Marceau thoroughly introduced the scale issue in geographic, or any researches related to spatial and temporal scale issue. It refers to the difficulty of understanding and using the “correct scales”, as phenomenon of interest may only occur in certain scales, and the study result might be various due to the use of alternative combinations of areal units at similar scales.
To explain scale issues in social science and natural science, Marceau focuses on MAUP (Modifiable Areal Unit Problem) in the realm of social science, and scale and aggregation problems at natural science. As progress has been made for the last few decades, the MAUP problem still remain unsolved (according to Marceau, by 1999) , but studies on how to control and predict its effects were developed to get close to solve the problem. While in natural science, HPDP (Hierarchical patch dynamics paradigm) was provided to solve the scale and aggregation problem in natural science, as a framework of combining bother vertical and horizontal hierarchy problem.
In the end of his introduction of scale issue, Marceau threw out three main steps of solving the scale issue: understanding scale dependence effect; the development of quantitative methods to predict and control the MAUP effects, as well as to explain how entities, patterns, and processes are linked across scales (aggregation problem); and to build a solid unified theoretical framework, which hypothesis can be derived and tested, and generalizations achieved (Marceau, 1999).
The most intriguing part of this article to me is the MAUP problem. Although progress has been made to control and predict its effects on the study result, from some of the recent urban geographic studies I have been read, the MAUP problem is still unsolved. And sometimes, ignored when researchers talked about their sampling process. From Marceau’s explanation, I do realize that it is important to address scale issues, such as MAUP in both social and natural science studies, in order to figure out whether the study result is solid and spatially valid, and avoid unexpected spatial bias.