Scribd
Upload
61 views

Stat 234 Chang. Section 02, 391255: Ben Jacobson March 6, 2012

1. This document contains statistical calculations and analyses for hypothesis tests with varying sample sizes, means, variances, and confidence intervals. 2. Hypothesis tests include one-sample z-tests to compare a sample mean to a hypothesized population mean and two-sample t-tests to compare means between two independent samples. 3. Confidence intervals calculated include 95% and 99% CIs for differences between means, proportions, and averages using pooled and Welch-Satterthwaite-adjusted degrees of freedom approaches.

Uploaded by

Ben Jacobson
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
61 views

Stat 234 Chang. Section 02, 391255: Ben Jacobson March 6, 2012

1. This document contains statistical calculations and analyses for hypothesis tests with varying sample sizes, means, variances, and confidence intervals. 2. Hypothesis tests include one-sample z-tests to compare a sample mean to a hypothesized population mean and two-sample t-tests to compare means between two independent samples. 3. Confidence intervals calculated include 95% and 99% CIs for differences between means, proportions, and averages using pooled and Welch-Satterthwaite-adjusted degrees of freedom approaches.

Uploaded by

Ben Jacobson
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Stat 234; Chang.

Section 02, 391255


Ben Jacobson
March 6, 2012

12.1

H0 : = 30, 000, Ha : > 30, 000, = 7500, = .01


7500
Critical value zc = z1 = 2.3263, x
c = 30000 + 2.3263
= 30872.3805
400

= P r(RetainH0 | = 31000, = 7500) = P r(X < 30872.3805| = 31000, = 7500)


P r(Z < 0.3403) = 0.3668.
= P r(Z < (30872.380531000))
7500

400

Thus Power = 1 = 0.6332

12.2

Here we want =0.50 when the true mean is 31000; that is


= P r(RetainH0 | = 31000, = 7500) =
<x
P r(X
c | = 31000, = 7500) = .50
That means that x
c = 31, 000.T henzc = 3100030000
= 43 = z1
7500

100

So 1 =0.90879 and = 0.09121.

12.8

a. H0 : = 13
b. Ha : 6= 31
c. P value using continuity correction:


28 .5 24
7
= 2P r(Z ) = .38
P-value = 2P r(K 28|H0 ) = 2P r Z
4
8
d. He does not have ESP (because the P-value exceeds the significance level)

4
1.
2.
3.
4.

12.22
H0 : = 1.2
Ha : 6= 1.2
= .05
x
s = 1.2275, ts =

(1.22751.2)
s
n

(.0275)
.03864

12

= 2.4653

5. P-value=P r(|T11 )| > ts ) = 2P r(t > 2.4653) = 2 tcdf (2.4653, 99999999, 11) = .0314 <
so reject H0 .
Using table 4 we get .02 < P value < .04.
6. There is sufficient evidence .02 < P value < .04. to conclude that population average
thickness is not 1.2 millimeters.

14.4

The differences Before-After are: d: 6 1 9 12 -1 9


and the standard deviation s=5.06 with t5.95 =4.0321.
The sampled differences have mean d=6
The 99% confidence interval for the average decrease in the puzzle completion time is
99%C.I.:
s

D = d = t5.995
6
So

= 6 8.3 Or: -2 seconds D 14 seconds.

14.6

a.

s
SE(xy) =

s2y
s2x
+
=
nx ny

4.72 2.32
+
= 100
25
45

b.Here df=25-1=24, t7.975 = 2.0639 and the 95% CI is of the form:


(
x y) t24.975 SE(xy) or 95%CI : 4 2.065 or 6.1 < x y < 1.9

14.10

Here nx = 24, x
= 15, sx = 2.5, ny = 24, y = 12.5, sy = 3.5, and null hypothesis: H0 :
x y = 0 and alternative
q hypothesis: Ha : x y 6= 0.
2

s2

We have: SE(xy) = nsxx + nyy = .87797 and test statistic: ts = SExy = 2.8475.
(
x
y)
Reject H0 if |ts | > t23.975 = 2.0687
Conclusion: Reject H0 ; There is a difference in the mean number of problems. Open book
tests make a difference.
2

14.14

a. Here

s2x
nx

= 2 and

s2y
ny

= 0.75 Welchs formula: df= (2+0.75)


= 12
22
0.752
7

11

b. The 95% confidence interval for x-y is of the form (


x y) tdf.975 SE(xy) .

95%CI : 15 2.1788 1 + 0.75 = 15 3.613 or 11.4 x y 18.6.

14.20

a. The pooled standard deviation is:


s
r
r
(nx 1)Sx2 + (ny 1)Sy2
7 42 + 11 32
221
sp =
=
=
= 3.4238
(nx + ny 2)
8 + 12 2
18
b. The 95% confidence interval x y is (
x y) t18.975 sp

1
nx

1
ny

(35 20) 2.1009(3.4238)(0.4564) = 15 3.283 or 11.717 x y 18.283.

10

14.41

p
a. K1 has a binomial distribution with mean 1 = 1001 and standard deviation 1001 (1 1 ).
Z1 , the standardized form of K1 , will be approximately standard normal, because n = 100
is moderately large as long as 1001 and 100(1 1 ) both are greater than 5.
b. = P r(C|H0 ) = 0.05.
c. 1 = P r(C|HA ) = P r(K1 > 59.8|1 = 0.3) + P r(K1 , 40.2|1 = 0.3) = P r(Z >
59.830

) = 0.987.
21
d. Let X1 = 1 if a randomly selected adult answers yes to the first question and 0 otherwise. Let X2 = 1 if a randomly selected adult answers yes to the second question and 0
otherwise. Let Di = X1 X2 . Then: E(Di ) = 1 2 and
V ar(Di ) = V ar(X1 + X2 ) = V ar(X1 ) + V ar(X2 ) 2Cov(X1 X2 )
And
Cov(X1 , X2 ) = E(X1 X2 )E(X1 )E(X 2) = 12 1 2
If we put it all together:
V ar(Di ) = 1 (1 1 ) + 2 (1 2 ) 2(12 1 2 )
K1 K2 is the sum of n = 100 independent Di so according to the central limit theorem,
K1 K2 is approximately normally distributed with mean and variance 100 times those of
Di . Finally ZD is just the bootstrapped approximation of standardized K1 K2 so it is

approximately standard normal.


e. An approximate (1 )100% confidence interval for 1 2 is:
p
CI : p1 p2 z2 = 100(p1 (1 p1 ) + p2 (1 p2 ) 2(p12 p1 p2 ))
f. ZD defined in part d can be used to test this hypothesis. The critical region or rejection
region is:
C = (ZD : |ZD | > 1.96)
g. An approximate 95% confidence interval 1 2 is:
p
(p1 (1 p1 ) + p2 (1 p2 ) 2(p12 p1 p2 ))
1 2 1.96
100
0.364 1 2 0.236
h. Yes, ZD = 4.7 is in the critical region. Based on this random sample of 100 adults,
there is sufficient evidence to support the hypothesis that there is a difference in the adults
populations response rate to these two questions at the 5% significance level.

You might also like