Is the World Becoming a Better or a Worse Place? A Data-Driven Analysis

Abstract

Is the World becoming a better or a worse place to live? In this paper, we propose a tool that can help to answer the question by combining a number of global indicators belonging to multiple categories. The proliferation of statistical data about various aspects of the World performance may suggest that it should be "easy" to evaluate the overall success of human enterprise on this planet. Moreover, it also points out the intrinsic importance in the selection of indicators. However, people have different values, biases, and preferences about the importance of various indicators, making it almost impossible to find an objective answer to this question. To address the variety and the heterogeneity of available indicators and world views, we present the analysis of global World performance as a multi-criteria decision problem, making sure that the assessment method remains as transparent as possible. By dynamically selecting a set of indicators of interest, defining the weights that we attach to various indicators and specifying the desired trends associated with each indicator, we make the assessment adaptive to individual values. We also try to deal with the inherent bias that may exist in the set of indicators that are chosen. As a study case, from various data sets that are openly available online, we have selected several that are most relevant and easy to interpret in the context of the question in the title of the paper. We demonstrate how the choice of personal preferences, or weights, can strongly change the result. Our method also provides analysis of the weights space, showing how results for particular value sets compare to the average and extreme (optimistic and pessimistic) combinations of weights that may be chosen by users.

Life before COVID-19: how was the World actually performing?

“…The target system is modelled by selecting a number of categorised indicators, which are global indicators in this study. The model also assumes weights and semantics associated with indicators and it’s the input for the computational method (Pileggi 2020 ). Interpretations are based on both qualitative and quantitative metrics.…”

“…Criteria selection: macro-categories and representative indicators The normal approach (adopted also in previous work (Pileggi 2020 ) as well as by many reputable studies and publications, such as Our World in Data ) is to group the different indicators in classes which represent, therefore, an abstracted categorization of the considered indicators. It is very useful, especially considering the great availability of data, dependencies and the need to consider multiple aspects together.…”

“…1 ), given M categories and N indicators, we are assuming and cardinality 1:1. Such a simplification allows an easier weighting and modelling within the method adopted (Pileggi 2020 ). We are assuming the definition of categories and the selection of representative indicators as an intrinsic bias, which is referred to as selection bias .…”

“…For instance, they may reflect the opinions within a given group or community, normally elicited by surveys or interviews. Alternatively, weights may be inferred by capturing input parameters by the users of tools that adopt the method (Pileggi 2020 ). Either ways, to be relevant, the weighting should be based on a significant number of samples.…”

“…Previous work and background This paper is based on the method proposed in Pileggi ( 2020 ) which adopts a Multi-Criteria Decision Analysis (MCDA) philosophy (Ishizaka and Nemery 2013 ; Velasquez and Hester 2013 ). That paper focuses on the method in itself, which is explained in detail and applied to a number of examples using real data.…”