Analyzing the Yourtopia Dataset
The following post is from Dirk Heine and Guo Xu, members of the Yourtopia project team.
Last year, the Open Economics Working Group submitted Yourtopia, a crowd-sourced indicator of social progress, to the World Bank Apps4Development competition and has been awarded the third prize. Yourtopia allows users to assign weights on different dimensions of development (e.g. economy, health and education). Based on the weights submitted by all users, we constructed a robust aggregate weighting, reflecting a global “consensus weighting”, which can used as a consensus measure of development. One year later and after more than 4,000 submitted weightings, where do we stand? And perhaps most importantly, how does our “consensus weight” compare to conventional indices, such as the Human Development Index (HDI)?
The results are quite remarkable: Compared to the default weights of the HDI where economy, health and education receive equal weights (33% each), our consensus weight assigns 30% to economy, 34% to health and 36% to education. Two things are worthwhile pointing out:
1) The HDI weights, even though ad-hoc and arbitrary, are nearly identical to the weights obtained by crowd-sourcing. Taking measurement errors into account, we cannot reject that our consensus weights are equal to the HDI weights. Despite the criticism, the HDI appears to be quite robust.
2) Looking at the point estimates only, the consensus weights also suggest that education is the most important dimension of development, followed by health. This is not surprising as human capital plays a crucial role in fostering economic growth. The economy is merely a means towards expanding capabilities.
Finally, we were also able to explore cross-country variation: By matching the IP addresses of the users against their country of residence, we were able to merge individual weights to country-level means. Correlating the country-level averages against other socio-economic variables enables us to address interesting questions: For example, are weightings associated with country-level variables? Are people from richer countries more likely to assign higher weights to economy, or vice versa?
The figure below plots a country’s GDP per capita level against the average country-level consensus weight for “economy”. A high value for the weight indicates that more importance is given to the economy as an indicator of development. The plot suggests a significant negative relationship: People in rich countries tend to assign less importance to the “economy” dimension, while people in poor countries perceive the economy to be more important. If this is indeed the case, we have another reason to re-consider GDP per capita as a measure of social progress.
Of course, the results should be taken with a grain of salt: The submitted weights are obviously subject to selection bias, which can be substantial in developing countries as access to internet is relatively limited. In addition, measurement errors are likely to confound the results as users were allowed to submit several times. While the large sample size can help alleviate some of these concerns, the results should be seen as tentative.
Are you interested in our project?
Help us analyze the Yourtopia dataset! We have released the dataset and are looking forward to more sophisticated analyses! We are also currently working on Yourtopia 2. If you would like to join the project or come along for a hackday, please contact us at economics [at] okfn.org.