Chapter

2 Exploratory Data Analysis

Research Methods in Psychology

I. FOUNDATIONS OF RESEARCH ISSUES

  1. John T. Behrens PhD1,
  2. Kristen E. DiCerbo PhD2,
  3. Nedim Yel3,
  4. Roy Levy PhD4

Published Online: 26 SEP 2012

DOI: 10.1002/9781118133880.hop202002

Handbook of Psychology, Second Edition

Handbook of Psychology, Second Edition

How to Cite

Behrens, J. T., DiCerbo, K. E., Yel, N. and Levy, R. 2012. Exploratory Data Analysis. Handbook of Psychology, Second Edition. 2:I:2.

Author Information

  1. 1

    Cisco, Networking Academy & Corporate Affairs, Mishawaka, Indiana, USA

  2. 2

    Avondale, Arizona, USA

  3. 3

    Arizona State University, Division of Educational Leadership and Innovation, Tempe, Arizona, USA

  4. 4

    Arizona State University, School of Social and Family Dynamics, Tempe, Arizona, USA

Publication History

  1. Published Online: 26 SEP 2012

Abstract

Exploratory Data Analysis (EDA) is a quantitative data analytic tradition based on the original work of John Tukey. EDA provides a framework for a broad range of data analytic activity and addressing the broad range of forms of data and design that applied researchers face. The core conceptual and computational tools of EDA include the use of graphics and interactive data display (Revelation), an emphasis on model building, diagnosis and evaluation (Residuals), addressing the fundamental measurement issues associated with various distributions (Re-expression) and undertaking procedures that are resistant to misleading or erroneous results because of the vagaries of real-world data (Resistance). While these tools provide a grounding for all analysis, EDA emphasizes data-driven learning from the data to compliment cook-book application of hypothesis testing procedures that can overlook important unanticipated aspects of data and their impact of modeling and estimation. It is argued that the EDA is essential both in early stages of research where hypothesis and model formulation need to be well informed, as well as in later stages where model misfit and blatant “missing the picture” can occur. The slow integration of this approach into standard software and practice is discussed.

Keywords:

  • exploratory data analysis;
  • graphical display;
  • visualization