[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: e-postcard
"Clive Nicholas" <email@example.com>
Re: st: e-postcard
Mon, 7 Jul 2008 17:58:14 +0100
Johannes Geyer replied:
> if you have not thought about that, it may be useful: you could use the
> UEFA coefficient in your regression, they've calculated this measure at
> least since 1960 (http://www.xs4all.nl/~kassiesa/bert/uefa/index.html) -
> you could argue that it is a proxy for unobserved country effects
> (strength of the national league).
I thought about this shortly after firing off my posts, but you are
quite right to raise this. These are certainly worth including:
indeed, Bert Kassies has calculated these tables from the very
beginning, i.e. 1956-7!
Two caveats about the use of UEFA Coefficients:
(1) even including this variable probably does not do away with the
need for fixed-country effects in pooled TSCS models (if only they
(2) many of Europe's big leagues have lots of foreign players playing
in them, so I can envision a peer-reviewer slapping me down for saying
that this variable is a good proxy measure for the strength of a
country's national league. It _is_ if your league is in, say, Sweden,
the Czech Republic or Bulgaria. It is _not_ if it's in, say, England,
Spain or Belgium.
As an aside, another (dummy) variable that would be worth including is
whether or not a country is a member of the EU. This variable proved
to be positively significant in most of my pooled models estimating
the degree of success of teams at previous European Championships
since 1980. Still another might be whether or not a country's FA has
established a 'Centre of Excellence', but this information might be
hard to gather for all 53 countries.
Many thanks once again, and comments (back) are welcome.
[Please DO NOT mail me personally here, but at
"Courage is going from failure to failure without losing enthusiasm."
-- Winston Churchill
* For searches and help try: