Explanation of tabstat command

[Nick Hale]

Hello,

I am a little confused when running the "tabstat" command. For example, I entered the command "tabstat male marry single, by(smoke) statistics(mean sd n) longstub format (%9.1g)" on a dataset of mine. Take male for example, the dummy variable is smoke (column is either 0 or 1), and the 0 column has mean=.33, sd=.47, N=40200; and the 1 column has mean=.93, sd=.25, N=9552. The total column below them has mean=.45, sd=.5, and N=49752.

What do the statistical (mean, sd, N) variables actually mean within the 1 and 0 columns? And within the total?

[Nick Cox]

33/100 non-smokers are male and 93/100 smokers are male.

So, the number of males is an integer close to

. di 40200 * .33 + 9552 * .93
22149.36

and the overall proportion of males is something like

. di (40200 * .33 + 9552 * .93) /(40200 + 9552)
.44519537

The proportion of smokers is

. di 9552/ (40200 + 9552)
.19199228

So, which country?

Flipping the problem round, so that smoke is the outcome may be closer to what you want to know.

[Joro Kolev]

You might want to show a sample of your data using -dataex-, and phrase your questions with respect to the results on the data you have shown.

I am not on very clear what you are asking. Here:

. sysuse auto, clear
(1978 Automobile Data)

. tabstat price, by(foreign) statistics(mean sd n) longstub format (%9.1g)

foreign      variable |      mean        sd         N
----------------------+------------------------------
Domestic        price |      6072      3097        52
Foreign         price |      6385      2622        22
----------------------+------------------------------
Total           price |      6165      2949        74
-----------------------------------------------------

. summ price

    Variable |        Obs        Mean    Std. Dev.       Min        Max
-------------+---------------------------------------------------------
       price |         74    6165.257    2949.496       3291      15906

So you get the statistics you have requested by the groups defined by foreign, and then on the bottom you get the statistics you have requested for the whole dataset.

[Nick Hale]

Here is a screenshot of my dataset