A large body of recent quantitative work on the 'diversity detriment' hypothesis finds that ethno-religious diversity is linked with a host of societal ills, implying in turn a strong challenge to multiculturalist theory and policies. Given the stakes, the appropriate conceptualization and measurement of ethno-religious divisions is a matter of considerable importance. This paper provides a synthetic critique of how major measures each capture the notion of 'divisions' and relate to each other conceptually and empirically within a divided society. Furthermore, instead of presenting temporal snapshots of divisions at the national level, as in most previous work on the topic, we explore how measures perform at more localized levels of analysis and over time, drawing on individual level census data from Mindanao, the Philippines. We highlight four conceptually 'big' issues we believe deserve emphasis and further investigation: the sensitivity of measures to the choice of construct, categorization methodology, passage of time, and spatial variation. We provide guidance and discuss the key implications of these points both for quantitative scholars working with these measures and for qualitatively inclined empiricists and normative theorists wishing to interpret, evaluate or otherwise engage with the quantitative research on the merits of diversity.