Doubts in coding "mark" command

[Brian Yalle]

Hi,
I'm using a cross-sectional survey data. I run regression various model specifications, and I want to compare them. The complete model only considers 7 variables of all those available in the survey:

logit depvar var1 var2 var3 var4 if var5==var6-1

As can be seen, the model has 4 independent variables and 2 variables used as conditional (var5, var6). The specifications will include or exclude each of this 4 variables. As example,

logit depvar var1 var3
logit depvar var2 var4 if var5==var6-1
logit depvar var2 var3 var4

In order to know that the difference is caused by specification and not different samples, I found mark command useful.
The command line using mark woud be like this:

mark nomiss
markout nomiss depvar var1 var2 var3 var4

but I have some concerns:

Should the command mark only include independent variables -excluding conditional vars like var5 and var6-?

I read in the manual reference there is a specific command for survey-characteristic variables syvmarkout.
So, I run some codes to compare the results:

• Without considering survey design:

. mark nomiss
. markout nomiss depvar var1 var2 var3 var4
. tabulate nomiss
     nomiss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |     13,787        4.25        4.25
          1 |    310,623       95.75      100.00
------------+-----------------------------------
      Total |    324,410      100.00

• Considering survey design:

. svyset, clear
. svyset cluster [pw = pweight]
. mark nomiss
. svymarkout nomiss depvar var1 var2 var3 var4
too many variables specified

Taking account of this result, I tried again but this time without any specific variable.

. svyset, clear
. svyset cluster [pw = pweight]
. mark nomiss
. svymarkout nomiss
. tabulate nomiss
     nomiss |      Freq.     Percent        Cum.
------------+-----------------------------------
          1 |    324,410      100.00      100.00
------------+-----------------------------------
      Total |    324,410      100.00

I guess this result is because the command svymarkout applies to all variables.
In view of these three results, I think they might differ by survey design but could someone explain this better?

Thanks in advance,
Brian