Re: st: Assistance with manipulating a social network dataset?

[Nick Cox <njcoxstata@gmail.com>]

Your last question is addressed in

SJ-8-4  dm0043  . Tip 71: The problem of split identity, or how to group dyads
        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  N. J. Cox
        Q4/08   SJ 8(4):588--591                                 (no commands)
        tip on how to handle dyadic identifiers

Nick

On Tue, Oct 11, 2011 at 11:05 PM, Brandon Olszewski
<olszewski.brandon@gmail.com> wrote:

> I have a social network dataset, and I can’t figure out how to perform
> the proper manipulations. People in rows were asked if they know
> people listed in columns. In cells, “0” indicates two people don’t
> know each other, and “1” indicates otherwise. So what I have looks
> like this:
>             Adam Beth  Charlie
> Adam    1       1       0
> Beth            0       1       0
> Charlie 0       1       1
>
> Note that while Adam claims to know Beth, Beth doesn’t claim the same,
> and while Beth says she doesn’t know Charlie, he says otherwise. For
> my purposes, I want to assume that if anyone says they know someone
> else, to treat it as a “1” both ways.
>
> The software I want to use (Sonoma) wants the data in one of two
> formats. Here’s the wide option, which offers only one half the
> matrix, with “1” coded in the diagonal and “.” coded in the bottom
> half, with max values for combinations in cells:
>            Adam        Beth    Charlie
> Adam    1       1       0
> Beth           .        1       1
> Charlie .       .       1
>
> Question 1: How would I do this in Stata? I looked at -help mata-, but
> I don’t even know if that’s the right direction. Is it? If not, how
> might I do it? This option seems more difficult for me (given my
> familiarity with Stata’s functionality) than the “long option” below.
>
> Here’s the long option, which seems more feasible for me, given my
> level of skill. Note that each combination is listed just once, again
> with maximum values:
> Adam    Adam    1
> Adam    Beth            1
> Adam    Charlie 0
> Beth            Beth            1
> Beth            Charlie 1
> Charlie Charlie 1
>
> Question 2: I can get the data to long format fine no problem. But end
> up with duplicates of combinations, as Adam is asked about Ben, and
> Ben is asked about Adam (i.e. a total of 9 observations, rather than
> the six above). How could I drop duplicate combinations, saving only
> the max value for each? While I am pretty familiar with the
> -duplicates- set of commands, I’m running into the problem that I
> don’t know how to use the command since combinations go both ways,
> where Adam-Beth is a duplicate of Beth-Adam. I’ve also thought about
> it substituting numbers for people (i.e. 1-2 & 2-1), but that doesn’t
> change my problem that I can’t figure out how to tell Stata to treat
> those as duplicates.
>

*
*   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/