## asdoc: Cutomizing the regression output | MS Word from Stata | Confidence Interval, adding stars, etc.

Category:asdoc,Blog,Uncategorized

Version 2.3 of asdoc adds the following features for reporting detailed regression tables.

1. Reporting confidence interval

2. Suppressing confidence intervals

3. Suppressing the stars which are used to show significance level

4. Customization of significance level for stars

These features are discussed in details below. If you have not already studied the features of asdoc, you can visit this page that lists the table of contents of what asdoc can do. You can also read this one paragraph introduction to asdoc. The following line of code will install this beta version of asdoc from our website

net install asdoc, from(http://fintechprofessor.com) replace

help asdoc

## Details of the new features

The new features related to creating detailed regression tables with asdoc are discussed below with details.

### 1. Confidence interval

I received several emails and comments on blog posts suggesting the addition of confidence intervals (CI) to the detailed regression tables created by asdoc. In version 2.3 onwards, confidence intervals are shown by default. This means that we do not have to add an additional option to report CI. See the following example.

sysuse auto, clear

asdoc reg price mpg rep78 headroom trunk weight length turn , replace

### 2. Suppressing the confidence interval

If confidence intervals are not needed, we can use option **noci**. For example

asdoc reg price mpg rep78 headroom trunk weight length turn , replace noci

### 3. Suppressing stars

Similarly, if we are not interested in reporting significance stars, we can use option **nostars**. For example,

### 4. Setting custom significance level

The default significance levels for reporting stars are set at : *** for p-values <=0.01; ** for p-values <=0 .05, and * for p-values <=0.1. However, now we can set our own levels for statistical significance using option **setstars**. An example of setstars option looks like:

setstars(***@.01, **@.05, *@.1)

As we can see from the above line, **setstars **separates each argument by a comma. Each argument has three components. The first component is the symbol (in our case it is *) which will be reported for the given significance elve. The second component is the @ sign that connects the significance level with the symbol. And the third component is the value at which the significance level is set. So if we want to report stars such that

* for p-value .001 or less

** for p-value .01 or less

*** for p-value .05 or less

We shall write the option setstars as

setstars(*@.001, **@.01, ***@.05)

Continuing with our example, let us use the above option to report our defined level of stars.

asdoc reg price mpg rep78 headroom trunk weight length turn , replace setstars(*@.001, **@.01, ***@.05)

## 17 Comments

## Ya Zheng

January 7, 2019at 8:16 pmDear Dr. Attaullah Shah,

I am studying your powerful Stata package of asdoc. It doesn’t seem to be available for logistic regression when I want to output OR, right?

Thank you!

## Attaullah Shah

January 7, 2019at 9:42 pmDear Zhengya

Yes, it can be used with logistic regression. See this thread, post 4 and #9 where I have shown how to do it.

https://www.statalist.org/forums/forum/general-stata-discussion/general/1437027-exporting-logistic-regression-outputs-or-and-ci-to-excel

## Ya Zheng

January 7, 2019at 11:20 pmDear Dr. Attaullah Shah,

Thank you for getting back to me. The package will work if I wanna get OR and the 95%CI for OR?

## David Straarup

January 23, 2019at 7:06 pmDear Dr. Attaullah Shah

Tried to update to ver. 2.3 as you recommend.

I have still ver. 2.2??

## Attaullah Shah

January 24, 2019at 8:46 amDavid StraarupDid you install asdoc from SSC from this website? I have not yet uploaded version 2.3 to SSC as I need to test it further. Version 2.3 is available on this website, to download this beta version, you can type the following in the Stata command window

Update: The new version is now available on SSC, to install

## David Straarup

January 24, 2019at 12:11 pmYes, tried update from your location.

The version is still ” Oct 23, 2018, ver 2.2.0″ in “help asdoc”!

## david straarup

January 24, 2019at 12:22 pmDon´t know why….now it works with CI-interval.

Thank you!

## Dong

April 5, 2019at 6:29 pmDear Dr. Attaullah Shah,

I am studying your powerful Stata package of asdoc. When the label or title include Chinese, the result appear messy code in the word. It doesn’t seem to support unicode ?

Thank you!

## Attaullah Shah

April 5, 2019at 6:35 pmDong: I replied to a similar query on Statalist here – posts 45-46, where Chen Samulsion asked about the possibility of passing Chines characters from Stata to a Word file using asdoc. My reply was that getting Chinese characters in an RTF file is a pain in the neck. Usually, getting Chinese characters programmatically into an RTF file does not seem straightforward, one has to first convert the letters to Unicode and then render it properly in the RTF file.I think similar is the case with Nordic characters. Right now, I am struggling with finding a way out.

## Dr Joanna F Dipnall

June 1, 2019at 3:42 amI also think you should check to report the correct std errors with the transformed coefficients in logit. The standard error reported by logit with the or optionis computed using the delta method. For example, to manually compute the standard errors:

use http://www.stata-press.com/data/r15/bangladesh.dta, clear

logit c_use age i.urban, vce(cluster district) nolog

logit c_use age i.urban, vce(cluster district) or nolog

* after -logit, or-, e(b) and e(V) still store the result of the coefficients rather than the ORs

mat V=e(V)

display sqrt(exp(_b[age]*2)*el(V,1,1))

Stata uses the delta method here because the logit coefficients are approximately normally distributed comparing with the ORs – See:

https://www.stata.com/support/faqs/statistics/delta-rule/

## Attaullah Shah

June 9, 2019at 5:54 pmDr JoannaI have uploaded a revised version of asdoc, it solves the mentioned problem and brings within few other updates. To update, please type the following line and paste in Stata

## Ali

July 30, 2019at 5:45 pmDear Dr. Attaullah Shah,

Thank you for your guidance with asdoc.

When you change stars for a significance level (P-values), the footnote below the table doesn’t automatically correspond to the star changes in the table?

## Attaullah Shah

July 30, 2019at 6:14 pmPlease install the updated version of asdoc, that has an update to setstars() option. The new version also changes the footnote accordingly.

The new version of asdoc can be installed from my site. Copy and paste the following line in Stata and press enter.

Please note that the above line has to be copied in full. After installation of the new version, then restart Stata.

Please do remember to cite asdoc. To cite:

In-text citation

Tables were created using asdoc, a Stata program written by Shah (2018).

Bibliography

Shah, A. (2018). ASDOC: Stata module to create high-quality tables in MS Word from Stata output. Statistical Software Components S458466, Boston College Department of Economics.

## Eva Asselmann

August 16, 2019at 12:37 amDear Professor Shah,

thank you very much for your asdoc feature, which is just wonderful!! I am wondering how I can change the confidence interval for regression tables from 95% to 99%? I would like to report the 99% upper and lower bounds, but cannot find any information on how to do this with asdoc at your homepage.

Thank you very much for your help and advice!

## Attaullah Shah

August 16, 2019at 12:41 amHello Eva

You can use the level(99) option to report 99% confidence interval

Example

Please cite asdoc in your research.

In-text citation

Tables were created using asdoc, a Stata program written by Shah (2018).

Bibliography

Shah, A. (2018). ASDOC: Stata module to create high-quality tables in MS Word from Stata output. Statistical Software Components S458466, Boston College Department of Economics.

## Ashleigh Morrice-West

August 22, 2019at 12:46 pmDear Dr. Attaullah Shah ,

I was hoping to ask for your help in using your asdoc program in stata. I have been really impressed with how nicely the output is generated in word for multivariable models, certainly much better than what I’ve been able to achieve in the past from various attempts at esttab/estout/regout etc. I am undertaking my PhD at the University of Melbourne. I have read the help file but still have a few questions.

My first issue that I’m having is that I am having is that i am running poisson regression models (or perhaps nbreg) and I would like the output to be the exponential of the coefficient to represent IRR’s. I read the following link of discussion for logistic regression and by your conversation it appears that with the logit command odds ratios are presented in the table. I have “eform” in my code which seems to be being ignormed when I add asdoc in front of the model. I’m not sure if there is another way to overide the coefficients?

https://www.statalist.org/forums/forum/general-stata-discussion/general/1437027-exporting-logistic-regression-outputs-or-and-ci-to-excel

Secondly, I am trying to tabulate the univariable models to include in a supplementary file for my publication from the univariable screening process as previously I have copy-pasted and with a very large number of variables this is becoming very laborious. I essentially need each univariable model to have the e^coefficient 95%ci and p-value without a constant.

The code I was running was going to have for about 20 variables(where I am looking at the effect of trainer and workload factors on race performance with the exposure as the number of races undertaken)

asdoc glm Career_All_Wins i.trainer_category , family(poisson) link(log) exposure(Career_All_Starts) vce(robust) eform cnames(Variable) title(Univariable test) dec(2) save(test_table2) notse nostars label replace drop(Constant)

asdoc glm Career_All_Wins log_numhorses_1 , family(poisson) link(log) exposure(Career_All_Starts) vce(robust) eform cnames(Variable) dec(2) save(test_table2) notse nostars label append drop(Constant)

*I was hoping to generate each variable as another line in the same table but the way I have coded this it starts a new table for each variable

*the notse I thought from the help file would drop the t statistic and the standard error but these columns still stay in the output per this code

*drop(_cons) and drop(Constant) do not seem to work

* I thought the label command would use the variable labels but it outputs the 0.category, i.category etc

* the other thing I was trying to do was to get multiple models to output side by side to have the predictor on the left and then different outcomes sequentially (i have outcomes of wins, places and prize money in separate models). I thought the wide command might help but its not really what I’m after. I am happy to merge columns of different model tables later though in word if this is not possible.

I wondered if I could loop the code over the list of variables, but I haven’t been able to find any examples online of people trying this.

Apologies for the lengthy email. Any assistance would be greatly appreciated.

## Mamun

October 7, 2019at 7:42 amCan you please help me to report pvalue instead of standards errors in nested table outcome of asdoc.