Re: st: tricky data merge/joinby problem

[David Kantor <kantor.d@att.net>]

At 10:30 AM 3/4/2011, Dimitry wrote:
> I have two files that I would like to merge. The first contains data
> on city blocks and block groups (BGs) and fraction of population
> variable. A simplified version of the data looks like this:
>
> bid bgid  fracpop
> 11  1    .5
> 12  1    .5
> 21  2    .3
> 22  2    .2
> 23  2    .5
>
> For example, BG 1 contains 2 blocks, each of which has half of the BG
> 1's population (fracpop==.5). The unique identifier in this file is
> bid.
>
> I would like to merge the data above with panel data file2 that
> contains block group populations over time. This data looks like:
>
> bgid dateyq bgpop
> 1    2010q1 100
> 1    2010q2 105
> 1    2010q3 106
> 1    2010q4 125
>
> Here bgid and dateyq are the identifiers. The final goal of merging is
> to come up with a population for each block by allocating bgpop using
> the weights in fracpop. For example, for BG 1, this would yield:
>
> bid bgid dateyq bpop
> 11  1    2010q1 50
> 12  1    2010q1 50
>
> Does this require the dreaded m:m merge with bgid as the id as the
> first step? That appears to work (although I only checked a few
> cases). Or is is better to expand the first file into a panel and then
> merge on bgid and dateyq? Or should I use -joinby bgid using
> file2.dta-? I am not sure which is the most efficient solution.

I'm not sure what you mean by m:m merge, as an actual -merge- with a 
many-to-many correspondence usually leads to meaningless pairings.
And I'm not sure what you meant by "expand ... into a panel".
But the situation does look like -joinby-.
HTH
--David

*
*   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/