3.2 Data and output measurement
We use a panel dataset that contains player-game day level information on the German
Soccer League (1. Bundesliga) in the seasons 2006/07 and 2007/08.18 For each of the
216 matches in each season, we have information about all players who participated in
the match - either on the field or on the reserve bench -, and about various dimensions
of their performances. In addition, we gathered data on national team matches of the
same players between the World Cup 2006 and the Euro 2008 using publicly available
sources;19 the latter data we used exclusively for constructing a variable summarizing a
player’s past national team exposure.
We keep only those players in the dataset for whom we have observations in the
German Soccer League both before and after the official Euro 2008 qualification date in
November 2007, and in Season 06/07, and who were on the field at least once in the two
seasons. Moreover, we exclude goalkeepers as they have very different tasks than field
players, and most of the performance measures we will use are not applicable to them.
Table 1 provides an overview of the players’ nationalities. The treatment group ”Euro
2008” consists of all players whose nations participated in the Euro 2008 (except for those
of Austrian or Swiss nationality), and the control group consists of all other players. About
half the players are of German nationality, but the others originate from all over the world.
Incidentally, the German Soccer League was the best represented in the Euro 2008, with
active players in fourteen out of sixteen teams. Tables 2 and 3 present descriptive statistics
for players from participating and non-participating nations, respectively; all statistics
refer to club matches in the German League. On average Euro 2008 - Europeans are
younger, occupy midfield positions more frequently, and have lower outputs than players
in the control group. Several of these comparison would be reversed if Germans were
excluded from the sample.
We think of unobservable effort as choices such as training intensity and lifestyle
(nutrition, sleeping habits,...), as well as concentration and motivation on the soccer field.
To measure observable individual output, we rely on the following measures:20
18 The data was kindly provided by Impire, a company specialized in collecting and selling sports data.
19We relied on the following websites: ESPNsoccernet.com, FIFA.com, Kicker.de, and Worldfoot-
ball.net.
20In addition, we also have information on the fouls committed and suffered by each player. Fouls
suffered could be interpreted as a positive performance measure, the idea being that stronger players are
harder to stop. Fouls committed can be viewed as a measure of destructive effort. This is the approach
14