(1)
> puzzled by a problem when writing an ado file. Suppose I have
> var1 var2 var3 var4 var5......vark, and I want to do the
> following loop:
> ************************************************************
> *********
> var1
> var2
> var3
> ...
> ...
> ...
> vark
> var1 var2
> var1 var3
> var1 var4
> ...
> ...
> var1 vark
> var2 var3
> var2 var4
> ...
> ...
> var(k-1) vark
> ...
> ...
> var1 var2 var3
> var1 var2 var4
> var1 var2 var5
> ...
> var1 var2 var3....vark
> ******************************************************************
> Basically, what I want to do is like step wise exhausting
> all combinations
> in a systematic way from univariate, bivariate, trivariate, to
> multivariate....Or, I can say for every variable in the
> variable list, there
> is indicator variable associated with it. I either take
> this variable in or
> out in each run. And there should be 2^k possibilities. I
> have no idea how
> to handle that. COuld anyone give me some hint? Many
> thanks in advance.
(2)
I think I might not have explained my problems clearly. I
have k indicator
variables (coded as 1 or 0) and I would like to know the
response patterns
(for example for latent class analsis) to these k
variables. For example,
var1 var2 var3 ....vark
1 0 0 0
0 1 0 0
...
1 1 0 0
...
...
...
1 1 1 1
I would like to know for each response pattern, how many
cases are there,
and programmed into an ado file. My key problem here is
how to run through
all the combinations (univariate, bivariate, and trivariate)
One posibility is that I used the following cods (or
reviced version to fit
into an ado file)
******************************************************************
clear
for num 1/6: set obs 100\ gen xX=invnorm(uniform()) \ gen DxX=xX>0.6
gen pattern=0
local i=1
while `i'<6 {
replace pattern=pattern+Dx`i'*10^(6-`i')
local i=`i'+1
}
aorder
list Dx1-Dx6 pattern
sort pattern
list pattern
gen count=1
collapse (sum) count, by(pattern)
***********************************************************
The resulting data matrix looks like:
============================
pattern count
0 16
10 10
100 5
110 6
1000 8
1010 7
1100 2
1110 1
10000 11
10010 3
10100 2
10110 2
11000 2
11010 2
11100 1
100000 7
100010 1
100100 2
101000 4
101010 1
110000 1
110010 2
110100 2
111000 1
111010 1
=================================
Here the problem is that it only presents the response
pattern that has at
least one case and it's hard to handle its order (now is
list in numerical
order: from small to big)
But what if I want to go through "each" combination (2^k
possible ways) in a
sysmatic way and list all response pattern freqeuncy though
some of them
have zero cases. What I meant by a systematic way is like:
************************************************************
*********
var1
var2
var3
...
...
...
vark
var1 var2
var1 var3
var1 var4
...
...
var1 vark
var2 var3
var2 var4
...
...
var(k-1) vark
...
...
var1 var2 var3
var1 var2 var4
var1 var2 var5
...
var1 var2 var3....vark
******************************************************************
or in binary coding
****************************************************************
1 0 0 0 0 .....0
0 1 0 0 0 .....0
0 0 1 0 0 .....0
...
...
...
0 0 0 0 0 .....1
1 1 0 0 0 .....0
1 0 1 0 0 .....0
1 0 0 1 0 .....0
...
...
1 1 1 0 0 .....0
1 0 1 1 0 .....0
...
...
...
...
1 1 1 1 1 .....1
***********************************************
Here I didn't present some summarize command that could
grab case number for
that response pattern. But basically I will run through
each combination
and calculate the frequency for that particular combination
though there
might be zero cases. Thanks a lot
1. To get a tabulation of patterns with some instances,
