Quantifying uncertainty in segregation measurement
In the vast literature on segregation measurement, only a handful of papers attempt to address the issue of inference – computing confidence intervals for example for the index of dissimilarity and related measures. Quantifying uncertainty is important if we are to address basic questions about whether there have been genuine changes in segregation over time or between cities. For example, has there been a statistically significant fall in religious segregation in Northern Ireland, or is the apparent decline due to random variation? Is there a statistically significant different between segregation in Glasgow compared with Edinburgh, or is the difference due to white noise?
Image: Flickr – Danny Fowler