RE: st: table formatting in stata

[Haluk Vahaboglu <vahabo@hotmail.com>]

Hi Sergiy,
I would like to thank you again for the below code you provided and the ado "ttesttab".
May I take your attention to the difference between outputs obtained by executing the code below and the "ttesttab" ado.

The code you provided below returns "n" of variables correctly, which means this code treats missing values independently for each row (variable).
But the ttesttab command treats missing values in a different way.

sysuse auto, clear
replace price = . in 1

The code returns:
price    51
length    52
weight    52
headroom    52
 
ttesttab returns:
price    51    
length    51    
weight    51   
headroom    51  

I believe missing values should be treated case-wise as returned with the code. Could you help to fix this problem with the "ttestab" ado.

Regards    






	
	
	
	
@page
{;}
.ExternalClass P
{margin-bottom:0.08in;}






	
	
	
	
@page
{;}
.ExternalClass P
{margin-bottom:0.08in;}



Prof. Dr. Haluk Vahaboğlu
Istanbul Medeniyet
Üniversitesi, 
Göztepe Eğitim ve Araştırma Hastanesi
Enfeksiyon Hastalıkları
ve Klinik Mikrobiyoloji ABD
Dr. Erkin Caddesi  34730
Kadıköy / Istanbul  TURKIYE



> Date: Thu, 6 Jun 2013 18:38:57 -0400
> Subject: Re: st: table formatting in stata
> From: serjradyakin@gmail.com
> To: statalist@hsphsun2.harvard.edu
> 
> How about this? Sergiy
> 
> sysuse auto, clear
> 
> local vars "price length weight headroom"
> local gvar "foreign"
> 
> 
> local n=`:word count `vars''
> matrix R=J(`n',7,.)
> matrix list R
> 
> forval i=1/`n' {
>   local w `:word `i' of `vars''
>   quietly ttest `w', by(`gvar')
>   matrix R[`i',1]=r(N_1)
>   matrix R[`i',2]=r(mu_1)
>   matrix R[`i',3]=r(sd_1)
>   matrix R[`i',4]=r(N_2)
>   matrix R[`i',5]=r(mu_2)
>   matrix R[`i',6]=r(sd_2)
>   matrix R[`i',7]=r(p)
> }
> 
> matrix coleq R = " " "Group 1" " " " " "Group 2" " " " "
> matrix colnames R = N mean SD N mean SD p
> matrix rownames R = `vars'
> 
> matrix list R , format(%4.3f)
> 
> On Thu, Jun 6, 2013 at 6:18 PM, Haluk Vahaboglu <vahabo@hotmail.com> wrote:
>> Sergiy,
>> Thank you for your contribution.
>> Unfortunately, tabstat do not generate a "p" value of T Test. I need n mean sd & "p" on a line.
>> Therefore, neither tabstat not tabstatmat resolve this problem.
>>
>>
>> Prof. Dr. Haluk Vahaboğlu
>> Istanbul Medeniyet
>> Üniversitesi,
>> Göztepe Eğitim ve Araştırma Hastanesi
>> Enfeksiyon Hastalıkları
>> ve Klinik Mikrobiyoloji ABD
>> Dr. Erkin Caddesi 34730
>> Kadıköy / Istanbul TURKIYE
>>
>>
>>
>>
>> ----------------------------------------
>>> Date: Thu, 6 Jun 2013 16:13:32 -0400
>>> Subject: Re: st: table formatting in stata
>>> From: serjradyakin@gmail.com
>>> To: statalist@hsphsun2.harvard.edu
>>>
>>> Yes, it is possible:
>>>
>>> sysuse auto,clear
>>> tabtstat price weight length, stats(mean sd)
>>>
>>> If you need it transposed, (and I assume you do) then add an option
>>> save to the above, then use Austin Nichols' -tabstatmat- to put
>>> results into matrix and transpose it. Type:
>>>
>>> findit tabstatmat
>>>
>>> and follow instructions.
>>>
>>> Best, Sergiy
>>>
>>>
>>> On Thu, Jun 6, 2013 at 5:11 AM, Haluk Vahaboglu <vahabo@hotmail.com> wrote:
>>>> TD P { margin-bottom: 0in; }P { margin-bottom: 0.08in; }
>>>>
>>>>
>>>> Hello Everybody,
>>>> In our research papers we
>>>> mostly present continuous data in a table with comparison to cases vs
>>>> controls as "n mean SD and p (from Ttest)". Here is a brief
>>>> example of such a table:
>>>>
>>>>
>>>>
>>>> ................Cases / Controls
>>>> ........N mean (SD) N mean (SD)...p
>>>> Var1
>>>> Var2
>>>>
>>>>
>>>>
>>>> With STATA I use these
>>>> commands
>>>>
>>>>
>>>>
>>>> 1. "estpost tabstat age
>>>> VAR1 VAR2, by(C/C) s(n mean sd) c(s)"
>>>> 2. "esttab ., main(mean)
>>>> aux(sd) unstack noobs nonote wide compress label"
>>>> 3. "estpost ttest age
>>>> VAR1 VAR2, by(C/C) une w"
>>>> 4. "esttab ., cell("p
>>>> (fmt(3))")"
>>>>
>>>>
>>>>
>>>> and get two separate
>>>> tables "n mean SD" in one and "p" in the other. I afterwards
>>>> combine these two in one table formatted as mentioned above.
>>>>
>>>>
>>>>
>>>> Is there any simple way to
>>>> get a table with n-mean-SD-p all in one line for multiple continuous
>>>> variables?
>>>> Thanks again
>>>> note: sorry for the subject line of my previous message
>>>>
>>>> *
>>>> * For searches and help try:
>>>> * http://www.stata.com/help.cgi?search
>>>> * http://www.stata.com/support/faqs/resources/statalist-faq/
>>>> * http://www.ats.ucla.edu/stat/stata/
>>>
>>> *
>>> * For searches and help try:
>>> * http://www.stata.com/help.cgi?search
>>> * http://www.stata.com/support/faqs/resources/statalist-faq/
>>> * http://www.ats.ucla.edu/stat/stata/
>>
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>> *   http://www.ats.ucla.edu/stat/stata/
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/ 		 	   		  

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/