Matthew Lebo

1

• Matthew Lebo
• Stony Brook University
• ITV Seminar
• Ohio State University
• Friday April 28, 2006

2
Strategic Party Government Party Influence in
Congress 1789-2000

• Matthew Lebo
• Stony Brook University
• Adam J. McGlynn
• Stony Brook University
• Gregory Koger
• University of Montana

3
When are parties in Congress influential?

• There is a great deal of historical variation in
  the influence of parties.
• Sometimes parties are weak and seem to fit
  Krehbiels Wheres the Party description.
• Sometimes they are strong enough to coerce
  members to cast votes that endanger their
  reelection.

4
Why does the influence of Congressional parties
vary?

• Prevailing answer is Conditional Party Government
  (CPG).
• Ideology or preferences is the key.
• A party is strengthened when members ideologies
  become more aligned.
• Or, when the ideological distance between the
  parties grows.

5
Conditional Party Government

• Congressional parties should be most influential
  when two internally cohesive parties disagree on
  a wide range of issues.
• Members will then delegate more authority to
  party leaders.
• Party influence varies across issues and contexts
  (Froman and Ripley 1965), over time (Cooper and
  Brady 1981), or both (Rohde 1991).

6
WEAK
7
Getting Stronger...
8
Stronger...
9
STRONG
10
STRONGER
WEAKER
11
Some problems with the CPG model

• What are party labels beyond ideological groups?
• If this is all there is to parties, would we
  expect parties to mean anything to legislators?
• What is the interplay between influence and
  elections?
• Parties do not operate in a vacuum what about
  the competition between the parties?

12

• Key criticism of CPG by Krehbiel If party
  leaders are only active on a subset of issues
  that already unite party members, what marginal
  effect do they have on policy outcomes?
• Cox and McCubbins (1993, 2005) stress that
  parties are most interested in electoral victory.
  
• Parties work together to enhance their reputation
  and their common label.
• Strategic Party Government build on both the CPG
  model and the Cox and McCubbins focus on
  elections.

13
Strategic Party Government

• Actors Legislative Parties
• Ultimate Goal Maximize seat share
• Proximate Goal Win important votes.
• Challenge Choose party strength and effort to
  maximize legislative victories. 
• But Dont be too strong!

14

• Parties try to win elections by winning
  legislative contests.
• This is an assumption we test.
• Parties that lose votes look bad for several
  reasons.
• So, to win legislative contests, parties engage
  in multiple activities (e.g. vote-buying, agenda
  shaping)
• This is the untested assumption of our model.
• These attempts by parties to increase their
  influence are costly.
• Why?

15

• 1. Opportunity cost effort spent to win one
  contest makes winning other contests more
  difficult. Also detracts from other areas of
  legislating and dealing with constituents.
• 2. As legislators delegate more power to their
  leaders, they lose the ability for
  credit-claiming.

16

• 3. Most importantly, we should expect that a
  legislators default voting position is in line
  with her constituents. Being pulled away from
  that vote means being pulled away from her
  constituents. Come election time, this could cost
  her reelection.
• (Carson (2005) for example, finds that increased
  party unity increases the probability that a
  quality challenger will emerge.)

17
Thus, we build upon Conditional Party Government
with
STRATEGIC
Party Government

• Parties want electoral successes.
• To do this, they want to win legislative
  contests.
• But, they must be mindful, coercing stronger
  party unity can pull a legislator from her
  district and can be costly come election time.

18

• So, parties want to be strong, but not for its
  own sake just strong enough to win.
• And what determines how strong they need to be to
  win?
•  
• The opposition its relative size and strength.

19
What we do

• Test 2 major aspects of SPG alongside CPG
• 1. That a (or the) major determinant of party
  influence is the strategic interaction of the two
  major parties over time.
• 2. That increased party influence can be
  electorally damaging.

20
Hypotheses

• SPG
• Party influence will be determined by the level
  of influence of the other party.
• This is so strong that the two parties level
  of unity will be in a long-run equilibrium.
• As party size increases, party influence will
  decrease.
• Winning legislative contests will improve a
  partys electoral fortunes.
• Increasing party influence will hurt a partys
  electoral fortunes.

21
Hypotheses

• CPG 
• Party influence will increase as the two
  parties diverge ideologically.
• Party influence will increase as a party is
  more ideologically cohesive.

22
How we do it

• Using Congressional Roll Call and election data
  from 1789-2000.
• Doing so for both the House and the Senate.
• Applying dynamic modeling techniques to SPG as
  well as CPG, a dynamic theory that has never been
  tested as such.

23
Key findings

• Party Unity
• - is directly linked to winning votes.
• - varies strongly with opposition unity.
• - is in a long-term equilibrium relationship
  between the parties (fractionally
  cointegrated)
• - all the above is true in both chambers and
  when controlling for ideology and issue agenda
  effects.

24
Key findings II

• Parties gain seats as the percentage of party
  votes they win increases.
• - House and Senate
• Parties lose seats as their level of party voting
  unity increases.
• - House and Senate since the 17th Amendment

25
In sum

• Influence as a double-edged sword
•       Winning votes helps electorally but
•       Party Unity costs seats
• Central Claim
•     A both a relative surplus of unity and a
  relative deficit are costly, parties seek to
  match each others effort.
• Auxiliary Claim
•       Activity level also varies with party
  size. The larger the party, the easier it is to
  win without effort.

26
Measures

• Influence Yearly mean of party members unity
  scores the of times they vote with their
  party on party votes.
• Sources of Unity Preferences, plotting strategy
  together, agenda manipulation, media relations,
  vote-buying/arm-twisting.

27
Two sets of models

• First, the dependent variable is voting cohesion.
• Second, the dependent variable is size of
  delegation.

28
(No Transcript)
29
(No Transcript)
30
But wait, Democrats and Republicans in 1789?

• Democrats are opponents of the Washington
  administration, the (Jeffersonian) Republican
  party, supporters of Andrew Jackson, and members
  labeled Democrat or Independent Democrat,
  using Kenneth Martis (1989) coding.
• Republicans are supporters of the Washington
  Administration, Federalists, pro-Adams and
  anti-Jackson factions, Whigs, and members labeled
  Republicans and Independent Republicans by
  Martis.

31
CPG Variables

• Ideological cohesion within parties and
  ideological distance between the parties.
• DW-NOMINATE scores in two dimensions.
• The 1st dimension is liberal-conservative which
  covers most votes.
• The 2nd deals mostly with slavery in the 19th
  century North/South issues generally.
• Standard deviation of each NOMINATE dimension for
  each party gives us ideological cohesion for
  each.
• Absolute difference between the scores of the
  median members of each major party gives
  inter-party differences.

32
(No Transcript)
33
Time Series Properties

• We want to be sure our variables are stationary.
  That is, do they return to a constant mean or do
  they exhibit strong persistence?
• Good reason to expect that they possess elements
  of both and are fractionally integrated.

34

• Grangers aggregation theorem
• Yj,t aj Yj,t-1 et where aj ß(0,1) and ejt
  N(0,s2)
• At time t each legislative seat j will rely on
  its previous value to a different extent,
  depending upon aj
• Such a series will be composed of long, but not
  perfect memory.
• This makes perfect sense for us since the
  dynamics of each seat may be unique.
• For some seats, the legislator changes. For
  others, behavior may evolve differently over
  time.

35

• ARFIMA models defined as
• So, we expect tests of the fractional
  differencing parameter, d, should be between 0
  and 1.
• We do find this for our unity series as well as
  our ideological series.To avoid threats to
  inference, we need to use fractional differencing
  for all of our variables.
• That is, difference each by its own value of d
  before estimating our models.

36

• We do not just want to estimate the coefficient
  between these two.
• Remember this hypothesis
• This is so strong that the two parties levels
  of unity will be in a long-run equilibrium.
• Equilibrium suggests that any random shock that
  separates the series will be short-lived.
• The series will return to equilibrium through
  error correction.
• Take Yt a ß Xt et
• Measure effect of ECMt-1 Yt-1 - a ß Xt-1

37

• This is how me measure Cointegration two or
  more non-stationary series can be combined to
  create a stationary one.
• Fractional cointegration relaxes this so that
  cointegration exists if some combination of the
  variables has simply a lower level of
  integration.
• Cointegration is a useful way to study
  action-reaction relationships.

38

• So we are interested in whether Republican unity
  will affect Democratic unity contemporaneously
  and whether movements away from each other are
  short-lived due to error correction.
• Test of Cointegration show that Republican unity
  and Democrat unity are cointegrated in each of
  the House and Senate.
• Thus, error-correction mechanisms are appropriate
  in the models that follow.
• Here come the tables

39
Table 1 Strategic Party Voting in the House
1789-2000 ARFIMA Model of Yearly
Data_______ Independent I.V. Differenced
by Coefficient p Variable (1-B)d , d
(s.e.)________________________
Constant 0.585 .223 (0.479) Republic
an Unity 0.78 0.510 .000 (0.062) FECM
0.54 -0.328 .000 (DemUnity
RepUnity)t-1 (0.065) Democratic
Size 0.75 -0.311 .000 (0.086) Democra
tic Majority 1.00 -0.221 .893 (1.650) Id
eological Distance N. 1st 1.10 -2.614 .816
(11.204) Ideological Distance N. 2nd
0.69 -9.909 .978 (4.883) Dem.
Ideological Cohesion 1.12 -32.971 .115 NOMINA
TE 1 (27.443) Dem. Ideological Cohesion
0.86 -37.868 .022 NOMINATE
2 (18.645) Dem. Ideological Cohesion
0.86 -73.187 .000 NOMINATE
2t-1 (16.537) Democrat Unityt-2 0.69 0.166 .
001 (0.054) ________________________
____ Durbin Watson Statistic 1.92 Centered R2
0.47 N209 Significant at .05 level,
Significant at .01 level, Significant at .001
level (all one-tailed tests).
40
Table 2 Strategic Party Voting in the Senate
1789-2000 ARFIMA Model of Yearly
Data_____________ Independent I.V. Differenced
by Coefficient p Variable (1-B)d, d
(s.e.) ________________________ Constant 0.615
.191 (0.469) Republican
Unity 0.70 0.337 .000 (0.056) FECM
0.43 -0.539 .000 (DemUnity
RepUnity)t-1 (0.100) Democratic
Size 0.91 -0.370 .000 (0.083) Democra
tic Majority 1.00 5.715 .000 (1.603)
Ideological Distance N. 1st 0.95 7.058 .185
(7.859) Ideological Distance N. 1stt-2
0.95 20.695 .005 (7.870) Ideological
Distance N. 2nd 0.80 2.246 .300 (4.278)
Dem. Ideological Cohesion 1.12 0.256 .510
NOMINATE 1 (21.211) Dem. Ideological Cohesion
1.12 -54.423 .003 NOMINATE
1t-3 (19.886) Dem. Ideological Cohesion
0.81 -48.456 .000 NOMINATE
2 (12.020) Rep. Ideological Cohesion
0.77 22.732 .013 NOMINATE
2 (10.186) Democrat Unity t-1 0.69 0.148 .042
(0.085) Democrat Unity
t-2 0.69 0.129 .013 (0.057)
Durbin Watson Statistic2.09, Centered R2 0.47,
N208 Significant at .05 level, Significant
at .01 level, Significant at .001 level (all
one-tailed tests).
41
What do they tell us?

• Relationship between the Parties is the dominant
  explanation for changes in unity over time.
• What does the ECM tell us?
• Democratic size is very significant also.
• CPG variables are OK at times, not great.
• Any questions at this point?

42

• So, parties exert more or less influence on their
  members in order to compete with their
  opposition.
• Why?
• Because winning votes helps them come election
  time.
• So why dont they try for perfect unity?
• Because if would pull legislators away from their
  districts.
• Really?

43
Table 3 The Electoral Effects of Party Unity and
Party Success on Democratic Chamber Share U.S.
House of Representatives_________________________
__________________________________
  Independent Coefficient (s.e.) p Variable
  Constant -1.992 (1.731) .253   Dem
ocratic Party Unity in Previous
Congress -0.245 (0.110) .014   Democratic Win
Rate in Previous Congress 7.500 (4.176) .038   M
idterm Election with GOP President 9.984 (2.474)
.000   Midterm Election with Dem.
President -5.760 (2.296) .007   Presidential
Election with Democratic Win 5.016 (2.348) .018
  1822 Election -33.425 (8.720) .000
The dependent variable is the percentage of
the House won by Democrats (differenced by
0.54). Significant at .05 level,
Significant at .01 level, Significant at
.001 level (all one-tailed tests).   Durbin
Watson Statistic 1.91 Centered R2 0.35 N104
44
Table 4 The Electoral Effects of Party Unity and
Party Success ARFIMA Models of the U.S.
Senate Independent 1st 66th
Congress Variable Coefficient (s.e.) p_____
  Constant -2.343 (2.708)
.391 Democratic Party Unity 0.011 (0.140)
.935 in Previous Congress  Democratic Win Rate
23.015 (6.680) .001 in Previous
Congress  Midterm Election 5.646 (3.895)
.076 with GOP President  Midterm Election
-5.480 (3.355) .054 with Dem.
President  Presidential Election
5.536 (3.572) .063 with Democratic
Win  Democratic Sizet-3 0.081 (0.121) .502
__________________________________________________
_____________________________   Durbin Watson
Statistic 2.16 Centered R2 0.23 N 62
The dependent variable is the percentage of the
Senate won by Democrats (differenced by 0.78).
Significant at .05 level, Significant at .01
level, Significant at .001 level (all
one-tailed tests).
45
Table 4 The Electoral Effects of Party Unity and
Party Success ARFIMA Models of the U.S.
Senate Independent 67th 106th
Congress Variable Coefficient (s.e.) p_____
  Constant -4.222 (1.591) .012
Democratic Party Unity -0.346 (0.163)
.021 in Previous Congress  Democratic Win
Rate 10.837 (3.897) .004 in Previous
Congress  Midterm Election 8.880 (2.311)
.000 with GOP President  Midterm Election
-8.072 (2.388) .001 with Dem.
President  Presidential Election
9.713 (2.395) .000 with Democratic
Win  Democratic Sizet-3 -0.317 (0.128)
.018 ___________________________________________
____________________________________   Durbin
Watson Statistic 1.81 Centered R2
0.55 N 40 The dependent variable is the
percentage of the Senate won by Democrats
(differenced by 0.78). Significant at .05
level, Significant at .01 level,
Significant at .001 level (all one-tailed tests).
46
Table 4 The Electoral Effects of Party Unity and
Party Success ARFIMA Models of the U.S.
Senate Independent 72nd 106th
Congress Variable Coefficient (s.e.) p_____
  Constant -2.185 (1.928) .267
Democratic Party Unity -0.593 (0.193)
.002 in Previous Congress  Democratic Win
Rate 14.347 (4.284) .001 in Previous
Congress  Midterm Election 6.862 (2.597)
.007 with GOP President  Midterm Election
-8.081 (2.369) .001 with Dem.
President  Presidential Election
7.723 (2.671) .004 with Democratic
Win  Democratic Sizet-3 -0.350 (0.137)
.008 ___________________________________________
____________________________________   Durbin
Watson Statistic 1.90 Centered R2
0.56 N 35 The dependent variable is the
percentage of the Senate won by Democrats
(differenced by 0.78). Significant at .05
level, Significant at .01 level,
Significant at .001 level (all one-tailed tests).
47
Electoral Consequences of Winning Votes

• In the House For each additional 10 of the
  votes the Democrats win, they will win a 0.75
  greater share (3.4 seats) in the next Congress
  (/- 0.69, plt.05).
• In the Senate
• From 1931 to 2000 a 10 increase in the portion
  of party votes won by Democrats is followed by a
  1.43 gain in seat share.
• In the indirectly elected Senate a 10 increase
  in win rate associated with a 2.3 increase in
  seat share.
• So, the importance is declining with the
  introduction of the 17th Amendment.

48
Effects of Party Influence

• House
• Each additional point of Unity costs the
  Democrats just under 0.25 in their share of the
  House (roughly 1.13 seats out of 435) in the next
  Congress.
• Senate
• For the period 1931 to 2000, a 1 increase in
  party unity is associated with a .593 decrease
  in chamber seat share (i.e. .593 seats in a
  100-seat Senate).
• And of course, party unity helps win legislative
  contests

49
Table 5 Predictors of Democratic Defeats on
Party Votes Pooled Logit Models by Party
Status, House Senate
SENATE HOUSE Independent
Democratic Majority
Democratic Minority
Democratic Majority Democratic
Minority Variable
Coefficient (SE)
Coefficient (SE)
Coefficient (SE) Coefficient (SE)

(z) (z) (z)
(z)
_________ Constant 19.91
(2.37) 12.65 (1.86) 22.29 (3.14)
11.15 (1.59) 8.39 6.79
7.09 7.02 Democrat
Unity -17.21 (1.46) -10.54 (1.57) -21.53 (2.63)
-12.63 (1.46) -11.82
-6.77 -8.18
-8.67 Republican Unity 13.63 (1.30) 14.90
(1.62) 15.36 (1.67) 16.44
(1.97) 10.46 9.19
9.18 8.33 Democratic
Size -0.35 (0.04) -0.29 (0.04) -0.38 (0.05)
-0.24 (0.04) -8.02
-7.84 -7.07 -6.21 Democratic
President -0.97 (0.53) 0.37 (0.43) 0.01
(0.29) -0.71 (0.61)
-1.81 0.86 0.05 -1.16
_______________________________________________
______________________ N 13013 11347 15245 10
198 Prob gt .00 .00 .00 .00 Pseudo
0.72 0.67 0.77 0.67 Log Pseudo-Likelihood -
2315.12 -2069.45 -2213.56 -1926.73  
Significant at .05 level, Significant at .01
level, Significant at .001 level (all
one-tailed tests). Robust standard errors are
clustered by Congress.
50
Conclusions

• Winning votes helps parties electorally.
•  
• But, doing so is costly if it requires parties to
  exert influence.
•  
• This keeps parties level of unity in a long-run
  equilibrium.
• The behavior of the parties is not merely
  conditional on their ideology, it is strategic.

51
Next steps

•  
• Move to the individual-level.
• Midwest 2006 with Jamie Carson and Greg Koger.

52

• Table 6. House Incumbents Share of the Two-Party
  Vote, 1956-2004
• 1972-2004
  1956-2004
• Coefficient Standardized
  Coefficient Standardized
• (robust s.e.) Coefficient
  (robust s.e.) Coefficient
• Ideological extremism 0.79 (2.685)
  .015 -1.365 (2.122) -.025
• District partisanship 0.302(0.061)
  .379 0.435 (0.052)
  .531
• Quality challenger -3.493(0.535)
  -.160 -5.563 (0.449)
  -.237
• Spending gap -1.986(0.340) -.396
  --- ---
• Freshman -0.976 (0.319) -.054
  -1.695(0.317) -.086
• Presidential approval -.002 (.065)
  -.002 0.004 (0.058)
  .004
• Midterm election -2.458 (1.351) -.180
  -3.140(1.314)
  -.214
• In party -2.191 (1.342) -.119
  -2.972(1.153) -.152
• Party unity -0.103(0.024) -.168
  -0.072(0.024) -.112
• Constant 55.159(3.511) 50.348(3.036)
• N 4843 7802

53
Does the nature of the equilibrium relationship
change over time?

• APSA 2006 with Greg Koger.
•  
• ECM model
•  
•  
• Assumption is a constant.
•  
• It captures the speed of error correction with a
  single value.

54
But, what if we allow it to vary over time?

• Adapting the Dynamic Conditional Correlation
  model of Engle (2002) to the ECM case gives us
  Dynamic Error Correction.

55
In RATSStart with tests of d and cointegration

• House
• difference hdemco1 / hdemco1d
• _at_rgser hdemco1d
• dhat -0.31000 ? 1 0.69
• dse 0.05893
• difference hrepco1 / hrepco1d
• _at_rgser hrepco1d
•  
• dhat -0.22000 ? 1 0.78
• dse 0.05893

56

• Senate
• difference demco1 / demco1d
• _at_rgser demco1d
•  
• dhat -0.31000 ? 1 0.69
• dse 0.05893
•  
• difference repco1 / repco1d
• _at_rgser repco1d
•  
• dhat -0.30000 ? 1 0.70
• dse 0.05893

57

• House
• Democratic Unity d 0.69
• Republican Unity d 0.78
• Democratic Size d0.75
• linreg hdemco1 / ecmh
• constant hrepco1 hdemsize

58

• Annual Data From 178901 To 200001
• Variable Coeff Std Error
  T-Stat Signif
• 
  
• 1. Constant 43.55239662 4.94289521
  8.81111 0.00000000
• 2. HREPCO1 0.65426744 0.05861964
  11.16123 0.00000000
• 3. HDEMSIZE -0.46762610 0.05428008
  -8.61506 0.00000000
•  
• difference ecmh / ecmhd
• _at_rgser ecmhd
•   gives
• dhat -0.46000 ? 1 .54

59

• Senate
• Democratic Unity d 0.69
• Republican Unity d 0.70
• Democratic Size d0.91
•  
• linreg demco1 / ecms
• constant repco1 demsize

60

• Annual Data From 178901 To 200001
• Variable Coeff Std Error
  T-Stat Signif
• 
  
• 1. Constant 71.22988127 3.55581914
  20.03192 0.00000000
• 2. REPCO1 0.38390765 0.05286073
  7.26262 0.00000000
• 3. DEMSIZE -0.67178568 0.04661296
  -14.41199 0.00000000
•  
• difference ecms / ecmsd
• _at_rgser ecmsd
•  
• gives
• dhat -0.57000 ? 1 .43

61

• So, evidence of cointegration in both the House
  and Senate. Stronger in the Senate.
•  
• Next, estimate and then adjust for level of
  fractional integration for each variable in the
  model.
• Tests of stationarity and estimates of d.

62
Table S1 Yearly Data Descriptive Statistics and
Summary of Stationarity Diagnostic Tests -HOUSE
63
Table S1 Yearly Data Descriptive Statistics and
Summary of Stationarity Diagnostic Tests -SENATE
64
Table S2 Estimates of d obtained from (0,d,0)
models - HOUSE
65
Table S2 Estimates of d obtained from (0,d,0)
models - SENATE
66
In RATS

• The RGSER procedure estimates Robinsons d.
• And, the FIF procedure fractionally differences
  the variable.
• For example
• difference HMEDMEDN2 / HMEDMEDN2d
• _at_rgser HMEDMEDN2d
• dhat -0.31000
• dse 0.05893
• _at_fif(d-.31) hmedmedn2d / hmedmedn2df
• This gives us a variable that is
  level-stationary and can be included safely in
  our models.

67
This is done for all of our variablesIn the
House

• _at_fif(d-.31) hdemco1d / hdemco1df
• _at_fif(d-.22) hrepco1d / hrepco1df
• _at_fif(d-.25) hdemsized / hdemsizedf
• _at_fif(d.12) hdemsn1sd / hdemsn1sdf
• _at_fif(d-.04) hdemsn1md / hdemsn1mdf
• _at_fif(d-.14) hdemsn2sd / hdemsn2sdf
• _at_fif(d-.22) hdemsn2md / hdemsn2mdf
• _at_fif(d-.09) hrepsn1sd / hrepsn1sdf
• _at_fif(d.21) hrepsn1md / hrepsn1mdf
• _at_fif(d-.17) hrepsn2sd / hrepsn2sdf
• _at_fif(d.00) hrepsn2md / hrepsn2mdf
• _at_fif(d.1) hmedmedn1d / hmedmedn1df
• _at_fif(d-.31) hmedmedn2d / hmedmedn2df

68
Including the ECM

• The ECM is not level stationary and including it
  is problematic.
• So, fractionally difference it to create an FECM.
• linreg hdemco1 / hecm
• constant hrepco1 hdemsize
• difference hecm / hecmd
• _at_rgser hecmd
• _at_fif(d-.46) hecmd / hecmdf

69
PARTY UNITY MODEL IN THE SENATE Complete
Model linreg demco1df / res constant
demco1df1 demco1df2 repco1df ecmdf1
demsizedf demmajd medmedn1df,
medmedn1df2 medmedn2df demsn1sdf
demsn1sdf3 demsn2sdf repsn2sdf   Linear
Regression - Estimation by Least
Squares Dependent Variable DEMCO1DF Annual Data
From 179301 To 200001 Usable Observations
208 Degrees of Freedom 194 Centered R2
0.470697 R Bar 2 0.435228 Uncentered
R2 0.472670 T x R2 98.315 Mean of
Dependent Variable 0.5266076200 Std Error of
Dependent Variable 8.6295549872 Standard Error of
Estimate 6.4852247086 Sum of Squared
Residuals 8159.2790671 Regression
F(13,194) 13.2707 Significance
Level of F 0.00000000 Log Likelihood
-676.75401 Durbin-Watson
Statistic 2.089672   Variable
Coeff Std Error T-Stat
Signif
1.
Constant 0.61470113
0.46867125 1.31158 0.19121102 2.
DEMCO1DF1 0.14802383
0.08531191 1.73509 0.08431297 3.
DEMCO1DF2 0.12920104
0.05742509 2.24991 0.02557796 4. REPCO1DF
0.33742456 0.05642845
5.97969 0.00000001 5. ECMDF1
-0.53858521 0.09978027 -5.39771
0.00000020 6. DEMSIZEDF
-0.37038971 0.08324465 -4.44941
0.00001449 7. DEMMAJD
5.71546419 1.60297959 3.56553
0.00045729 8. MEDMEDN1DF
7.05784119 7.85931291 0.89802
0.37028662 9. MEDMEDN1DF2
20.69453654 7.86994256 2.62957
0.00923499 10. MEDMEDN2DF
2.24612639 4.27792438 0.52505
0.60014759 11. DEMSN1SDF
0.25611600 21.21053778 0.01207
0.99037823 12. DEMSN1SDF3
-54.42325306 19.88575690 -2.73680
0.00678047 13. DEMSN2SDF
-48.45603348 12.02019459 -4.03122
0.00007968 14. REPSN2SDF
22.73236731 10.18608517 2.23171
0.02677857 ---------------------------------------
-------------------------------
70
Table 3 SUR ARFIMA Models of Agenda Hypothesis
Yearly Data in the House Senate 1789-2000
Independent House Model Senate
Model Variable Coefficient p Coefficient
p Constant 0.657 .159 0.648
.151 Republican Unity 0.447 .000 0.314
.000 (DemUnity RepUnity)t-1
(ECM) -0.332 .000 -0.517 .000 Democra
tic Size -0.312 .000 -0.373
.000 Democratic Majority -1.02 .505 3.862
.004 Ideological Distance N. 1st -4.381
.675 8.049 .136 Ideological Distance N. 1st
t-2 -- -- 23.71 .000 Ideological
Distance N. 2nd -7.472 .097 1.75 .326 Dem.
Ideological Cohesion NOMINATE 1 -41.441
.053 -1.147 .476 Dem. Ideological Cohesion
NOMINATE 1 t-3 -- -- -33.604
.031 Dem. Ideological Cohesion NOMINATE
2 -30.452 .039 -46.29 .000 Dem.
Ideological Cohesion NOMINATE 2t-1 -71.420 .000
-- -- Rep. Ideological Cohesion
NOMINATE 2 -- -- 20.60 .026 Democrat
Unityt-1 -- -- 0.126 .104 Democrat
Unityt-2 0.142 .005 0.078 .140
________________________________________________
__________________________________________________
_