Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# RE: st: RE: Comparing variables by observation

 From Nick Cox To "'statalist@hsphsun2.harvard.edu'" Subject RE: st: RE: Comparing variables by observation Date Tue, 21 Sep 2010 20:18:00 +0100

```(1) and (2) are not legal variable names.

-by obscode- will do no harm, but will make no difference to the outcome as -obscode- does not feature on the right-hand side.

As pointed out very recently,

cond(5>= X1 & X1>= X2 & X2>=X3 & X3>=4, 1, 0)

is just the same in result as

5>= X1 & X1>= X2 & X2>=X3 & X3>=4

I'd check for missings on your Xs.

[_n] is unnecessary in the last command.

Nick
n.j.cox@durham.ac.uk

Anisa Shyti

I tried the following codes:

by  obscode, sort: gen (1) = cond(5>= X1 & X1>= X2 & X2>=X3 & X3>=4, 1, 0)
by  obscode, sort: gen (2) = cond(X1<=3 & 5<= X3, 1, 0)

I also tried:
gen byte cond2 = X5t[_n] <=3 & X7t[_n] >=5

Is this approach correct? I will try what you both suggest.
Otherwise, I'll have to look for a logical problem..

On Tue, Sep 21, 2010 at 8:57 PM, Nick Cox <n.j.cox@durham.ac.uk> wrote:

> No code of yours allows no comment on that.
>
> Responding to the algebra given, this kind of problem is one in which Stata's syntax is inescapably long-winded. To save checking on precedence rules I tend to parenthesise aggressively:
>
> (5 >= X1) & (X1 >= X2) & (X2 >= X3) & (X3 >= 4)
>
> (X1 <= 3) & (5 <= X3)
>
> What you wrote is not illegal but it will be interpreted in terms of 1s and 0s as explained at length in
>
> http://www.stata.com/support/faqs/data/trueorfalse.html
>
> which will usually yield something quite different from what you want.
>
> Nick
> n.j.cox@durham.ac.uk
>
> Anisa Shyti
>
> I have three variables X1 X2 X3
> (taking values from 1 to 10) and I need to compare their values, by
> observation, according to several conditions (which are theoretically
> derived) of the form, examples: (1) 5>= X1 >=X2 >=X3 >=4; (2) X1<=3
> and 5<= X3.  I need to evaluate each condition on the data by counting
> the number of times it is met, in order to classify the observations.
>
> I tried several approaches, but the total number of classified
> observations exceeds the total number of observations.  So, I am
> guessing I have some redundant counts - maybe I am doing something
> wrong with the coding (given that the conditions are correct).

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