Exemplary versus statistical evidence? Response to Berg-Schlosser

We appreciate Dirk Berg-Schlosser’s discussion of our article. He makes a number of positive comments and also expresses some doubts, raising important issues that should be clarified. In this limited space, we will not address all of the diverse matters he discusses, but will concentrate on his major points. Berg-Schlosser expresses concerns about our measurements, arguing that the indicators we used to measure a society’s available resources, its emphasis on emancipative values and its degree of effective democracy are based on ‘guesstimates’. This is true in the sense that any measure in the social sciences involves a certain amount of measurement error. Berg-Schlosser’s critique, however, is pointless insofar as he fails to provide evidence for his claim that there are better indicators of a society’s resources, its emphasis on emancipative values and its degree of effective democracy than the ones we used – and we doubt that he could, because any alternative indicators of which we are aware correlate strongly with the ones we used (usually at r = 0.8 or 0.9). For example, Berg-Schlosser criticizes our measure of effective democracy, asserting that Singapore would obtain a medium democracy score using this measure. This conclusion is false. We measure effective democracy by combining two separate components. First, we use the Freedom House scores for liberal and political rights as an indicator of the extension of rights. Then we weight the extension of rights by the elite corruption scores provided by Transparency International, using the inverse of these scores as a measure of ‘elite integrity’ or ‘rule of law’. We consider ‘elite integrity’ as the factor deciding how effective or ineffective a given extension of rights is in light of actual elite behavior. Mathematically, this is expressed by multiplying the extension of rights with the degree of ‘elite integrity’. This procedure prevents high levels in one of the two components compensating for low levels in the other, because multiplication is sensitive to discrepancies. Hence, Berg-Schlosser is wrong in claiming that our measure would locate Singapore at a medium level