Number format which does not add zeros to natural numbers

[Mor Zahavi]

I want Stata to show decimals only for numbers which are not natural. For instance - I want the number 5.257 to be presented as 5.25, but the number 1 to be presented as 1 (and not as 1.00).

How can I do it?

[Maarten Buis]

See help format Here is an example:

. di %9.2f 1.00
     1.00

. di %9.2f 1
     1.00

. di %9.2g 1
        1

. di %9.2g 1.00
        1

[Mor Zahavi]

Thanks Maarten.

This does not seem to solve my issue.

I am looking for a format which would turn the original numbers (left) to the formatted numbers (right):


1.23456 = 1.23

1 = 1

And not:


1.23456 = 1.23

1 = 1.00

[Nick Cox]

What do you want this format for? Listing of data, listing of results in table, display on graphs? You may need to create your own string variable to display, say

gen show = cond(whatever == round(whatever), strofreal(whatever), strofreal(whatever, "%3.2f"))

[Mor Zahavi]

I need it for a table of descriptive statistics.


MWE:

input v1 v2
1 3
1 3
3 3
end
su

I want that instead of 1.666667 I would have 1.66
But that the 3 below will not be converted to 3.00

Is it possible?

[Maarten Buis]

The problem you are running into is that summarize does not have a format() option, but you can get the same statistics with tabstat, which does have a format() option.

. clear

. input v1 v2

            v1         v2
  1. 1 3
  2. 1 3
  3. 3 3
  4. end

.
. sum

    Variable |        Obs        Mean    Std. Dev.       Min        Max
-------------+---------------------------------------------------------
          v1 |          3    1.666667    1.154701          1          3
          v2 |          3           3           0          3          3

. tabstat *, format(%9.3g)

   stats |        v1        v2
---------+--------------------
    mean |      1.67         3
------------------------------

[Mor Zahavi]

Perfect thanks Maarten.

[Mor Zahavi]

Actually it's not perfect. This is what I get when I format with %9.3g:

[Nick Cox]

A negative principle:

Stata can't know which numbers are special to you unless you tell it.

I don't have a different solution here, because I don't understand your preferences. (in #8 which column is the one you want?)

What I have found is that sometimes you have to do all the work. For example, if a graph axis runs from 0 to 1, then sometimes I really want to see 0 0.25 0.5 0.75 1 -- and certainly not 0.0 or 0.00 or whatever. That is an (a)esthetic preference rather than anything else, but I just spell it out to be certain of getting what I want.

yla(0 "0" 0.25 "0.25" 0.5 "0.5" 0.75 "0.75" 1 "1", ang(h))

[Mor Zahavi]

Thanks Nick. The column I'm looking for in #8 is the following:

.522
.062
3.08
.332
.064
.095
.107

Which means - I want 3 decimal numbers, but without trailing zeros.

Your code shows exactly what I want, and I am aware of your graph axis format program which is exactly what I am looking for, however I need to apply this format in other instances than graph axis.

It is quite a surprise that there is no built-in format option which removes trailing zeros. It is definitely required in other instances than graph axis, and seems like a straightforward requirement.

[Nick Cox]

It does not surprise me. It seems that you want 3 decimal places, except that you then want to elide trailing zeros. You can do that by a string manipulation, but otherwise it's a hybrid rule of your own invention. 1.23 and 1.230 are not equivalent in general, as 1.23 is rounded to 2 decimal places and 1.230 to 3 decimal places.

My example has different flavour as I know that what I want to show is exact to the detail given. But when values are data or results, the story is different either way.

[Mor Zahavi]

Dear Nick.

I just think that Stata should provide the option to improve the display of the attached table. I need decimal places but not when the number is zero or one (or any other natural number).

[Nick Cox]

gen show = cond(x == floor(x), strofreal(x, "%1.0f"), strofreal(x, "%4.3f"))

[Mor Zahavi]

Dear Nick,

Thanks for your reply.

I do not understand how to use this command for a table of summary statistics?

Right now I use -

estpost su var1 var2
eststo est_1
esttab est_1 using file_1.tex, replace ///
cells("count(fmt(%12.0fc)) mean(fmt(3)) sd(fmt(3)) min(fmt(3)) max(fmt(3))")

[Andrew Musau]

esttab is from Stata Journal, as you are asked to explain. You have received some good responses from both Maarten and Nick, but because you do not disclose what your end-goal is, you are not able to resolve your problem. While Maarten is right that summarize does not have a -format()- option, esttab does, and you are clearly not using it.

min(fmt(%9.3g))