As we have seen in the last chapter, in many studies of schooling that examine processes and outcomes, the outcome and process variables are only measured on one occasion (i.e., using cross-sectional data). A limitation of static, cross-sectional data, however, is that they are not well suited to investigations of processes that are assumed to be dynamic. This is because it is difficult to establish the proper time ordering necessary to address causal relationships in cross-sectional analyses. It therefore becomes more difficult to rule out alternative explanations. Analyses based on cross-sectional data may only lead to a partial understanding of processes at best, and misleading interpretations at worst (Davies & Dale, 1994). As suggested in the chapter on qualitative methods (chap. 9), time is a key factor in understanding how policy processes unfold, as well as how their impacts may be observed. Limitations of data and method have in the past restricted the quantitative analysis of policy processes. Increasingly, however, both the concepts and methods are becoming available that can provide a more rigorous and thorough examination of longitudinal data. Although there has been considerable development of longitudinal data analysis techniques for use with experimental, quasi-experimental, and nonexperimental research designs, there is still a need to make the techniques more accessible to those interested in conducting policy research.