Mead 27
activist, civil rights, and religious organizations, do continue to raise progressive issues, but even
they pay rising attention to paternalist questions. So the changes in discourse are broad-based.
Determinants of the Agenda
As suggested above, the reasons for the agenda shift ultimately lie outside welfare politics—in
trends in national politics and the society. But some correlates of the key framings can be found
within the current data. I built models of whether witnesses mentioned progressivism or paternalism
as their leading issue, and whether they mentioned these issue types at all. Tables 4-7 show the
results (see Appendix for descriptive statistics on the variables).
Since a paternalist emphasis seems to grow over the whole span of the welfare controversy, I
interpret the stage of reform as a continuous, five-step variable. As an explainer, this term
dominates most of these models, accounting for the lion’s share of the explained variation all by
itself.
Table 4: Model of whether a witness cited progressivism as leading issue:
Coeff
Std.
error
z score
P>|z|
Effect
Stage of welfare reform:
-.554
.062
-8.94
.000
-.568
Witness was left on leading issue:
.575
.155
3.70
.000
.204
Hearing occurred in Senate:
-.403
.134
-3.01
.003
-.131
Witness was from the administration:
-.758
.241
-3.15
.002
-.289
Witness was an academic:
-1.14
.354
-3.22
.001
-.416
Constant:
2.39
N
527
Log likelihood
-255.0
Pseudo R
2
.237
Note: Effects are Clarify estimates of the change in the chance that a witness cites progressive as
leading issue resulting from a change in this variable from its lowest to highest value.
Table 4 shows that the chance a witness would adopt progressivism as his or her leading issue
was strongly depressed by the stage of reform, although somewhat raised if the witness took a left
38
These models were assembled with limited fitted and confined to terms significant at .05 or better. I
began with the stage of reform term, then added ideology and—if significant—the Senate term. I then added
such witness terms as were significant when added to the terms already in the model.
Convention