The Effect of Spatial and Temporal Heterogeneity on the Design and Analysis of Empirical Studies of Scale-Dependent Systems

DOI: 10.1086/510925

 

For convenience references to scale in this paper are referred here as levels.

This paper presents a review and synthesis of issues associated with empirical study designs that attempt to make inferences across multiple levels. The general idea is that empirical field studies need to be better designed in order to effectively inform parameters used in multi-level statistical and theoretical models. While this reviewed literature is related to marine studies, authors suggest that considerations could be applied to non-marine systems.

The crux of the problem discussed in this paper is that traditional field studies are often conducted at the smallest possible level in order to identify the correct mechanistic process. When conducted properly, the field study has adequately isolated any other potential mechanisms operating at the similar level. This isolation may result in unknown consequences when attempting to scale up to a higher level, especially if mechanisms have any potential interactions.

At the smallest level there are two general types of field studies manipulative and correlative. The main issue with manipulative experiments is that they require many replications through space and time to allow for scaling up to a higher level. While correlative studies must overcome issues with pseudo correlations obfuscating true relationships.
The proposed solution to issues regarding either experimental approach is to design experiments as hybrids of the two systems. The first approach would be to nest lower-level (small scale) manipulative experiments into higher level (broader scale) correlative studies.  The second approach, would be to conduct both types of experiments at the same scale. This second approach would provide insight into rather or not experimental responses can be related to the current level. A third approach calls for a series of studies conducted over an integrated framework. Allowing the build up of information from small scale studies. However, if not conducted properly the third approach could fall short when attempting to connect studies. Authors present a flow chart diagram that assists in determining which of the three approaches would be most suitable to the study objective.