t always have to be this way! Promoting accessibility and transparency of foreign aid data is a primary goal of IPD, and the necessity of such improvements grew increasingly apparent throughout our project.
By the end of this analysis, I had developed a newfound understanding of the types of problems that any group working with development data must deal with. When is the available data adequate? When do we search for more available data? When do we collect new data? Each research project demands a different answer. However, increased accessibility and transparency of data would address this shortcoming in every case.
Jonathan Walsh is an undergraduate studying biomedical engineering at the University of Texas at Austin.
RSS Feed