Statistical Comparisons in Time and Space

When thinking about the future, the question of the indicators used to construct analyses and serve as a basis for projections is a key one. Many indicators are imperfect, as can be seen from the recurrent discussions on the creation of wealth, or on the measurement of a country’s national income, which we know to be inadequate for estimating its population’s well-being. But even beyond the nature of the indicators selected, the question of the pertinence of the comparisons to be made from them in both time and space is another focal issue that gets little attention. Yet, as is demonstrated here by Jean Baneth, an economist and former director of the World Bank’s International Economics Department, we must be cautious about this and not make statistics say more than they can. His article explains that the conclusions of such statistical comparisons may vary substantially, depending on the methodology used to calculate price indices for the purpose of making comparisons over time or between countries, on the mode of comparison adopted for the prices of goods or income levels (nominal levels or purchasing power parity), and even on the level of quality of goods compared. In this article Baneth presents a number of subtleties (if not, indeed, biases) to be found in the interpretation of statistical comparisons, so as to provide those who use them with some clues for avoiding the pitfalls attaching to them or, at the very least, for being aware of their limits.