Altruism with Social Roots: An Emerging Literature

DESARROLLO Y SO

SEGUNDO SEMESTRE DE 2006, PP. 245-260.

Table 3 presents summary statistics for the regression variables.

Using these three variables we estimate the impact of social proximity,
social integration and reciprocity on giving. As expected, there are
high correlations between variables (ρ_pi(f)pi(j) = 0.69, ρ_pi(f)fi = -0.34 &
ρ_fipi(j)= -0.32) which may indicate multicollinearity.

Table 4 presents the regression results. Columns 1 to 3 explore each
variable separately. The effect of degree-out on giving is marginal
and both social proximity and reciprocity are highly significant. Also
note that the latter has a coefficient which is twice as high as that of
the former.

To explore in depth the effects of reciprocity and social integration
we use p_i( j), a variable which assigns value zero to those individuals
who knew they were never going to be matched to a friend and the
inverse of degree-out (1/f_i) for subjects facing friends. Given the
multicollinearity problem we must study social proximity and reci-
procity separately. Column 4 jointly explores social proximity and
integration and column 5 uses social integration and reciprocity. The
results are clear: the model presented in column 5 is highly explanatory.

Furthermore, the model presented in column 6 (with high
multicollinearity) suggests that social proximity offers no additional
predictive power once social integration and reciprocity are included
in the regression. In sum, regressions 5 & 6 show that p_i( j) has a
positive and significant coefficient but p_i( f) has no effect when both f_i
and p_i( j) have been introduced. So there is no friendship effect on
giving once degree-out and reciprocity have been taken into account.

¹⁹ For subjects playing with strangers and friends respectively.

<<   <   8   9   10   11   12   13   14   >   >>