descriptive statistics table

[Meike Wenzl]

It seems that if I have solved a problem in Stata, I create new ones:for my analysis I standardized several variables in Stata (according to Dawson, 2014). So now I have a lot of means that display the e-function. I have never seen those tables in other papers and now I am worried that maybe standardizing was not the best decision after all... any suggestions what I can do or how I can get rid of the e-function in the excel table?

Variable	Mean	Std. Dev.
workforcechange	.024971	.1593084
ROAchangec	1.19e-10	1,000
Narcissism Indexc	-1.42e-08	1,000
ROAchange x narcissim Index	.0111959	1054418,000
Firm sizelc	-8.76e-10	1,000
Revenuec	-3.95e-10	1,000

[Marcos Almeida]

If I understood correctly, your question is not related to modeling, but on how to get rid of the scientific notation.

For this, you may type:

. format ROAchangec Narcissism Firm Revenuec %10.0f

Best,

Marcos

[Meike Wenzl]

Marcos:

thanks for your reply. I did type your code but nothing changed

[Rich Goldstein]

please read the FAQ (what is "Dawson (2014)"?)

this appears to be a question about Excel, not Stata - unless you are having the same issue in Stata; if you are, please see "help cformat"

[Meike Wenzl]

Dawson (2014) is a paper which suggested to standardize variables in a regression.

All the variables I standardized are expressed with the e-function. I do have the same problem in Stata, but nothing I do seems to help. I also already read the help format file, but nothing that is suggested there helps me to get rid of the e-display.

[Marcos Almeida]

Sorry, I didn't pay attention to the negative sign.

You may try this:

. set obs 10
number of observations (_N) was 0, now 10

. gen x = runiform()

. gen y = runiform()

. gen z = x/1000000000

. gen w = y/1000000000

. list

     +-------------------------------------------+
     |        x          y          z          w |
     |-------------------------------------------|
  1. | .1649395   .2278204   1.65e-10   2.28e-10 |
  2. |  .932945   .5782465   9.33e-10   5.78e-10 |
  3. | .3999315   .7533595   4.00e-10   7.53e-10 |
  4. | .9881987   .8570072   9.88e-10   8.57e-10 |
  5. | .9287856   .9322746   9.29e-10   9.32e-10 |
     |-------------------------------------------|
  6. | .6640378    .324447   6.64e-10   3.24e-10 |
  7. | .0368038   .1637711   3.68e-11   1.64e-10 |
  8. | .3336498    .958201   3.34e-10   9.58e-10 |
  9. | .7824295   .6008608   7.82e-10   6.01e-10 |
 10. | .0170049   .9733476   1.70e-11   9.73e-10 |
     +-------------------------------------------+

. format z w %20.15f

. list

     +-------------------------------------------------------------+
     |        x          y                   z                   w |
     |-------------------------------------------------------------|
  1. | .1649395   .2278204   0.000000000164940   0.000000000227820 |
  2. |  .932945   .5782465   0.000000000932945   0.000000000578247 |
  3. | .3999315   .7533595   0.000000000399931   0.000000000753359 |
  4. | .9881987   .8570072   0.000000000988199   0.000000000857007 |
  5. | .9287856   .9322746   0.000000000928786   0.000000000932275 |
     |-------------------------------------------------------------|
  6. | .6640378    .324447   0.000000000664038   0.000000000324447 |
  7. | .0368038   .1637711   0.000000000036804   0.000000000163771 |
  8. | .3336498    .958201   0.000000000333650   0.000000000958201 |
  9. | .7824295   .6008608   0.000000000782430   0.000000000600861 |
 10. | .0170049   .9733476   0.000000000017005   0.000000000973348 |
     +-------------------------------------------------------------+

Best,

Marcos

[Meike Wenzl]

Marcos:

Thank you so much! That worked perfectly.

I am not familiar with descriptive statistic tables that show such a small mean. Is that ok to post those really small means or is it rather advisable to post the e-funciton ones?

[Marcos Almeida]

I believe you should follow the literature related to you study. That said, you could re-scale the variables, use elasticities or employ - margins - followed by - marginsplot - for a visual presentation. You may check how to perform all this on the help files and the Stata manual. Since your theme is quite far from my field, I prefer to stay at this level of recommendation.

[Meike Wenzl]

Thanks!

[Marcos Almeida]

Just a final thought, that passed unnoticed at the first inspection of your results.

Apparently, you're deailing with centered means, if I take the "c" in the end of the variable names as a sign for centering... Is that true?

If so, you shouldn't expect something different, rather than having the mean of a centered variable around zero (decimals will depend on rounding and precision issues, related to the type of stored variables in Stata).

Furthermore, you have several variables with SD = 1 and mean around zero, due to the standardization. In short, that's the same we should expect under a standard uniform distribution.

Hopefully that helps to solve the issues raised in your query!

Best,

Marcos

[Meike Wenzl]

Marcos:

thanks for clarifying. Yes I standardized the variables with the mean =0 and the s.d. =1. I never came across such means and s.d.'s in journals so I thought I might have done something wrong. But I guess I can just include 0.0000 then in my descpriptive analysis table.

[Nick Cox]

On the contrary, once you have standardized variables to mean 0 and SD 1 there is nothing of interest in reporting that directly. The point of standardization is to use such variables in contexts where you think that helps in some way.