How to save descriptive results in Stata data (.dta) format?

[Biswa bhusan]

Dear user, I used this code:
sysuse auto.dta
mean price, over ( foreign )

My results are:
--------------------------------------------------------------
-
Over | Mean Std. Err. [95% Conf. Interval]
-------------+------------------------------------------------
price |
Domestic | 6072.423 429.4911 5216.449 6928.398
Foreign | 6384.682 558.9942 5270.608 7498.756
-------------------------------------------------------------

Now i want to save these mean, standard deviation and confidence interval in Stata data format (.dta). How to do it? if there is code please share.

[Amin Sofla]

You might want to take a look at here.

[Nick Cox]

. sysuse auto
(1978 Automobile Data)

. statsby , by(foreign) : ci means price
(running ci on estimation sample)

      command:  ci means price
            N:  r(N)
         mean:  r(mean)
           se:  r(se)
           lb:  r(lb)
           ub:  r(ub)
        level:  r(level)
           by:  foreign

Statsby groups
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..

. list 

     +-------------------------------------------------------------------+
     |  foreign    N       mean         se         lb         ub   level |
     |-------------------------------------------------------------------|
  1. | Domestic   52   6072.423   429.4911   5210.184   6934.663      95 |
  2. |  Foreign   22   6384.682   558.9942    5222.19   7547.174      95 |
     +-------------------------------------------------------------------+

[Biswa bhusan]

Dear Nick, The code you have provided might need to rearrange. I used the code:
statsby , by( foreign ) : means price

Please suggest me the in the next query:
I want to estimate means if rep78==3 and rep78==2. Then, i want to save the both Stata files into one Stata file. After that i need to plot means of both rep78==3 and rep78==2 across the foreign. Please suggest me how to do?

statsby , by( foreign ) : means price if rep78==3
statsby , by( foreign ) : means price if rep78==2

[Nick Cox]

My code seems to save what you want so I am not clear why you want to "rearrange" it.

The extra twist of working with further subsets can be met with

sysuse auto, clear 
statsby if inlist(rep78, 2, 3), by(foreign rep78) : ci means price

[Bruce Weaver]

Hi Biswa. I can't find the old thread right now, but the use of -mean- to get the CIs for two or more groups has come up before, because -mean- uses one critical value of t for all of the groups, and it is the critical t with df = N_total - 1. Run the following code to see that this happens for your example.

clear *
sysuse auto.dta
ttest price, by(foreign)
mean price, over ( foreign )
matrix list r(table)

Compare the CIs from the -ttest- output to those from -mean-. And notice that for both foreign and domestic vehicles, -mean- uses the same critical t-value with df = 73.

r(table)[9,2]
            price:     price:
         Domestic    Foreign
     b  6072.4231  6384.6818
    se  429.49109  558.99417
     t  14.138647  11.421732
pvalue  1.353e-22  6.520e-18
    ll  5216.4486   5270.608
    ul  6928.3976  7498.7556
    df         73         73
  crit  1.9929971  1.9929971
 eform          0          0

Unless this is really how you want to compute the CIs (which I doubt), I suggest using another method, such as -statsby-, as suggested by Nick in #3 and #5.

HTH.