Please follow these instructions carefully when preparing your tables for the production process. Tables are by far the most complex aspect of our typesetting process; by attending to the seemingly anodyne details below in detail, you help make the process run more smoothly, and will reduce your typesetting costs if we are converting your manuscript to LaTeX.
The most important rules to follow below are highlighted in bold.
In general
- When reporting regression results, tables should report both coefficient estimates and estimated standard errors. Standard errors should be placed in a row below the corresponding coefficient estimate, not in a separate column.
- Authors may generally choose the level of precision at which to report numerical estimates, subject to the restriction that no more than three digits be reported after the decimal point.
- Use leading zeros in presenting decimal numbers between -1 and 1, e.g., 0.234 not .234
- To the extent possible, avoid using exponential notation (e.g., for very small numbers). Similarly, avoid reporting the standard error of a coefficient estimate as (0.000). This can often be accomplished by scaling the variable in question.
- Any “stars” denoting levels of significance (i.e., p<.05, p<.01) should be placed immediately to the right of the coefficient (not to the right of the standard error).
- Such signifiers may not be more than one character wide, so use different symbols for different levels.
- Do not include more than two levels of significance tests as a general rule.
- In almost all cases, do not report significance tests to more than two digits after the decimal point (e.g., do not report p < 0.001, but instead report p<0.01).
- Include a note explaining the chosen levels of significance
- Variable descriptions should be short but comprehensible, so that they can be made to fit in a single row. Do not report variable names used in statistical software (e.g., EDUC)
- Limit the number of notes to the table
Additional guidelines for tables with regression output
- Present the standard errors of coefficient estimates, not t-statistics or z-statistics
- Each column should be numbered in the form (1) (2) etc. Additional short column labels are permitted, for example to denote different dependent variables. However, avoid repetition of the same label in sequential columns
- Except in rare cases when needed for clarity, the left-most column should not have the heading “Variable”
- Variable names should be in the same row as the coefficient estimate
- To the extent possible, avoid large and complex tables.
- Coefficient estimates for control variables that are not discussed in the text may be moved to an appendix or in some cases omitted
- In general avoid presenting more than six columns of regression estimates in a single table
- Be selective in which ancillary statistics to report, and how to report them. For example, if the number of observations is the same for all models, that number can be reported in a note to the table instead of a row with redundant numbers.