Effects of Measurement Error on 2SLS (PA)

The Effects of Presidential Approval on Partisanship

Nonrandom Error and 2SLS

TITLE THE EFFECTS OF ERROR ON TSLS ESTIMATES: 1980 NES WAVES 1-2
! Analysis of how party identification influences and is influenced
! by presidential approval. Data are from the January and June
! waves of the 1980 National Election Study Panel Survey.
! Party ID is measured by the traditional 7-pt scale; approval,
! by a 5-pt scale ranging from 1-5. Here, lagged approval
! and party identification serve as instruments.
!
! This command file replicates the analyses reported in
! Green, Donald Philip. 1990. The Effects of Measurement
! Error on Two-Stage, Least-Squares Estimates. In James A. Stimson,
! ed., _Political Analysis_, v.2. Ann Arbor: University of Michigan Press.
!
! Analysis 1. Assumes no error.
! Analysis 2. Assumes a measurement error variance of .4 for party ID.
! Analysis 3. Assumes a measurement error variance of .4 for all variables.
! Analysis 4. Assumes within-wave error covariance of .3
! Analysis 5. Assumes across-wave error covariance of .2
!——————————————————————
! Analysis 1: No Measurement Error
! (Table 2, scenario 1)
!——————————————————————
DA NI=4 NO=637 MA=CM
LA
*
‘PID.1’ ‘PID.2”APP.1”APP.2’
CM
*
3.9597366
3.4984005 4.1897308
.8614970 .8807105 2.3841316
.9548518 1.0040431 1.3066038 2.1886792
SE
2 4 1 3/
MO NY=4 NE=4 LY=FI,ID BE=FR,FU TE=FI,DI PS=FR,SY
PA BE
*
0 1 1 0
1 0 0 1
0 0 0 0
0 0 0 0
PA PS
*
1
1 1
0 0 1
0 0 1 1
VA 0 TE(1,1) TE(3,3)
VA 0 TE(2,2) TE(4,4)
OU TO SE ndec=3
!——————————————————————
! Analysis 2: Random Error in Party ID Alone
! (Table 2, scenario 2)
!——————————————————————
DA NI=4 NO=637 MA=CM
LA
*
‘PID.1’ ‘PID.2”APP.1”APP.2’
CM
*
3.9597366
3.4984005 4.1897308
.8614970 .8807105 2.3841316
.9548518 1.0040431 1.3066038 2.1886792
SE
2 4 1 3/
MO NY=4 NE=4 LY=FI,ID BE=FR,FU TE=FI,DI PS=FR,SY
PA BE
*
0 1 1 0
1 0 0 1
0 0 0 0
0 0 0 0
PA PS
*
1
1 1
0 0 1
0 0 1 1
VA .4 TE(3,3)
VA 0 TE(2,2) TE(4,4)
OU TO SE ndec=3
!——————————————————————
! Analysis 3: Random Error in Each Observed Variable
! (Table 2, scenario 3)
!——————————————————————
DA NI=4 NO=637 MA=CM
LA
*
‘PID.1’ ‘PID.2”APP.1”APP.2’
CM
*
3.9597366
3.4984005 4.1897308
.8614970 .8807105 2.3841316
.9548518 1.0040431 1.3066038 2.1886792
SE
2 4 1 3/
MO NY=4 NE=4 LY=FI,ID BE=FR,FU TE=FI,DI PS=FR,SY
PA BE
*
0 1 1 0
1 0 0 1
0 0 0 0
0 0 0 0
PA PS
*
1
1 1
0 0 1
0 0 1 1
VA .4 TE(1,1) TE(3,3)
VA .4 TE(2,2) TE(4,4)
OU TO SE ndec=3
!——————————————————————
! Analysis 4: Random Error and Within-Wave Nonrandom Error
! (Table 3, scenario 3)
!——————————————————————
DA NI=4 NO=637 MA=CM
LA
*
‘PID.1’ ‘PID.2”APP.1”APP.2’
CM
*
3.9597366
3.4984005 4.1897308
.8614970 .8807105 2.3841316
.9548518 1.0040431 1.3066038 2.1886792
SE
2 4 1 3/
MO NY=4 NE=4 LY=FI,ID BE=FR,FU TE=FI,sy PS=FR,SY
PA BE
*
0 1 1 0
1 0 0 1
0 0 0 0
0 0 0 0
PA PS
*
1
1 1
0 0 1
0 0 1 1
FI TE(1,1) TE(2,2) TE(3,3) TE(4,4) TE(2,1) TE(4,3)
VA .4 TE(1,1) TE(3,3)
VA .4 TE(2,2) TE(4,4)
VA .3 TE(2,1) TE(4,3)
OU TO SE ndec=3
!——————————————————————
! Analysis 5: Random Error and Across-Wave Nonrandom Error
! (Table 4, scenario 1)
!——————————————————————
DA NI=4 NO=637 MA=CM
LA
*
‘PID.1’ ‘PID.2”APP.1”APP.2’
CM
*
3.9597366
3.4984005 4.1897308
.8614970 .8807105 2.3841316
.9548518 1.0040431 1.3066038 2.1886792
SE
2 4 1 3/
MO NY=4 NE=4 LY=FI,ID BE=FR,FU TE=FI,sy PS=FR,SY
PA BE
*
0 1 1 0
1 0 0 1
0 0 0 0
0 0 0 0
PA PS
*
1
1 1
0 0 1
0 0 1 1
FI TE(1,1) TE(2,2) TE(3,3) TE(4,4) TE(3,1) TE(4,2)
VA .4 TE(1,1) TE(3,3)
VA .4 TE(2,2) TE(4,4)
VA .2 TE(3,1)