Anda di halaman 1dari 8

STA 302 / 1001 Summer 2014

Term Test #1

LAST NAME : ____Solutions__ FIRST NAME: ________________
STUDENT # : ___________________
ENROLLED IN (tick one): STA 302 STA 1001

INSTRUCTIONS:
Time: 90 minutes
Aids allowed: calculator
A t-distribution table is provided on the last page
There are 3 questions on 8 pages, including this cover sheet and the t-table
Please carry all possible precision through a numerical question
Give final answers to 3 decimals.

Some formulae:
( )( )
( )
{ }
( )
{ }
( )
{ }
( )
( )
( ) ( ) ( )
{ }
1 1
1 0 1 2
2 2
1 1
2 2
2
1 0 2 2
1 1
2
2
0 1 2
1
1
2 2
2
2
1
1 1 1
2
1
,

n n
i i i i
i i
n n
i i
i i
n n
i i
i i
n
i
n i
i
i
n n n
i i i i
i i i
h
X X Y Y X Y nXY
b b Y b X
X nX X X
X
Var b Var b
n
X X X X
X
Cov b b SSTO Y Y
X X
SSE Y Y SSR Y Y b X X
Y Va

= =
= =
= =
=
=
= = =

= = =

| |
|
= = +
|

\ .
= =

= = =
=



{ }
( )
( )
( )( )
( ) ( )
{ }
{ }
( )
( )
2
2 1
2
2 2
1
1 1
2
2 2
2
1
1

1
n
h i i
i
h
n
n n
i
i
i i
i i
h
h h
n
i
i
X X X X Y Y
r Y r
n
X X
X X Y Y
X X
pred Var Y Y
n
X X


=
=
= =
=
| |

|
= + =
|
( (

\ .

| |

|
= = + +
|

\ .


For TA use only
Q1 Q2 Q3 Total
10 10 30 50




Page 1 / 8
1. Multiple Choice Questions: circle the best answer (no explanation needed)
(10 marks)

I. For the simple linear regression model
0 1 i i i
Y X = + + ,
A. Y
i
is random but known, and X
i
is a known constant
B.
0
is random and unknown, and X
i
is a known constant
C.
1
is random but known, and
i
is random and unknown
D. Y
i
is a known constant, and
i
is random and unknown


II. For a particular data set, suppose the Y variable (Temperature) is changed
from to via the following: =
9
5
+32. Which of the following
statements is FALSE?

A. The slope will change.
B. The p-value for the slope will not change.
C. MSE will not change.
D. R
2
will not change.


III. For a particular data set, a new point (

) is added such that

and

<

. Which of the following is necessarily TRUE?



A. SSR will decrease and b
0
will increase
B. SSE will increase and b
0
will decrease
C. MSE will increase and b
1
will decrease
D. R
2
will decrease and b
1
will remain the same


IV. Which of the following is NOT an appropriate use of a linear regression
model?

A. Description and Control
B. Interpolation
C. Prediction
D. Determination of Causation

V. Which of the following is an assumption of the Normal linear regression
model:
0 1 i i i
Y X = + +

A. The X
i
s come from a Normal distribution
B. The e
i
s are uncorrelated
C. The Y
i
s have constant variance
D. The Y
i
s are independent of the
i
s
Page 2 / 8
2. A simple linear regression model is fit to initial data ( )
1
,
n
i i
i
X Y
=
, and the estimated
regression coefficients ( )
0 1
, b b are calculated. Let the sample means of these data
be
1
1 n
n i
i
X X
n
=
=

and
1
1 n
n i
i
Y Y
n
=
=

. Suppose that the Gauss-Markov
assumptions hold, and that we can model the errors as

~(0,
2
).
(10 marks)
I. Show that

= 0. Recall that

. (3)


= (

)(


0

1

) (1)

= (

)(

)
0
(

)
1
(

)(

) (1)

=


0
(0)
1


= 0 (1)


II. Show that SSTO =SSR +SSE. Recall that

= 0. (4)

= (

)
2
=

2
(1)
=

2
+

2
+2

(1)
= + +2


= + +2

(
0
+
1

) 2

(1)
= + +0(2
0
2

) +2
1


= + (1)



Suppose instead that we use a regression model

=
1

.
III. Show that the Least Squares estimate of
1
is
1
=

2
(3)

=

2
= (

)
2
(1)

1
= 2(

= 0 (1)
X
i

2
= 0

(1)


Page 3 / 8
3. The data analyzed below come from an engineering experiment at an electronics
manufacturing plant where several mobile phones were tested until they failed.
The following variables were measured from a sample of phones:
temperature: the temperature in C
lifetime: the amount of time until the phone failed (in months)
A plot of the data (not shown) indicates that the Normal linear regression model is
appropriate. The SAS output from two procedures is given below. Note that some
values are missing, and some have been replaced by letters. (30 marks)

I. Find the 11 missing lettered values (A through K) in the output. (11)
(1 mk each)

A. 1
B. 28
C. 29
D. 1599.835
E. 1599.835
F. 4.884
G. 0.035
H. 8.507
I. 1
J. 52.556
K. 10753
The REG Procedure
Dependent Var i abl e: l i f et i me

Number of Obser vat i ons Used 30

Analysis of Variance
Sumof Mean
Sour ce DF Squar es Squar e F Val ue Pr > F

Model (A) (D) (E) (F) (G)
Er r or (B) 9153. 46681 326. 90953
Cor r ect ed Tot al (C) 10753

Root MSE 18. 08064 R- Squar e 0. 1488
Dependent Mean 52. 55592 Adj R- Sq 0. 1184
Coef f Var 34. 40267

Parameter Estimates
Par amet er St andar d
Var i abl e DF Est i mat e Er r or t Val ue Pr > | t |
I nt er cept 1 68. 92578 8. 10274 (H) <. 0001
t emper at ur e (I) - 0. 96672 0. 43700 - 2. 21 0. 0353

The MEANS Procedure

Var i abl e Mean St d Dev Cor r ect ed SS Mi n Max
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
t emper at ur e 16. 9333333 ( - - - - - - ) ( - - - - - - ) 5. 0 30. 0
l i f et i me (J) 19. 2562586 (K) 13. 95 96. 89
Page 4 / 8
II. What is the sample standard deviation of the temperature? (2)

() =

1

=

2
{
1
}

1

=

(18.08064)
2
(0.437)
2

29

= .

III. What is the two-sided p-value for testing whether the correlation coefficient
is zero (H
0
: =0)? (1)


= 0.035








IV. Would you reject the null hypothesis that the intercept is 0 (H
0
:
0
= 0 vs.
H
a
:
0
0) at the 5% level? Give the test details. (2)

=

0

0
{
0
}
=
68.92578 0
8.10274
= 8.507

= 30 2 = 28

= 2 (0, 0.025) < 0.05

Reject at 5%.




Page 5 / 8
V. Would you reject the null hypothesis that the intercept is 55 (H
0
:
0
= 55 vs.
H
a
:
0
55) at the 5% level? Give the test details. (2)

=

0

0
{
0
}
=
68.92578 55
8.10274
= 1.72

= 30 2 = 28

= 2 (0.025, 0.05) = (0.05, 0.1)


Do not reject at 5%.




VI. What percentage of the variation in lifetime can we attribute to temperature?
(1)


14.88%









VII. Predict the expected value of the lifetime for a phone at the freezing point of
water (0C). (2)


( 5)






Page 6 / 8
VIII. Predict the expected value of the lifetime for a phone at 10C. Build an
appropriate 95% interval around this prediction. (5)

=
0
+
1

= 68.92578 0.96672(10) = 59.2586 (1)


=

2

2
{
1
}
=
(18.08064)
2
(0.437)
2
= 1711.85 (1)

=
2

+
(

)
2

= (18.08064)
2

1
30
+
(1016.93333)
2
1711.85
= 20.077 (1)

95%

0.975,28

(1)

= 59.2586 2.048(4.48074) = (. , . ) (1)









IX. Would a 95% CI for the slope include zero? Explain how you know this
without constructing the interval. (2)

. (1)
,
> 0.05, . (1)







X. Based on this SAS output, is there a linear relationship between temperature
and lifetime? How did you arrive at this conclusion? (2)

(1),
, . (1)

Page 7 / 8

Page 8 / 8

Anda mungkin juga menyukai