18
In the analysis of the 1996 Romanian election the Issue attitude variable refers to
the salience of the same opposition-promoted issues as before; to general performance
evaluation in the 2000 Romanian election, an issue which then certainly favored the
opposition campaigning on governmental incompetence; to anticommunist attitudes,
which the main government party tried to appeal to, in the 1994 Hungarian election; to a
set of welfare state issues promoted by the opposition in the 1998 Hungarian election;
and to retrospective economic evaluations - that the then government appealed to - in the
2002 Hungarian election. The coding of the variable is such that it should always be
positively related to a pro-government vote. However, a positive interaction effect of
Public television exposure and Issue attitude on Vote choice would signal pro-
governmental priming where – as in the 1994 and 2002 Hungarian elections – the issue
in question was promoted by the government, and pro-oppositional priming in all other
elections.
The relevant parameter estimates – once again omitting the effects of the control
variables and the intercepts – are shown in separate tables. For instance, Table 2 shows –
for all surveys in question - the direct effects of public television on party sympathies,
and of party sympathies on vote choice. On top of each causal arrows of interest in Figure
2, a number is printed that shows which table displays the relevant parameter estimates.
Tables 2 to 5 about here
The cross-sectional data seems to reveal much the same pattern for the 1996
Romanian and the 1994 and 1998 Hungarian elections as the panel data did. That is,
public television seems to have had a direct anti-governmental effect on vote choice in
the 1994 Hungarian election (cf. Table 5), and no significant indirect effect either through
party affect or issue attitudes (cf. Tables 2 and 3). In the 1998 Hungarian election and the
1996 Romanian election exposure to public television had no direct effect on the vote (cf.
Table 5), but an indirect and pro-governmental effect via party affect (cf. Table 2) and, in
the Romanian election, also via issue attitudes (cf. Table 3). The results regarding
priming are broadly consistent with the findings based on panel data: no significant effect
in Romania in 1996, and pro-governmental priming in both the 1994 and the 1998
Convention