Get access

Causal diagrams and change variables

Authors

  • Eyal Shahar MD MPH,

    Corresponding author
    1. Professor, Division of Epidemiology and Biostatistics, Mel and Enid Zuckerman College of Public Health, University of Arizona, Tucson, AZ, USA
    Search for more papers by this author
  • Doron J. Shahar

    1. Undergraduate student, Departments of Physics and Mathematics, College of Science, University of Arizona, Tucson, AZ, USA
    Search for more papers by this author

Dr Eyal Shahar, Division of Epidemiology and Biostatistics, Mel and Enid Zuckerman College of Public Health, The University of Arizona, 1295 N. Martin Avenue, Tucson, AZ, USA, E-mail: shahar@email.arizona.edu

Abstract

Background  The true change in the value of a variable between two time points is often assumed to be a cause or an effect of interest. To our knowledge, this assumption is based on intuition, rather than on any formal theoretical justification.

Methods  We used causal directed acyclic graphs to explore the causal properties of a change variable, and critically examined competing structures.

Results  Based on the proposed causal structure, a change variable (true change) is no more than a derived variable. It does not cause anything and is not of causal interest.

Conclusions  A true change is not a variable in the physical world. Therefore, modelling the change between two time points is justified only in a few situations.

Ancillary