[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: RE: RE: Displaying values less than one with a zero in front of the comma
In regards to #1, I think we should all develop a taste for including
the leading zero. Whether or not that requires a law in Stata or
elsewhere is open to discussion.
As regards economics. What's important in results are (usually in
decreasing order) significance, sign, and size. This doesn't require
more than 2, maybe 3, digits in the mantissa. You're right, the logic
of significant digits doesn't differ from elsewhere, just the
reliability of the data and---more crucially---the model. Since this
is probably true for most social scientists using Stata and it's
probably true that most people using Stata are
social scientists, singling out economics was unnecessary.
On Feb 21, 2005, at 5:51 PM, Nick Cox wrote:
Whoa, just a small point please! I said I like this myself, not that
it should be the law!
I second two of Nick's ideas. (1) Always use the leading zero,
although having the option to decide whether or not it's included in
all formats is fine.
(2) More flexibility in applying different formats to different parts
of the display. I'm frustrated, in particular, when I want to use a
fixed format but different variables' coefficients in my regression
either have a different number of significant digits or (since the
concept of significant digits is questionable in economics) different
magnitudes of the least significant digit. Example 1: I'd rather see
I am intrigued here. I don't understand the comment about economics.
Does the logic of showing economic numbers differ from that elsewhere?
0.0032 instead of either 0.0231 and 0.0032 or 2.3E-2 and 3.2E-3.
Example 2: I'd rather see 2.3 and 0.32 than either 2.31 and 0.32 or
2.3 and 3.2E-1. ["and" takes precedence over "or" in parsing these
* For searches and help try: