Graphical Methods for Exploratory Analysis of Complex Data Sets

• Published in: Bioscience Vol. 57; no. 8; pp. 673 - 679
• Main Authors: YEAGER, RONNIE L, PARKHURST, DAVID F, HENSHEL, DIANE S
• Format: Journal Article
• Language: English
• Published: Circulation, AIBS, 1313 Dolley Madison Blvd., Suite 402, McLean, VA 22101. USA American Institute of Biological Sciences 01.09.2007; Oxford University Press
• Subjects: 2,3,7,8-tetrachlorodibenzo-p-dioxin; Bar graphs; Biochemistry; BIOLOGIST'S TOOLBOX; Body Weight; Brain; Brain research; central nervous system diseases; Cerebrum; chickens; Computer software; Data analysis; Datasets; DEPARTMENTS; developmental toxicology; Dioxin; Dioxins; Dosage and administration; enzyme activity; Enzymes; exploratory data analysis; Graphic methods; Graphs; Health aspects; Histograms; Hypotheses; light effects; Light emitting diodes; mathematical models; metabolism; Poultry; Software; Statistics; toxic substances; Toxicology; trellis graphics; United States; developmental toxicology; trellis graphics; exploratory data analysis; light effects; 2,3,7,8-tetrachlorodibenzo-p-dioxin
• ISSN: 0006-3568; 1525-3244
• DOI: 10.1641/B570807

In scientific investigations, performing initial exploratory studies can help in generating hypotheses and deciding what to study in more detail later. Little-known modern graphical methods can be powerful tools for exploration of complex data. We illustrate several such types of graphs and discuss what we have learned from them. The data come from a study designed to quantify physical and enzymatic changes in both sides of chicken brains after developmental exposure to dioxin, with half the eggs treated with red light. The data corresponded to 160 combinations of dose, light presence or absence, brain side, and brain dimension or biochemical activity, making the graphical methods especially helpful for assessing the presence and extent of asymmetry in the responses. We provide graphical examples using a pairs plot and several trellis plots in which the data appeared consistent or inconsistent with prior expectations and suggested unexpected findings that will motivate new research.