Theme NexT works best with JavaScript enabled
1
2
3
4
5
cd "F:\General Database of Social Science\Courses\Courses Materials\波士顿学院\纵贯调查分析\数据"
use hrs_hours,clear
reshape long r@workhours80 r@poorhealth r@married r@totalpar r@siblog h@childlg r@allparhelptw, i(hhid pn) j(wave)
keep if wave < 3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
xtset hhidpn wave
*用滞后期Y解释当期Y
reg rworkhours80 l.rworkhours80
est store m1
esttab m1,nogap
----------------------------
(1)
rworkhours80
----------------------------
L.rworkho~80 0.737***
(79.38)
_cons 5.340***
(15.03)
----------------------------
N 5897
----------------------------
t statistics in parentheses
* p<0.05, ** p<0.01, *** p<0.001
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
reg rworkhours80 l.rworkhours80 l.rallparhelptw
est store m2
esttab m2,nogap
----------------------------
(1)
rworkhours80
----------------------------
L.rworkho~80 0.735***
(78.12)
L.rallparh~w -0.160*
(-2.23)
_cons 5.484***
(15.08)
----------------------------
N 5767
----------------------------
t statistics in parentheses
* p<0.05, ** p<0.01, *** p<0.001
reg rworkhours80 l.rworkhours80 l.rallparhelptw l.rpoorhealth l.rmarried l.rtotalpar l.rsiblog l.hchildlg raedyrs female age minority
est store m3
esttab m1 m2 m3,nogap
------------------------------------------------------------
(1) (2) (3)
rworkhours80 rworkhours80 rworkhours80
------------------------------------------------------------
L.rworkho~80 0.737*** 0.735*** 0.669***
(79.38) (78.12) (63.23)
L.rallparh~w -0.160* -0.0942
(-2.23) (-1.28)
L.rpoorhea~h -4.444***
(-7.46)
L.rmarried 0.421
(0.69)
L.rtotalpar 0.276
(0.95)
L.rsiblog -0.420
(-1.12)
L.hchildlg -0.522
(-1.31)
raedyrs 0.124
(1.59)
female -3.393***
(-7.35)
age -0.781***
(-10.97)
minority -0.741
(-1.39)
_cons 5.340*** 5.484*** 52.53***
(15.03) (15.08) (11.98)
------------------------------------------------------------
N 5897 5767 5457
------------------------------------------------------------
t statistics in parentheses
* p<0.05, ** p<0.01, *** p<0.001
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
drop _est_m1 _est_m2 _est_m3
reshape wide
reg rworkhours802 rworkhours801 rallparhelptw1 rpoorhealth1 rmarried1 rtotalpar1 rsiblog1 hchildlg1 age minority female raedyrs
est store m4
esttab m4,nogap
----------------------------
(1)
rworkhou~802
----------------------------
rworkhou~801 0.669***
(63.23)
rallparhel~1 -0.0942
(-1.28)
rpoorhealth1 -4.444***
(-7.46)
rmarried1 0.421
(0.69)
rtotalpar1 0.276
(0.95)
rsiblog1 -0.420
(-1.12)
hchildlg1 -0.522
(-1.31)
age -0.781***
(-10.97)
minority -0.741
(-1.39)
female -3.393***
(-7.35)
raedyrs 0.124
(1.59)
_cons 52.53***
(11.98)
----------------------------
N 5457
----------------------------
t statistics in parentheses
* p<0.05, ** p<0.01, *** p<0.001
差分方法
将因变量设置为两期差分值
差分方法一直受到批评。其中一个原因是它们被认为是不可靠的——如果时间1和时间2的因变量是正相关的(几乎总是如此) ，那么差分方法的可靠性就会低于单个时间点的可靠性，如果跨时间的相关性很高，那么可靠性的降低就会很多。
差分方法没有考虑到回归到平均效应的情况——在极低的初始分数之后，随之而来的是分数的增加和分数的降低。因此，初始水平可能会发生变化，但是如果我们将滞后因变量加入到差分方法中，我们就回到了 LDV 模型，所以这个策略是没有用的:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
reshape long
gen diff= rworkhours80-l.rworkhours80
reg diff l.rallparhelptw
est store m1
reg diff l.(rallparhelptw rpoorhealth rmarried rtotalpar rsiblog hchildlg) raedyrs female age minority
est store m2
esttab m1 m2,nogap
--------------------------------------------
(1) (2)
diff diff
--------------------------------------------
L.rallparh~w -0.0147 -0.0267
(-0.19) (-0.34)
L.rpoorhea~h 0.264
(0.42)
L.rmarried 1.384*
(2.08)
L.rtotalpar 0.0907
(0.29)
L.rsiblog -0.790
(-1.94)
L.hchildlg -0.428
(-0.99)
raedyrs -0.131
(-1.57)
female 1.381**
(2.92)
age -0.476***
(-6.22)
minority -0.578
(-1.00)
_cons -2.792*** 25.22***
(-12.15) (5.40)
--------------------------------------------
N 5767 5457
--------------------------------------------
t statistics in parentheses
* p<0.05, ** p<0.01, *** p<0.001
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
for any poorhealth married totalpar siblog allparhelptw: gen Xdiff=rX-l.rX
for any childlg: gen Xdiff=hX-l.hX
reg diff allparhelptwdiff poorhealthdiff marrieddiff totalpardiff siblogdiff childlgdiff
est store m3
reg diff allparhelptwdiff poorhealthdiff marrieddiff totalpardiff siblogdiff childlgdiff raedyrs female age minority
est store m4
esttab m3 m4,nogap
--------------------------------------------
(1) (2)
diff diff
--------------------------------------------
allparhelp~f -0.0796 -0.0779
(-1.33) (-1.31)
poorhealth~f -2.450*** -2.417***
(-3.61) (-3.56)
marrieddiff -0.890 -0.790
(-0.65) (-0.58)
totalpardiff 0.572 0.430
(1.16) (0.87)
siblogdiff -1.649 -1.740
(-0.57) (-0.60)
childlgdiff 1.416 1.101
(0.85) (0.67)
raedyrs -0.0944
(-1.18)
female 1.263**
(2.68)
age -0.454***
(-5.97)
minority -0.936
(-1.64)
_cons -2.516*** 23.36***
(-9.67) (5.28)
--------------------------------------------
N 5229 5227
--------------------------------------------
t statistics in parentheses
* p<0.05, ** p<0.01, *** p<0.001
