Mon, Feb 13, 2017, noon:
Daniel Almirall, "Getting SMART about adaptive interventions"
Jennifer Johnson-Hanks (Department of Demography, University of California, Berkeley)
03/11/2013, at noon in room 6050 ISR-Thompson.
Social science relies heavily on aggregation - that is, on moving from individual cases or instances to higher-order categories. This presentation argues that (1) basic theoretical claims are embedded in how we make up the aggregates that we study, (2) important phenomena may look radically different at different levels of aggregation, (3) aggregates may have certain characteristics that apply only at that aggregate level, and cannot be reduced to the characteristics of the constituent parts, and (4) sometimes the apparent characteristics of the aggregate are artifacts of a bad process of aggregation and what look like relationships are not in fact relationships at all. The variation in how relationships at the micro level are reflected at the macro, and vice versa, may be systematic and meaningful. That is, when we know something about the causal processes behind the variables that interest us, we may be able to predict what kinds of macro-micro patterns will emerge as we move across levels of aggregation.