April 8, 2006


Common Mistakes

These are the most common mistakes students in G604 (PhD IO) make in writing up
regression analysis.

 1. Don't use computer codes as names or abbreviations for variables.

2. Don't put boxes around tables and figures. Do not use more horizontal and
vertical lines than are  useful.

3. Include a title page,  an abstract, a date, and your email address.

4. Say what software you use.

5. Do not report standard deviations and F-statistics unless you have some
reason to report them.

 6. Do not report numbers to more decimal places than is interesting (i.e., say
2.13, not 2.134292022).

7. Do not use keys to graphs or charts--- label the curves or regions directly,
so the reader does not have to move his eyes between a key and the chart itself.

8. If you are using time series data, include  time-period dummies or a time
trend, since serial correlation will be a big problem. Fixed effects are also
usually necessary for time-series cross-section data. Use robust standard
errors.

9. Tables should explain what is going on in them. If you exclude outliers, for
example, say how you define "outlier".

10. Label  each regression column in a table with a number.

11. Give sources for your data.