ERW mentions in a couple places that users of quantitative analysis have a range of conventions in writing about their methods and models. Here’s a quick summary of key things beyond what ERW contains.
Subscripts
Subscripts are used to differentiate variants of the same variable. They are almost always written in lower-case letters unless they are referring to a specific abbreviation, such as when we have two measures of a variable.
| Subscript | Interpretation |
| i | Index – used to identifies observations (case numbers). |
| j | A second case indexer. Often used in dyadic analysis, where two or more units are being studied as pairs. |
| ij and ji | Used to denote directed dyads. For example, trade flows might be tradeij and tradeji – trade from I to J and then trade from J to I, respectively. |
| t | Time – used to index time periods. |
| t -1 | Time at period t-1 (i.e., the period before the period in which the rest of the variables were measured). For example, we might want to use trade from the year before to predict trade this year, so we would include “Tradet-1” in the model. |
Font and Character Styles
ERW typically refers to variables X and Y with capital letters in plain font since I am referring to what are technically speaking random variables. In many more sophisticated methods books, especially econometrics ones, the notation is more complex. Both case and font style convey information about the structure of the quantity expressed.
| Font Style Case | Lower case | Upper case |
| Plain | Scalar | Random variable |
| Bold | Vector | Matrix |
Scalars almost always require subscripts, as do some vectors. (You can see why I opted for a simplified notation.)
Population parameters are expressed as lower-case Greek letters, and sample statistics are lower-case Roman letters.
In the context of statistical formulas, N regularly refers to population size and n to sample size; in the context of results tables, we typically default to capital N.
Variable names are written in the text in title case italics.