Midterm 1 - Solution Set

Matthew Strathearn
March 12, 2015

Question 1
(a)
Let f (x, , ) be the normal distribution PDF. Therefore;
1 x
f (x, , ) = ( ) (1)


1 v
x
F (x) = ( )dv (2)

v x
F (x) = ( ) (3)

x
F (x) = ( ) (4)

In other words, if x N (, 2 ), then x
N (0, 1). We want to find the
probability that the students grade falls between 40 and 64 when = 70 and
2 = 10.
P (40 x 64) = F (64) F (40)
64 70 40 70
= ( ) ( )
10 10
(5)
(1.90) (9.49)
.0287 0
.0287

1
(b)
The probability of a student scoring higher than 64 is as follows;
P (x > 64) = P (x ) P (x 64)
= 1 P (x 64)
64 70
= 1 ( )
10 (6)
1 (1.90)
1 0.0287
0.9713

Question 2
(a)
Assuming the normality of and that the regressor p is fixed among repeated
samples;

N (, V ar())
(7)
Where;
= n 2
V ar() (8)
i=1 (xi x
)2
As 2 is not known, thus has to be estimated with s2 , it then follows that

t= n tn2 . (9)
s/ i=1 (xi x
)2
The 95% confidence level is as follows;
P (|t2.5%,302 | t |t2.5%,302 |) = 95%
P (2.048 t 2.048) = 95%
(10)

P (2.048 n 2.048) = 95%
s/ i=1 (xi x
)2
Solving for the 95% confidence interval;
{ }
s2 s2
P 2.048 n
+ 2.048 n
= 95%
i=1 (xi x
)2 i=1 (xi x
)2
P {9.835284 2.048 1.375 9.835284 + 2.048 1.375} 95%
P {12.58 7.11} 95%
(11)
Therefore [12.58, 7.11]. Also note that the confidence interval does
not include zero and therefore the coecient is statistically dierent from zero.
There is a 95% chance that this confidence interval contains the true .

2
(b)
H0 : = 0 is the null hypothesis while Ha : = 0 is the alternative. From
earlier, the t-statistic can be calculated as such;

t= n
s/ i=1 (xi x
)2
9.835284 0 (12)
t=
1.375388
t = 7.150914
At the 5% significance level t2.5%,28 = 1.071. Since |t| > |t2.5%,28 | we reject
the null hypothesis in favour of the alternative, at the 5% level.

(c)
All that changes is the t-critical value. t5%,28 = 1.071, since t < t5%,28 we fail to
reject the null hypothesis at the 5% level. To determine the P-Value we calculate
the probability of observing a t-statistic at least as extreme as -7.150914.
P (t 7.150914| = 0) = 1 P (t 7.150914| = 0) = 1 0 = 1 (13)
Since p > .05, at the 5% level, we fail to reject the null hypothesis that = 0.

Question 3
(a)
Assumption 1: i , i = 1, . . . , n, are independent and identically distributed
(i.i.d.), and i N (0, 2 ).
Assumption 2: i and xi are independent for every i [1, n].
Assumption 3: xi , i = 1, . . . , n, are fixed regressors.

(b)

yi = + xi + i (14)
Using the sample moments;

1 1 1
n n n
(i ) =
(yi ) (xi ) = 0
n i=1 n i=1 n i=1
(15)
1 1 1 1 2
n n n n
(xi i ) = (xi yi )
(xi ) (x ) = 0
n i=1 n i=1 n i=1 n i=1 i

Solving the two equations yields;

= y x
(16)

3
 1 1 1 1 2
n n n n
(xi i ) = y x
(xi yi ) ( ) (xi ) (x ) = 0 (17)
n i=1 n i=1 n i=1 n i=1 i

1 1 1 2 1
n n n n
(xi yi ) y (xi ) = (xi ) x
(xi ) (18)
n i=1 n i=1 n i=1 n i=1
n
(x x )(yi y)
= i=1 n i (19)
i=1 (xi x
)2

(c)

yi y = (xi x
) + (i ) (20)
Substituting 23 into 22;
n
i=1 (xi x
)((xi x) + (i ))
= n
i=1 (xi x
)2
n
(x x )(i )
= +n i
i=1
(21)
(x i x)2
n i=1
(x x )E[i ]
n i
= + i=1
E[]
i=1 i x
(x )2
= since E[i ] = 0. This demonstrates the unbiasedness of
Therefore E[]

.

=+x +
( ) (22)

E[ + E[
] = + x( E[]) ] (23)
This implies that E[ = and E[
] = since E[] ] = 0.

(d)
Notice that n
(x x)(i )
= n i
i=1
, (24)
i=1 i x
(x )2
= , and
E[]

1 n
Var[ ] = n (xi x
)2 Var[i ] (25)
( i=1 (xi x
)2 )2 i=1

1 n
Var[ ] = n (xi x
)2 2 (26)
( i=1 (xi x
)2 )2 i=1

2
Var[ ] = n . (27)
i=1 (xi x
)2

4
Because i are i.i.d. N (0, 2 ) and xi , i = 1, . . . , n, are fixed, as a summation of
normal random variables is also a normal random variable, it then follows that
( )
2
N 0, n . (28)
i=1 (xi x )2