DESARROLLO Y SO
CIEDAD
SEGUNDO SEMESTRE DE 2006, PP. 245-260.
tor’s personal features (such us popularity, shyness, race, height, etc.)
and conclude that social distance variables are much more important
than personal variables12. In short,
R1”: Controlling for personal features, the shorter the social distance
the greater the giving. Social proximity has higher predictive power
than dictators’ personal attributes.
Branas-Garza, Cobo-Reyes, Espinosa, Jiménez and Ponti (2006)
(BCEJP) explore the effect of social integration, measured through sev-
eral network indexes, on dictator giving. This paper elicits a social net-
work within a group of students pursuing a BA in Economics (see Table
1) which is later used to analyze altruism in terms of dictator giving13.
The matching mechanism is controlled in order to create two specific
environments: subjects may face either friends or strangers (see Table 2)
but even in the first case subjects do not know the identity of the recipi-
ent, that is, pi(j) is positive but not equal to one14. The elicitation of the
latent social network allows to measure social integration using standard
network indexes -see Vega-Redondo (2005): clustering15, degree16 and
betweenness17 (a centrality measure). This paper obtains a salient result:
R2”: Social integration (as measured by betweenness) is positively
related to giving.
This result is relevant because it separates the pure effect of social
integration from the effect of reciprocity and the friendship effect. It
shows that with no potential gains in terms of reciprocity, more inte-
grated individuals are willing to cooperate more. In short, social inte-
gration seems to have a positive effect on giving which is independent
12 GMMTY also explore the importance of personal variables for network formation.
13 Branas-Garza, Cobo-Reyes, Jiménez and Ponti (2006) (BCJP) provide a detailed discussion
about experimental procedures for network elicitation. They compare simple devices, such
as surveys, with other more sophisticated ones such as the benefit-your-friend device,
reward+punishment coordination games, etc.
14 An exception is an individual who reveals that everyone else is his/her friend: fi = n - 1 then
pi(f ) = 1 but pi( j) takes the minimum value 1/(n - 1).
15 A measure of the likelihood that two friends of the subject in a node are friends of each other.
16 The count of the number of ties to other agents in the network.
17 The number of people who a person is connected to indirectly through their direct links.
253