Michael, the offending letter is 'b', as opposed to 's'. The government guy tried to say that Osama (bin Laden) was responsible for these atrocities. But what he actually said was that it was (Barack) Obama's fault. Outch.

Rodrigo,
as far as I know, the WoS site is designed to make mass-downloads by script difficult, if not impossible. At the end of the day, we asked a research assistant to download records in parcels of 500 which we then fed through a series of idiosyncratic, ill designed and undocumented scripts to extract the information we were after. These scripts were tuned to the research problem of the the day and are therefore not ready for prime time (or indeed for anybody else). I intend to rewrite the whole shebang in Python (which has a much clearer syntax plus pre-built structures to represent network data) if and when I find the time.

testcomment from wp

Thanks to Ben Stanley for pointing me towards to other links that provide a chapter-length treatment of the first problem: http://www.nd.edu/~rwilliam/xsoc73994/Margins01.p... and http://www.nd.edu/~rwilliam/xsoc73994/Margins02.p...

Thanks for your interest Teun. The book (edited by Uwe Backes and Patrick Moreau) will be published later this year with Vandenhoeck & Ruprecht (Göttingen). For the time being, it's probably best to cite the manuscript as a working paper, along with its full URL.

On second thought, Jacob Christensen is right when he argues that one should probably devote more attention to the regional elections in North Rhine-Westphalia (next Sunday) and be slightly less excited about the UK General Election:http://jacobchristensen.name/2010/05/04/nrw-vs-gb...

Then again, as Jacob puts it, the "UK is to Europe what New York is to the US", which certainly makes Westminster the Washington of the hearts.

Thanks for that Alexander. I tweaked an existing example to my needs, but your solution is obviously much more concise!

Thanks for your comment Gabriel - I wasn't aware of the end-of-line problem. Your code is very neat and useful but seems to assume that the data are already in a "long" format (one record per dyad). My code is really just a hack wrapped around reshape, which saves me the trouble of doing this in Perl.

For the benefit of other readers, here is a link to Gabriel's code on ssc:
http://ideas.repec.org/c/boc/bocode/s457075.html

And here is the link to the spiced-up version on his blog (the link in the original comment points back to this page): http://codeandculture.wordpress.com/2009/12/10/st...

Well, not exactly videos but at least a picture: http://www.kai-arzheimer.com/blog/2009/12/22/radi...

Hm, I assume that means you want to buy MLwiN?!? You should get in touch with the Centre for Multilevel Modelling at Bristol: http://www.cmm.bristol.ac.uk/