Download as txt, pdf, or txt
Download as txt, pdf, or txt
You are on page 1of 5

Problem set 3, Palumbo Matteo Gerardo (168911)

. reg narr86 black, r


1)
Linear regression

Number of obs
F( 1, 2723)
Prob > F
R-squared
Root MSE

=
2725
= 35.37
= 0.0000
= 0.0223
= .8496

-----------------------------------------------------------------------------|
Robust
narr86 |
Coef. Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------black | .3488323
.058657
5.95 0.000
.2338156
.4638489
_cons | .3482065 .0160955
21.63 0.000
.3166458
.3797671
-----------------------------------------------------------------------------Beta0, .3482065, is the percentage of arrests if not black. Beta1, .3488323 is b
y how much it increases if man is black. Both beta's are statistically significa
nt since pvalue is zero.
. reg narr86 hispan, r
Linear regression

Number of obs
F( 1, 2723)
Prob > F
R-squared
Root MSE

=
2725
=
6.92
= 0.0086
= 0.0028
= .85803

-----------------------------------------------------------------------------|
Robust
narr86 |
Coef. Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------hispan | .1103311
.041952
2.63 0.009
.0280702
.1925921
_cons |
.380394 .0181878
20.91 0.000
.3447308
.4160572
-----------------------------------------------------------------------------Beta0, .380394, is the percentage of arrests if not hispan. Beta1, .1103311 is
by how much it increases if man is hispan. Both beta's are statistically signifi
cant since pvalue is zero.
. reg narr86 black hispan, r
Linear regression

Number of obs
F( 2, 2722)
Prob > F
R-squared
Root MSE

=
2725
= 30.33
= 0.0000
= 0.0304
= .84624

-----------------------------------------------------------------------------|
Robust
narr86 |
Coef. Std. Err.
t
P>|t|
[95% Conf. Interval]

-------------+---------------------------------------------------------------black | .3987517
.058942
6.77 0.000
.283176
.5143273
hispan | .1924381 .0414864
4.64 0.000
.1110901
.2737861
_cons | .2982871 .0170711
17.47 0.000
.2648135
.3317606
-----------------------------------------------------------------------------This multiple regression is statistically significant since Prob > F
= 0.
0000. The model shows that if man arrested is neither black nor hispan, percenta
ge of arrests is .2982871(beta0). Ifman is black it will increase by
beta2, if hispanic by beta1.
2)
. reg bwght cigs faminc
Source |
SS
df
MS
-------------+-----------------------------Model | 17126.2088
2 8563.10442
Residual | 557485.511 1385 402.516614
-------------+-----------------------------Total | 574611.72 1387 414.283864

Number of obs
F( 2, 1385)
Prob > F
R-squared
Adj R-squared
Root MSE

=
=
=
=
=
=

1388
21.27
0.0000
0.0298
0.0284
20.063

-----------------------------------------------------------------------------bwght |
Coef. Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------cigs | -.4634075 .0915768
-5.06 0.000
-.6430518 -.2837633
faminc | .0927647 .0291879
3.18 0.002
.0355075
.1500219
_cons | 116.9741 1.048984 111.51 0.000
114.9164
119.0319
-----------------------------------------------------------------------------Beta0 is 116.9741. The higher the income, the higher the birthweight. The Opposi
te happens for cigarettes(negative correlation with birthweight). The model is s
tatistically significant since: Prob > F
= 0.0000.
. correlate cigs faminc
(obs=1388)
|
cigs faminc
-------------+-----------------cigs | 1.0000
faminc | -0.1730 1.0000
They're indeed negatively correlated.
. reg bwght cigs
Source |
SS
df
MS
-------------+-----------------------------Model | 13060.4194
1 13060.4194
Residual |
561551.3 1386 405.159668
-------------+-----------------------------Total | 574611.72 1387 414.283864

Number of obs
F( 1, 1386)
Prob > F
R-squared
Adj R-squared
Root MSE

=
=
=
=
=
=

1388
32.24
0.0000
0.0227
0.0220
20.129

-----------------------------------------------------------------------------bwght |
Coef. Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------cigs | -.5137721 .0904909
-5.68 0.000
-.6912861 -.3362581

_cons | 119.7719 .5723407 209.27 0.000


118.6492
120.8946
-----------------------------------------------------------------------------The effect of cigarettes is still around -0.5, we can conclude that faminc doens
t change this substantially.
In the bivariate regression(bwght-cigs), we can see that r-squared is very low (
0.02), this suggests NO causal effect.
3)
we wanna test: Ho: U2years=U5years (i.e. 1.5 and 4) against U5years>U2years
. mean affairs if yearsmarried==1.5
Mean estimation

Number of obs

88

-------------------------------------------------------------|
Mean Std. Err.
[95% Conf. Interval]
-------------+-----------------------------------------------affairs | .4659091 .1828236
.1025274
.8292908
-------------------------------------------------------------. mean affairs if yearsmarried==4
Mean estimation

Number of obs

105

-------------------------------------------------------------|
Mean Std. Err.
[95% Conf. Interval]
-------------+-----------------------------------------------affairs | 1.085714 .2462593
.5973729
1.574056
-------------------------------------------------------------Test statistics: . display (1.085714-0.4659091)/sqrt(0.2462593^2+0.1828236^2)
2.0208487, which for 5% significance level, is greater than the t observed. Ther
efore we can reject the null hypotesis.
b)
. reg affairs d_children, r
Linear regression

Number of obs
F( 1, 599)
Prob > F
R-squared
Root MSE

=
601
=
8.08
= 0.0046
= 0.0108
= 3.2836

-----------------------------------------------------------------------------|
Robust
affairs |
Coef. Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------d_children | .7598123 .2672709
2.84 0.005
.2349104
1.284714
_cons | .9122807 .2079019
4.39 0.000
.5039754
1.320586
-----------------------------------------------------------------------------The regression is statistically signiicant sice p-values tend to zero. Beta 0 is
the level of betrayals if there's no children. Otherwise, betrayals will be Bo+
b1.

We can say that there's no causal effect since the r-squared is pretty low (0.01
). Moreover, we can imagine that betrayals can happen even before children were
born.
4)
. reg nfatal spirits income drinkage youngdrivers
Source |
SS
df
MS
-------------+-----------------------------Model | 1406002.56
4 351500.64
Residual | 10488595.1 331 31687.5985
-------------+-----------------------------Total | 11894597.7 335 35506.2617

Number of obs
F( 4, 331)
Prob > F
R-squared
Adj R-squared
Root MSE

=
=
=
=
=
=

336
11.09
0.0000
0.1182
0.1075
178.01

-----------------------------------------------------------------------------nfatal |
Coef. Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------spirits | -74.98059 16.51275
-4.54 0.000
-107.4637 -42.49743
income | .0331938 .0056835
5.84 0.000
.0220135
.0443741
drinkage | -18.87723 11.46391
-1.65 0.101
-41.42854
3.67407
youngdrivers | 140.3551 462.0509
0.30 0.761
-768.5715
1049.282
_cons | 213.3899 276.7084
0.77 0.441
-330.9389
757.7186
-----------------------------------------------------------------------------The regression is statistically significant since Prob > F
= 0.0000. We ha
ve beta0=213.3899, which is the number of night-time vehicle fatalities given al
l the other variables zero. Youngdrivers
has a positive effect on fatalities, which suggests that 15-24 yers old people
cause more fatalities. Also it seems that the higher the income the higher the f
atalities. On the pther hand, drinkage has a negative effect
which means that states were drink age is high, fatalities are lower on average.
Anyway, youngdrivers and drinkage have a high p.value, i.e. we can't exclude th
e beta"i" to be 0.
. reg nfatal spirits income drinkage youngdrivers beertax unemp
Source |
SS
df
MS
-------------+-----------------------------Model | 3046927.74
6 507821.29
Residual | 8847669.93 329 26892.6138
-------------+-----------------------------Total | 11894597.7 335 35506.2617

Number of obs
F( 6, 329)
Prob > F
R-squared
Adj R-squared
Root MSE

=
=
=
=
=
=

336
18.88
0.0000
0.2562
0.2426
163.99

-----------------------------------------------------------------------------nfatal |
Coef. Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------spirits | -77.14126
15.3022
-5.04 0.000
-107.2438 -47.03877
income | .0578271 .0061621
9.38 0.000
.0457049
.0699492
drinkage | -9.109599 10.69445
-0.85 0.395
-30.14773
11.92853
youngdrivers | -420.6074 431.6744
-0.97 0.331
-1269.798
428.5828
beertax | 92.74566 21.15175
4.38 0.000
51.13593
134.3554
unemp | 32.56515 4.473538
7.28 0.000
23.76481
41.3655
_cons | -507.0885 271.6812
-1.87 0.063
-1041.54
27.3629
-----------------------------------------------------------------------------The model is still significant, and we can notice that beta0 itself is now signi
ficant (10% level), which wasnt before. Since this is the first add, we can look
at the r-squared critically and saying that this model seems

to explain more the fatalities. Both beertax and unemployment have a positive ef
fect on fatalities.

You might also like