Publication quality regression tables with asdoc in Stata – video example
Category:asdocCreating publication-quality tables in Stata with asdoc is as simple as adding asdoc to Stata commands as a prefix. asdoc can create two types of regression tables. The first type (call it detailed) is the detailed table that combines key statistics from the Stata’s regression output with some additional statistics such as mean and standard deviation of the dependent variable etc. This table is the default option in asdoc. The second table is a compact table that nests more than one regressions in one table (call it nested).
In this video post, I show how to use asdoc to produce the following nested table. 
56 Comments
Nicole Martin
June 8, 2018at 10:33 pmFirstly, thank you for making your Stata package available – it’s really speeding up my workflow compared to the other options available. I’m writing to make a suggestion that you consider adding an option that value labels can be reported in regression tables as well as variable labels.
Many thanks, and best wishes,
Emily
August 1, 2018at 1:36 amCan ” asdoc” export the OR and 95% CI from logistic regression analysis?
Attaullah Shah
August 1, 2018at 2:03 amDear Emily
Version 2.0 of asdoc will have this option.
Eric
August 16, 2018at 7:00 pmasdoc is an amazing service to the field. Appreciate it very much. Like Nicole, I would love a labels option for regression outputs. That function works great for asdoc sum, but would be of great value in this function too.
Thanks again.
Attaullah Shah
August 16, 2018at 9:37 pmDear Eric and Nicole, thanks for your suggestions. I am exploring the suggested possibilities. Since version 2 of asdoc is almost ready, I may come up with label option in version 3 or version 2.1.
Sher Khan
September 10, 2018at 11:39 amAppreciate for your great work ,thank you for making your Stata package available – it’s really speeding up my workflow and saved time. Honestly , i learned a lot .
i asked one question Can asdoc command used for advance Dynamic Model ? e.g asdoc reg .. How to used for Dynamic Model to create tables ? Your response with this regard is highly appreciated
Attaullah Shah
September 10, 2018at 2:24 pmSher Khan, can you tell which specific Stata command are you referring to?
Mike
September 18, 2018at 5:26 amI am excited to see this! However, I have a question. What would the code be if we want to put several models into the same table with different IVs but the same DV? So, say in the first model I want to include IV1, in the second model IV2, and so forth. But in model 6 I want to include IVs1-IVs5. What would this code look like if I’m nesting my model?
Attaullah Shah
September 18, 2018at 2:52 pmDear Mike
Let us say that our dependent variable is DV and our independent variables are named IV1, IV2, IV3, IV4, IV5, and IV6. You have asked that what if we run a separate regression on each independent variable and then include all independent variables in one model at the end. Then our codes would be
Justin
October 11, 2018at 1:56 pmThis is a wonderful tool! Thank you so much for making this available.
I see asdoc v2.0 allows ORs from ologit in the detailed table output. Is it possible to display ORs in the nested table output as well?
Attaullah Shah
October 11, 2018at 4:41 pmJustin: I missed the option eform when coding the nest option. Hopefully, I shall add it in the next update
Justin
October 12, 2018at 4:15 pmThat would so fantastic. Being able to use the reporting options, nested ORs and 95% CIs especially, to get tables directly to Word would save me hours of table-copying.
Attaullah Shah
October 12, 2018at 4:53 pmJustin: You mean to report the 95% CIs instead of t-values or standard errors?
Justin
October 16, 2018at 5:35 pmYes!
Attaullah Shah
October 16, 2018at 7:48 pmJustin: I have noted your suggestions and will consider them in future updates of the asdoc. Thanks.
Ayub
October 27, 2018at 7:39 amThank You so much sir for sharing your knowledge with.us.sir I have one question, how can I replace t-value on se value in regression tables, I tried but it shows se value in. perenthesis in word tables.plz, guide me. best regards
Attaullah Shah
October 27, 2018at 10:43 amAyub : I think you are referring to nested regression tables. If yes, then you can use the option rep(t), which will report t-values instead of standard errors. For example,
Dr Zumin Shi
November 10, 2018at 12:08 amI like your asdoc very much. In epidemiological research, we often focus on one independent variable (recoded into quartiles) and run multiple multivariable models (with different adjustment). The regression coefficients are often reported in one table (see below). Is it possible to do it using asdoc?
Q1
Q2
Q3
Q4
p for trend
Model 1
1.00
0.86 (0.52-1.42)
0.66 (0.40-1.11)
0.57 (0.33-0.98)
0.024
Model 2
1.00
0.85 (0.51-1.41)
0.64 (0.38-1.08)
0.58 (0.34-1.00)
0.028
Model 3
1.00
0.93 (0.54-1.61)
0.76 (0.42-1.37)
0.78 (0.43-1.43)
0.332
note: model 1 adjusted for age
model 2 further adjusted for education, and physical activity
model 3 further adjusted for hypertension.
p for trend was calculated using qartiles of fruit intake as an ordinal variable in logistic regression.
The table is based on six logistic regression models.
Bassem Hamandi
November 22, 2018at 10:24 pmThank you for developing this Stata package. I wonder if there is a way to export the 95% confidence interval reported after running stcox in Stata? My apologies if this is trivial but I could not find any details in the help file.
Thank you for any help you can provide.
Attaullah Shah
November 22, 2018at 11:29 pmBassem Hamandi : Thanks for your query. Adding CI to regression tables with asdoc is on my agenda for the next update. Can you please indicate whether you want the CIs with detailed regression tables or nested tables using asdoc. You can see the help file, Section 4.1 and 4.2
Bassem Hamandi
November 22, 2018at 11:32 pmThank you. I am hoping to report using detailed regression tables.
Juan Miranda
November 30, 2018at 8:16 pmDear Attaullah Shah, good afternoon.
I’m a Stata user, but some programming things I do not understand yet.
I’ve been looking at forums like https://stackoverflow.com/questions/53524560/looping-and-storing-beta-coefficients-after-regress-over-files#new-answer
answer to a question I have.
I am not able to make my loop save the results after regressing over files (betas, t, R2 and N), with all my datasets. I have 11 named 2005.dta to 2017.dta (except for 2009) contained in a folder (F:\datasets).
The objective is to use asdoc, and to generate a summary table of the estimates I make. The code that I have written is the following.
cd "F: \ datasets" local i: dir. files "* .dta" foreach file in `i '{ asdoc reg and x1 x2 x3, append nest rep (t | se) title (Table 1 MCO per year) dec (3) }But the result is that asdoc saves the results of a single database (only the first – 2005) 11 times and does nothing for the others.
If you could have time to help me solve this problem I would thank you very much.
Cheers Juan
Attaullah Shah
November 30, 2018at 8:16 pmJuan Miranda: The question you have asked on stackoverflow is on hold, so I cannot add my answer there. The loop you are using does not use asdoc on each data set that you have. So the correct code would be:
cd "F:\datasets" local i : dir "F:\datasets" files "*.dta" foreach file in `i' { use `file', clear asdoc reg y x1 x2 x3, append nest rep(t) title(Table 1 MCO per year) dec(3) save(reg) }Please also note that there should be no space between
repand(t | se). The same applies to other options of asdoc.Also note: the option
rep(t)is used to report t-statistics. Instead, if you want to report standard errors, you would userep(se). However, you cannot use bothrep(t|se). Perhaps, you copied this from the help file of asdoc, where the symbol pipe is used to denote that one of these can be used.Dr. Naimat Ullah Khan
December 18, 2018at 2:11 pmSalam Jee,
If I wish to generate a table with standard errors instead of p-values, what I have to do with asdoc command?
Attaullah Shah
December 18, 2018at 2:13 pmDear Dr. Naimat: Which regression table do you want to generate? The
nestedone or thedetailedone? I think you are referring to the nested tables.In the nested table, you can report either standard errors or t-values. For standard errors, you can use the option
rep(se)and for t-values, you can use the option
rep(t)for example
Dr. Naimat Ullah Khan
December 19, 2018at 12:43 amDear Sir Jee
Every time I use asdoc, I pray for you my heart.
Preshit Ambade
January 5, 2019at 4:16 amHello Dr. Shah,
Wonderful work on asdoc. It has made life easy for many students like me. I am using Fairle decomposition in stata with ado. file for it. I found the asdoc command doesnot work with it. If you get time please check it and update it in the next version. Thank you.
Regards
Preshit
Attaullah Shah
January 5, 2019at 7:04 amPreshit Ambade: Thanks for your comment and feedback. Can you please share an example data and the code that does not work with asdoc. I shall work on it.
HP
January 11, 2019at 11:06 pmHello Dr. Shah,
H have few questions?
1. I am wondering if we can use order option in asdoc regresssion as
. I want to use order option like estout.
2. Is there decimal alignment options?
Thank you for your help .
HP
Attaullah Shah
January 12, 2019at 8:13 amHP:
Thanks for your suggestions. I have noted them and will consider them in the next update of asdoc.
Attaullah Shah
January 31, 2019at 11:55 amHP
Turns out that we can use the keep option for ordering variables as well as keeping required variables. Let me present two examples in this blog post as this feature needs more attention.
HP
January 12, 2019at 5:38 pmThanks. One more question.
How to report Pseudo R2 in asdoc regression results.
Thanks.
HP
Krishanu Karmakar, PhD
January 24, 2019at 9:30 pmRecently I came across your -asdoc- routine in Stata and was trying to use it for my Econometrics courses. I think I came across a bug in the routine. If you have time kindly have a look at it. The stata code is copied below.
Description of the problem: When creating tables with multiple equations, where one of the equations may have a logged dependent variable, it is quite difficult to use the eform option with asdoc – mostly the output is wrong. Probably you did intend to make “eform” an option for asdoc. It still takes the option but makes only the first equation of the set exponentiated. No matter which command I add the “eform” option to, it is always only the first equation that is exponentiated. code below.
System: Windows 10, asdoc version 2.2.0
*************************************** version 14.2 sysuse auto, clear gen lnmpg=ln(mpg) *base models, no exponentiation tried asdoc reg mpg turn, eform() save(file.doc) reset nest replace asdoc reg lnmpg turn, save(file.doc) nest asdoc reg lnmpg turn, save(file.doc) nest /// title(table1: base models, no exponentiation tried) *eform() in first command - nothing happens asdoc reg mpg turn, eform() save(file.doc) reset nest asdoc reg lnmpg turn, save(file.doc) nest asdoc reg lnmpg turn, save(file.doc) nest /// title(table2: eform in first command - nothing happens) *eform() all last command - first equation exponentiated asdoc reg mpg turn, eform() save(file.doc) reset nest asdoc reg lnmpg turn, eform() save(file.doc) nest asdoc reg lnmpg turn, eform() save(file.doc) nest /// title(table3: eform in all commands - only first equation exponentiated) ***********************************************************Anyway, notwithstanding this small bug, thank you for writing a quite useful routine.
Attaullah Shah
March 3, 2019at 8:12 amDear Dr. Krishanu Karmakar
Thanks for pointing this out. I have fixed these issues in version 2.3.3, dated Feb 23, 2019. You can install the new version from SSC
Joanna Woersching
March 3, 2019at 7:47 amHi Dr. Shah,
Thank you very much for this wonderful program, will save me hours of table editing. I was hoping to create a nested table with odds ratio, se and confidence interval listed, are these options for asdoc nest? Thank you, Joanna Woersching
Attaullah Shah
March 3, 2019at 8:18 amJoanna Woersching
I replied to your query on the Statalist here
If that is not what you are looking for, please provide some details on your desired output.
Krishanu Karmakar
March 3, 2019at 6:08 pmThank you for taking a look at it and spending your valuable time on this. This command is a wonderful addition to user written commands of Stata.
Daniel
April 5, 2019at 8:35 pmWhen I use logit with or option, it provides the coefficient instead of the odds ratio. I check the numbers, and those were coefficient, and not OR. IS there a way to display the odds ratio?
Richard Makurumidze
April 7, 2019at 3:53 amGreetings
It seems asdoc does not work with the chi (chi -square) and or (odds ratio) in logistic regression. Is this correct or am making some error?
Attaullah Shah
April 7, 2019at 10:04 amSince your pointed needed some detailed replies, I have replied to your querry in this blog post.
Micah
April 8, 2019at 7:45 pmHi there!
– Is there any way to export ONLY the coefficients in the nested regression output table? Without including standard errors OR t-values?
– Or if not, is there a way to export the table so that the standard errors / t-values are in the same table cell as the coefficients? And not in the cell below them?
Thank you so much for all that you do!!
Attaullah Shah
April 9, 2019at 12:02 amMicah:
On your suggestion, I have added option
notseto the nested regression. With this option, the output will omit standard errors or t-values. The updated version can be installed by typing the following in Stata command window: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.
Kabir Ahmed
April 20, 2019at 7:56 amIs the asdoc for un-weighted estimates?
asdoc svy: table sex urb, col row sc format(%11.0f) //gives unweighted estimates
asdoc table sex urb, col row sc format(%11.0f) //gives unweighted estimates
table sex urb[pw=wgt_test], col row sc format(%11.0f) //gives weighted estimates
kindly clarify it.
Theo
June 18, 2019at 10:07 pmHi,
Is it possible to get the same table as with nest function, with the wide format? Missing t values or standard deviation , with p values in the wide version.
Thanks.
Attaullah Shah
June 18, 2019at 11:13 pmTheo
I did not get your question. Are you asking for a wide regression table with no t-values or se?
ALI UBAID
June 19, 2019at 11:11 amDear professor,
margins are not working in my analysis. Is this command is not compatible with asdoc? or I am doing it wrong? Kindly advise
……………………………
Attaullah Shah
June 19, 2019at 11:12 amPlease provide an example of what is not working with asdoc. You can refer me to an example data set and the code that you are using with asdoc.
ALI UBAID
June 19, 2019at 11:15 amapologies for bothering you, I found the correct command online it was helpful and issued resolved.
asdoc margins, dydx(rem_for_dummy), save(margins.doc) title(marginal effect) Average marginal effects Number of obs = 3,305 Model VCE : Linearized Expression : Pr(lab_per_dum), predict() dy/dx w.r.t. : rem_for_dummy ------------------------------------------------------------------------------- | Delta-method | dy/dx Std. Err. z P>|z| [95% Conf. Interval] --------------+---------------------------------------------------------------- rem_for_dummy | .0253485 .0122308 2.07 0.038 .0013765 .0493205 ------------------------------------------------------------------------------- (note: file margins.doc not found) Click to Open File: margins.docThank you for your response.
Theo
June 19, 2019at 5:09 pmHi,
Thanks for replying.
When you use the nested function, you get SD values in parentheses under the coefficients. You also get *** for p values. And you have the dependent variable in the columns.
With the wide function, where you have the dependent variables in the rows, you do not get the SD in parentheses under the coefficients. So, i would like in the wide option to have the same model output as with the nested function, only with the dependent variables in the rows, such as with the wide option. I would like the t values to be switched out with SD in the wide option also.
Hope this is a little more clear.
Attaullah Shah
June 19, 2019at 5:41 pmTheo
Seem you might benefit from reading the help file of asdoc. It has dozens of example on how to use the different options with the wide regression tables. What you are asking is already there. See the following example where I am reporting the SE under the regression coefficients and stars with coefficients for statistical significance.
You can read asdoc help file by
. Specifically, read Section 4.3 of the help file and examples 26 to 35.
Following are the examples of wide regressions with the mentioned options
Jana Tillmann
June 25, 2019at 12:59 amDear Professor Shah,
I am very thankful for your asdoc – breakthrough – program.
My name is Jana Tillmann and I currently write my Thesis. Hopefully, you can help me with this issue, it is very urgent as I have to hand in the thesis on Thursday.
What can I do to create a nested table with multiple additional regression statistics?
When I try to create nested regression tables, I can not find a solution to extend the tables with multiple additional regression statistics.
Here is the command I use:
If I use only one additional statistic:
–
RMSE:If I use stats() instead of stat(), I don’t get any additional statistics:
If I list them separately, it only gives me the latter one:
With a wide format, it does work:
What can I do to create a nested table with a few extra regression statistics?
Attaullah Shah
June 25, 2019at 1:00 amJana
The error you are getting is due to incorrect use of the stat() option.
Since asdoc report observations by default, there is no need to write N in the stat option.
The stat() option needs that you type comma between different statistics.
So the correct syntax would be
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.
Jana Tillmann
June 29, 2019at 1:24 pmDear Professor Shah,
I am very thankful for your asdoc – breakthrough – program.
My name is Jana Tillmann and I currently write my Thesis.
Hopefully you can help me with this issue, it is very urgent as I have to hand in the thesis on Thursday.
What can I do to create a nested table with multiple additional regression statistics?
When I try to create nested regression tables, I can not find a solution to extend the tables with multiple additional regression statistics.
Here is the command I use:
If I use only one additional statistic:
– RMSE:
If I use
stats()instead ofstat(), I don’t get any additional statistics:If I list them separately, it only gives me the latter one:
With a wide format, it does work:
What can I do to create a nested table with a few extra regression statistics?
Attaullah Shah
June 29, 2019at 1:27 pmI have fixed the issue.
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.
Dr. Emily C. Phillips Galloway
September 18, 2019at 10:43 amDear Dr. Shah,
Thank you for this wonderful program!
I’m attempting to create a regression table that contains multiple regression models, but am encountering this error:
func_nested_reg(): 3001 expected 25 arguments but received 28 : - function returned errorCould you tell me what I ma doing incorrectly? I’ve updated asdoc.
Thank you!
Emily
Attaullah Shah
September 18, 2019at 10:50 amDr. Emily
Due to the complex structure of the mixed command, asdoc finds it difficult to parse the command. As an alternative, asdoc offers the replay command for such cases. See this example