Understanding the nature of the variables used in an analysis is essential as different variable types will convey different kinds of information about a given entity. How the analyst or the software used for analysis treats this information will impact the validity of any results and interpretation. It is equally important to Identify what kind of data and/or information has been extracted from an experimental or observational design and to select an appropriate variable type to represent it. A description of variable types can be found in most introductory statistics textbooks. This page provide a superficial description of some variable types common to ecological data (Figure 1) and the practice of converting categorical variables into "dummy variables" (Figure 2).
Figure 1: Variable types often encountered in ecological data sets. a) Binary values (1 or 0) are used to describe dichotomous outcomes such as presence or absence, true or false. b) Integer values, or discrete quantitative values, are used to quantify discontinuous or countable entities. Values reflecting the abundance of whole organisms is an example of this data type. Arithmetical operations (e.g. addition, multiplication, etc) involving quantitative data values will produce meaningful results. For example, a sample with double the abundance of copepods observed in "Sample 1" will have six copepods. c) Non-countable (or not easily countable) but measurable quantities are represented by continuous quantitative variables. As with discrete quantitative data, arithmetical operations on this variable type will produce meaningful results. Temperature, pH, and mass are examples of entities that are best represented on by continuous quantitative variables. d) Variables which reflect a ranking or are ordered (e.g. reflect the assortment of entities into ordered intervals) are semiquantitative. Arithmetical operations involving ordinal data are not meaningful. For example, There is no rank that would represent "half" or "double" second place and there is no rank that represents the sum of first and second place. e) Entities or phenomena with no natural ordering or magnitude are described by categorical or nominal variables. The colours of entities, for example, are not ordered (even though they may correspond to phenomena that may be described by (semi)quantitative variables).
Figure 2: the creation of "dummy variables" from a multistate categorical variable. Herein, each state of a categorical variable is represented by a new binary variable. Objects (here, "Samples") that fall in a category acquire "1" (or "true") values and those outside the category acquire "0" (or "false") values. Depending on the analysis software used, this procedure may not be necessary.