the effect of unemploy remains significantly positive at the Union level. The
marginal effect of isb is still significantly positive for Greece and Luxembourg,
but no longer significant for Germany. Instead it is significantly positive for
Austria. It is now significantly negative for Ireland, but no longer significant
for Finland and the UK. The result is thus mixed. The marginal effect of
relinc is significantly positive for Portugal and Sweden. Note that it was
negative only for Portugal in Table 4.
As for the effect of education, we still find that a higher attainment is
related to a lower probability of being anti-immigration. However, we also
note that educational attainment is not important for more countries than
in Table 4, namely, Austria, Spain, Ireland, Italy, Luxembourg and Sweden.
We find that the exposure to the media is not an important determinant at
all in this subsample, while its marginal effect was significantly negative at
the Union level as well as for some countries in Table 4. The effect of having
immigrant friends continues to be generally negative, and the magnitude
is increasing in the number of such friends. But we also note that it is
not significant for more countries than in Table 4, i.e., Greece, Luxembourg,
Sweden and the UK. Surprisingly, the effect of racist is significantly negative
for Spain (i.e., the more racist a respondent was, the less restrictive he or she
was in Spain), although it remains generally positive for others.
The effect of citizen is no longer significant at the Union level. Neither
for Ireland. It becomes significantly positive for Luxembourg, while it is not
only significant but also negative for Denmark. The effect of belonging to
an ethnic minority group in the country now seems negative, while it was
mixed signwise in Table 4. The effect of having at least one parent born
23