Regression results displayed in scientific notation

[ChengLong Wang]

When I conduct regression analysis, the coefficients and confidence intervals on the regression results interface are both displayed in scientific notation. However, in the reference book, specific values are displayed normally. Is there any method to make them display the values normally?

[Nick Cox]

At a guess the problem lies in your choice of currency units for income. Scale by dividing by 1000 or 1 million or some other convenience factor.

Alternatively, income as response variable often makes less statistical (or even economic or financial) sense than log of income. An important detail here is whether your model makes any negative predictions (unless naturally you have some negative values).

[Andrew Musau]

See

help set cformat

Most likely, you have a fixed format at present. Consider:

di %9.3f -12419.2412345
di %9.3f 2412.33912345

Res.:

. di %9.3f -12419.2412345
-1.24e+04

.
. di %9.3f 2412.33912345
 2412.339

A general format will give you the wanted result:

di %10.7g -12419.2412345
di %10.7g 2412.33912345

Res.:

. di %10.7g -12419.2412345
 -12419.24

.
. di %10.7g 2412.33912345
  2412.339

[Nick Cox]

Showing us the results of something like

scatter realrinc age, by(female) mc(blue%10)

would be illuminating.

[ChengLong Wang]

At a guess the problem lies in your choice of currency units for income. Scale by dividing by 1000 or 1 million or some other convenience factor.

Alternatively, income as response variable often makes less statistical (or even economic or financial) sense than log of income. An important detail here is whether your model makes any negative predictions (unless naturally you have some negative values).

"Thank you for your response! As I am a beginner, I followed the original data from the book and ran the commands as instructed in the book. The original data did not involve any scaling of the income variable, so the results obtained should correspond to those in the book."

[ChengLong Wang]

Showing us the results of something like

scatter realrinc age, by(female) mc(blue%10)

would be illuminating.

[ChengLong Wang]

See

help set cformat

Most likely, you have a fixed format at present. Consider:

di %9.3f -12419.2412345
di %9.3f 2412.33912345

Res.:

. di %9.3f -12419.2412345
-1.24e+04

.
. di %9.3f 2412.33912345
2412.339

A general format will give you the wanted result:

di %10.7g -12419.2412345
di %10.7g 2412.33912345

Res.:

. di %10.7g -12419.2412345
-12419.24

.
. di %10.7g 2412.33912345
2412.339

Thank you for your response! Is there any way to directly restore the specific values in the regression results?

[Nick Cox]

I don't know the book in question but to me the plot in #6 raises serious questions about the analysis. Please give a reference in standard style following https://www.statalist.org/forums/help#references, i.e. authors, title, edition and date, publisher, etc.

Andrew Musau is indicating that you can change display formats and you need to do that if the task is exact reproduction. See the help for estimation options for options like cformat().

[ChengLong Wang]

I don't know the book in question but to me the plot in #6 raises serious questions about the analysis. Please give a reference in standard style following https://www.statalist.org/forums/help#references, i.e. authors, title, edition and date, publisher, etc.

Andrew Musau is indicating that you can change display formats and you need to do that if the task is exact reproduction. See the help for estimation options for options like cformat().

"Thank you very much for your guidance! The specific reference information is Mitchell, M. N. (2012). Interpreting and visualizing regression models using Stata (Vol. 558). College Station, TX: Stata Press. The specific chapter is 3.5 Main effects with polynomial terms on page 107."

the display formats is a good way to reproduction