19
multiplicative term on a subset of my data set where all of the hearings were held by
outlier committees only and then on the remaining non-outlier committee data. As stated
in H1, I should be able to reject the null hypothesis of no effect, the variable exhibiting a
positive influence, for the committee outlier set, but not for the non-outlier committees.
The next hypothesis, H2, deals with constraints interest group representatives are
under to stay true to the their members’ ideal positions and defy the committee, given that
an ideological difference exists and the group has been asked to testify. Therefore:
Pr(Y
Ti
=0) = |G
i
– C
m
|
[
i
(V
i
)
] |
Y
Ii
= 1
(2)
The term still refers to how strongly members of interest group i feel about their
positions on the issue, and V is the variance in the positions of these members. This latter
variable comes from a question asked of the lobbyist regarding how cohesive members
were in their positions on the issue. A closed-ended question, the response was coded 1
if members were not unified, 2 for somewhat unified, and 3 for very unified. H2 states
that these factors should only be interactive, they are multiplied together. Because
ideological divergence between this ideal position and that of the committee is also
required for Y
T
= 0, this term is then multiplied by the distance between the group’s ideal
position and the committee median. That this should only apply if the group has been
invited to testify, Y
I
= 1, will be determined by using sample selection in the statistical
model and is discussed below.
The construction of H3 is similar:
G
Pr(Y
Ti
=0) =
(G
i
– G
j
)
|
|G
i
– C
m
| 0, Y
Ii
= 1
(3)
i=1
Convention