others by about .1 at the Union level, although this is true at the country
level only for Germany, Finland and Italy. This might suggest that the
unemployed were more worried about the labor market effect of immigration
than the rest. This finding is consistent with O’Rourke and Sinnott (2006),
but Bauer et al. (2000) found that the status of being unemployed was not
an important determinant.
The exposure to the media on current affairs and politics seems to re-
duce the probability of preferring immigration restriction: the sign of the
estimated marginal effect of media is significantly negative at the Union
level. At the country level, it is significantly negative for Germany, Den-
mark, France and the UK. For instance, an additional hour spent on the
media is related to a decrease in the probability of being restrictive by about
.03 in France and the UK. The reason for the negative effect might be be-
cause longer exposure to the media on current affairs and politics increases
the chance of knowing various perspectives on the subject, resulting in less
biased opinions.
The relative income position does not seem to be an important deter-
minant of individual attitudes toward immigration: the estimated marginal
effect of relinc is significantly negative only for Portugal but at 10 percent.
We find that more educated persons are less likely to prefer immigration
restriction except for Spain and Portugal where the effect of education is not
statistically significant. A higher level of education seems to make people
more liberal. This confirms the findings by Hainmueller and Hiscox (2007).24
24 Scheve and Slaughter (2001) and Mayda (2006) show that this implies native workers’
concern with the labor market impact of immigration, consistent with the H-O model.
17