ECONOMICS 351* -- Output of Stata regress Command

M.G. Abbott

Reading Output of Stata regress Command

TOPIC:

Interpreting Output of Stata regress Command

DATA:

auto1.dta

(a Stata-format data file)

MODEL: pricei = 0 + 1 weighti + ui

(i = 1, ..., N)

. regress price weight

Source |
SS
df
MS
---------+-----------------------------Model |
184233937
1
184233937
Residual |
450831459
72 6261548.04
---------+-----------------------------Total |
635065396
73 8699525.97

Number of obs
F( 1,
72)
Prob > F
R-squared
Adj R-squared
Root MSE

=
=
=
=
=
=

74
29.42
0.0000
0.2901
0.2802
2502.3

-----------------------------------------------------------------------------price |
Coef.
Std. Err.
t
P>|t|
[95% Conf. Interval]
---------+-------------------------------------------------------------------weight |
2.044063
.3768341
5.424
0.000
1.292858
2.795268
_cons | -6.707353
1174.43
-0.006
0.995
-2347.89
2334.475
------------------------------------------------------------------------------

Source |
SS
df
MS = SS/df
---------+--------------------------------------------Model |
184233937 = ESS
1 = K1
184233937 = ESS/(K-1)
Residual |
450831459 = RSS
72 = NK
6261548.04 = RSS/(N-K) = $ 2
---------+--------------------------------------------Total |
635065396 = TSS
73 = N1
8699525.97 = TSS/(N-1) = S2Y

Number of obs
F( 1,
72)
Prob > F
R-squared
Adj R-squared
Root MSE

=
=
=
=
=
=

74
29.42
0.0000
0.2901
0.2802
2502.3

=
=
=
=
=
=

N
F-statistic for test of H0: 1 = 0 against H1: 1 0
p-value for F-statistic
R2
R2
$

-------------------------------------$ ($ j )
Std. Err. = se$ ($ j ) = Var
price |
Coef. = $ j
---------+---------------------------$ ($ 1 )
.3768341 = se$ ($ 1 ) = Var
weight |
2.044063 = $ 1
_cons | -6.707353 = 0
1174.43 = se( 0 ) = Var( 0 )
--------------------------------------

ECON 351*: Stata regress Command

351memo1.doc

Page 1 of 2 pages

ECONOMICS 351* -- Output of Stata regress Command

M.G. Abbott

The printed t-statistics are those for performing two-tail t-tests of the null
hypothesis H0: j = 0 against the alternative hypothesis H1: j 0.

The sample value of each t-statistic is the t-ratio:

tj =

$ j
= t-ratio for $ j
se$ ($ j )

(j = 0,1).

The null distribution of tj under H0: j = 0 is the t[N-2] distribution.

The column labelled P>|t| contains the two-tailed p-values for the tratios tj.

------------------------------P>|t| = Pr(| t | > | t j |)

t = t j = $ j se$ ($ j )
------------------------------5.424 = t 1 = $ 1 se$ ($ 1 )
0.000 = Pr(| t | > | t 1 |)

-0.006 = t = se( )
0.995 = Pr(| t | > | t |)
0

------------------------------The printed confidence intervals are the two-sided 95 percent confidence

intervals for each regression coefficient j (j = 0,1).

In general, the two-sided 100(1-) percent confidence interval for

regression coefficient j is:

[$ t
j

[ N 2]se$ ($ j ), $ j + t 2 [ N 2]se$ ($ j ) .

For the two-sided 95 percent confidence intervals, 1- = 0.95, = 0.05,

and /2 = 0.025.

--------------------------------------------------------$ t
[ 72]se$ ($ ), $ + t
[ 72]se$ ($ )
[95% Conf. Interval]

0 . 025

0 . 025

--------------------------------------------------------[ 72]se$ ($ )
[ 72]se$ ($ )
2.795268 = $ + t
1.292858 = $ t
1

0 . 025

0 . 025

-2347.89 = 0 t 0.025 [72]se( 0 )

2334.475 = 0 + t 0.025 [72]se( 0 )
--------------------------------------------------------Note: t 0 .025[ 72] = 1.9935

ECON 351*: Stata regress Command

351memo1.doc

Page 2 of 2 pages