22: Dependence in Data

Author

Affiliation

Smit, A. J.

University of the Western Cape

1 Introduction

The chapter on pseudoreplication focused on the failure to have independent experimental units at the design stage. Here, we broaden that concept to discuss statistical dependence that can be inherent in the data structure itself, even with a valid design. Recognizing and modeling this dependence is crucial for valid inference.

2 Key Concepts

• Temporal Autocorrelation: When observations collected closer in time are more similar to each other than those collected further apart.
• Spatial Autocorrelation: When observations collected closer in space are more similar to each other.
• Nested/Hierarchical Data: Data that is naturally grouped, such as students within classrooms, or plots within sites. Observations within the same group are typically not independent.
• The Problem: Failing to account for dependence leads to pseudoreplication at the data analysis stage, resulting in underestimated standard errors and inflated Type I error rates.

This chapter is a placeholder. Content will be developed further.