Vous êtes sur la page 1sur 5

MTH 202 : Probability and Statistics

Homework 8

3rd April, 2017

Submission : Submit solutions of 1, 3, 4, 5, 8 before 8th April, 2017.

=

H 0 : p = 0.9 and H 1 : p = 0.5, ﬁnd the values of α and β if this hypothesis is to be tested by means of a single observed value of X and the critical region consists of the points {0, 1}. [Ref : Exercise-14, Page-118, Hoel]

(2) A bag is known to contain eight balls, of which either one or two are white and the rest black. To test the hypothesis that there is only one white ball, balls are drawn until a white ball appears. Let X equal the number of balls drawn and ﬁnd f (x) under both hypothesis. Carry out the test if X = 4 and a good critical region of approximate size α = 0.25 is selected. [Ref :

, and the hypotheses

(1) Given f (x) =

p x (1 p), x

0, 1, 2,

Exercise-21, Page-119, Hoel]

Solution : The hypothesis are set as follows

H 0

H 1

 : 1 white and 7 black balls : 2 white and 6 black balls

We assume that the balls are drawn without replacement. Let f 0 (x) and f 1 (x) denote the densities of X in the situations H 0 and H 1 respectively. Then f 0 (x) and f 1 (x) are given by

f 0 (x) =

f 1 (x) =

1/8

0

if x = 1, 2, elsewhere

1/4

3/14

5/28

1/7

3/28

1/14

1/28

0

1

if x = 1 if x = 2 if x = 3 if x = 4 if x = 5 if x = 6 if x = 7 elsewhere

, 8

2

Now it is given that 0.25 = α = P (X C|H 0 ) = |C|/8, where

C is the critical region. This implies |C| = 2. Next to minimize

β =

from {1, 2,

and x 2

, 8} such that f 1 (x 1 )+f 1 (x 2 ) is maximum. Hence

C = {x 1 , x 2 } = {1, 2}.

accept H 0 .

C, which implies we can

P(X

C|H 1 ) we need to choose two points x 1

Finally 4

(3) Given

f(x; θ) =

(1 + θ)x θ

0

if θ > 0, 0 x 1 elsewhere

if the hypothesis H 0 : θ = 2 is to be tested by taking single observation on X and using the interval x < 0.25 as the critical region, (a) calculate the value of α, and (b) calculate the prob- ability of determining that H 0 is false if the true value of θ is 3. [Ref : Exercise-16, Page-118, Hoel]

(4) What critical region consisting of a single interval with α = 0.5 would you choose in previous problem if you wanted a critical region that minimizes β? [Ref : Exercise-17, Page-118, Hoel]

(5) Let X be normally distributed with mean µ and variance σ 2 = 4. If you are testing the hypothesis H 0 : µ = 1 against the alternative H 1 : µ = 6, by means of a single observed value of X, and if the critical region is chosen to be the interval X < 2, what are the values of α and β? [Ref : Exercise-18, Page-118, Hoel]

(6) A box is known to contain either three red and four black balls or four red and three black balls. Three balls are to be drawn and on the basis of their colors a decision as to the contents of the box will be made. If H 0 corresponds to three red and four black balls and if H 0 will be accepted unless three red balls are obtained, what are the values of α and β here? [Ref : Exercise- 19, Page-118, Hoel]

Solution : The hypothesis are set up as,

H 0

H 1

 : 3 red and 4 black balls : 4 red and 3 black balls

The critical region is given as 3 red balls. Hence

α = P (obtaining 3 red balls|H 0 ) = 3

3

/

7

3

= 1/35

3

and

β

= P (not obtaining 3 red balls|H 1 )

= 1 P (obtaining 3 red balls|H 1 )

= 1 4/35 = 31/35

=

1 4

3

/

7

3

(7) If the scores on a set of examination papers are changed by (a) adding 10 points to all scores, (b) increasing all scores by 10%, what eﬀects will these changes have on the mean and standard deviation? [Ref : Exercise-22, Page-119, Hoel]

Solution : Let N be the number of papers and x 1 , denote the scores. Then

µ =

1

N

N

i=1

x i , σ 2 =

1

N

N

i=1

(x i µ) 2

, x N

(a) The new set of value are x 1 = x 1 + 10,

Hence the revised mean and the standard deviations are

, x N

= x N + 10.

µ = µ + 10, (σ ) 2 = σ 2

since x i µ = x i µ for

(b) The new set of value are x 1 = 1.1x 1 ,

the revised mean and the standard deviations are

each i = 1,

, N . , x N = 1.1x N . Hence

µ = 1.1µ, (σ ) 2 = (1.1) 2 σ 2

(8) Given the density

f(x; θ) =

1

0

if 0 < x < θ, elsewhere

if you are testing the hypothesis H 0 : θ = 3 against H 1 : θ = 2 by means of a single observed value of X , (a) what would the sizes of the type-I and type-II errors be if you choose the interval x < 1 as the critical region? (b) What would the sizes of those errors be if you choose the interval 1 < x < 2 as the critical region? [Ref : Exercise-27, Page-119, Hoel]

(9) A professional pollster is hired to estimate the proportion of voters who favour a proposed bond issue. If the estimate must be correct to within 0.03 units of the true proportion with a probability of 0.90, how large the sample should be taken? [Ref : Exercise-31, Page-163, Hoel]

4

Solution : Here we wish to estimate pˆ = 0.5 (assigning equal probability to vote for or against the proposed bond) with a 90% conﬁdence interval. Hence we have

P pˆ 1.64 pˆ(1 pˆ)

n

< p < pˆ + 1.64 pˆ(1 pˆ) = 0.9

n

The conﬁdence is interval is thus

0.5 1.64 (0.5)(0.5) , 0.5 + 1.64 (0.5)(0.5)

n

n

For the accuracy up to 0.03 units we require

1.64 (0.5)(0.5)

n

0.03 =n 747

(10) A purchaser of bricks believes that the quality of the bricks is de- teriorating. From past experience the mean crushing strength of such bricks is 400 pounds with a standard deviation of 20 pounds. A sample of 100 bricks yielded a mean of 395 pounds. Test the hypothesis that the mean quality has not changed against the alternative that it has deteriorated. [Ref : Exercise- 41, Page-164, Hoel]

Solution : We set up the hypothesis H 0 : µ = 400 lbs against H 1 : µ < 400 lbs. From the description of the alternate hypoth- esis, the test is going to be left tailed. Hence the critical region C for the test with α = 0.05 is given by

P

X 400

100 t 0 = 0.05

20/

where

bution. Hence from the table of standard normal distribution t 0 = 1.645. This means

100 satisfy approximately the standard normal distri-

X400

20/

C

= (−∞, 400 20

10 × (1.645)) = (−∞, 396.71)

Since the sample mean 395 C, the null hypothesis H 0 is rejected.

(11) A biologist has mixed a spray designed to kill 70% of a certain type of insect. If a spraying of 200 such insects killed 130 of them, would you conclude that this mixture was satisfactory? [Ref : Exercise-42, Page-164, Hoel]

5

Solution : The hypothesis is set up as H 0 : p = 0.7 against the alternate hypothesis H 1 : p < 0.7, since it would not be unacceptable if the spray kills more insects. The test is therefore left tailed. The mean µ and the standard deviation σ for the test is given by

µ = 0.7, σ = (0.7)(1 0.7)

200

= 0.0324

pˆ0.7

Now assuming the sample is large, the random variable satisfy approximately the standard normal distribution. Using the table of standard normal distribution

0.0324

P pˆ0.7

0.0324

1.645 = 0.05

i.e. the critical region C for the test with α = 0.05 is given by

C = (0, 0.7 (0.0324)(1.645)) = (0, 0.647)

C, the null

hypothesis H 0 is accepted, which means the mixture is satisfac- tory.

(12) Suppose X and Y are the means of two samples of size n each from a normal population with variance σ 2 . Determine n so that the probability will be about 0.95 that the two sample means will diﬀer by less than σ. [Ref : Exercise-50, Page-164, Hoel]

Now since the sample mean 130/200 = 0.65