Professional Documents
Culture Documents
Inferences About
Population Variance
One Population scenario
1 2
3 4
5 6
1
Chi-Square Distribution Interval Estimation of 2
2
For example, there is a .95 probability of obtaining a .025
2 (chi-square) value such that .025
.975
2
2 .025
2
95% of the
possible 2 values
2
0 .975
2
.025
2
7 8
9 10
11 12
2
Interval Estimation of 2 Interval Estimation of 2
Our value
13 14
Our value
15 16
17 18
3
Hypothesis Testing Hypothesis Testing
About a Population Variance About a Population Variance
Left-Tailed Test Left-Tailed Test (continued)
•Hypotheses •Rejection Rule
H0 : 2 2
0 Critical value approach: Reject H0 if 2 (12 a )
Ha : 2 2
0
19 20
21 22
23 24
4
Hypothesis Testing Hypothesis Testing
About a Population Variance About a Population Variance
Example: Buyer’s Digest (B) Example: Buyer’s Digest (B)
Recall that Buyer’s Digest is rating ThermoRite Using the 10 readings, we will conduct a
thermostats. Buyer’s Digest gives an “acceptable” hypothesis test (with a = .10) to determine whether
rating to a thermostat with a temperature variance the ThermoRite thermostat’s temperature variance is
of 0.5 or less. “acceptable”.
We will conduct a hypothesis test (with a = .10)
Thermostat 1 2 3 4 5 6 7 8 9 10
to determine whether the ThermoRite thermostat’s
temperature variance is “acceptable”. Temperature 67.4 67.8 68.2 69.3 69.5 67.0 68.1 68.6 67.9 67.2
25 26
Our value
27 28
29 30
5
Hypothesis Testing Container Filling machine
About a Population Variance
Container-filling machines are used to package a variety of
Using the p-Value liquids, including milk, soft drinks, and paint. Ideally, the
• The rejection region for the ThermoRite amount of liquid should vary only slightly because large
thermostat example is in the upper tail; thus, the variations will cause some containers to be underfilled
appropriate p-value is less than .90 (2 = 4.168) (cheating the customer) and some to be overfilled
and greater than .10 (2 = 14.684). (resulting in costly waste). The president of a company
• Because the p –value > a = .10, we cannot that developed a new type of machine boasts that this
reject the null hypothesis. machine can fill 1-liter (1,000 cubic centimeters) containers
so consistently that the variance of the fills will be less than
• The sample variance of s 2 = .7 is insufficient 1 CC2. To examine the veracity of the claim, a random
evidence to conclude that the temperature sample of 25 l-liter fills was taken, and the results (cubic
variance is unacceptable (>.5). centimeters) recorded. These data are listed here. Do these
The exact p-value is .18156. data allow the president to make this claim at the 5%
significance level?
31 32
33 34
35 36
6
Hypothesis Testing About the Hypothesis Testing About the
Variances of Two Populations Variances of Two Populations
Two-Tailed Test Two-Tailed Test (continued)
•Hypotheses •Rejection Rule
H0 : 12 22
Critical value approach: Reject H0 if F > Fa/2
Ha : 12 22
where the value of Fa/2 is based on an
Denote the population providing the F distribution with n1 - 1 (numerator)
larger sample variance as population 1. and n2 - 1 (denominator) d.f.
•Test Statistic
2 p-Value approach: Reject H0 if p-value < a
F s1
s22
37 38
39 40
41 42
7
Hypothesis Testing About the Hypothesis Testing About the
Variances of Two Populations Variances of Two Populations
Test Statistic Determining and Using the p-Value
TempKing’s sample variance is 1.768
Area in Upper Tail .10 .05 .025 .01
ThermoRite’s sample variance is .700
2
F Value (df1 = 9, df2 = 9) 2.44 3.18 4.03 5.35
Fs 1 = 1.768/.700 = 2.53
s22 • Because F = 2.53 is between 2.44 and 3.18, the area
Conclusion in the upper tail of the distribution is between .10
We cannot reject H0. F = 2.53 < F.05 = 3.18. and .05.
There is insufficient evidence to conclude that • But this is a two-tailed test; after doubling the
the population variances differ for the two upper-tail area, the p-value is between .20 and .10.
thermostat brands. • Because a = .10, we have p-value > a and therefore
we cannot reject the null hypothesis.
43 44
An important statistical measurement in service A new highway has just been completed, and the government
facilities (such as restaurants and banks) is the must decide on speed limits. There are several possible
variability in service times. As an experiment, two food choices. However, on advice from police who monitor traffic,
delivery services (Zomato and Swiggy). Delivery times the objective was to reduce the variation in speeds, which it is
to TPK were observed, and the service times for each of thought to contribute to the number of collisions. It has been
100 customers were recorded. Do these data allow us to acknowledged that speed contributes to the severity of
infer at the 10% significance level that the variance in collisions. It is decided to conduct an experiment to acquire
service times of Zomato is lesser than Swiggy? more information. Signs are posted for 1 week indicating that
the speed limit is 70 mph. A random sample of cars’ speeds is
measured. During the second week, signs are posted
Data indicating that the maximum speed is 70 mph and that the
minimum speed is 60 mph. Once again, a random sample of
speeds is measured. Can we infer that limiting the minimum
and maximum speeds reduces the variation in speeds?
Data
45 46
Matched Pairs
47 48
8
Inferences About the Difference Between Inferences About the Difference Between
Two Population Means: Matched Samples Two Population Means: Matched Samples
With a matched-sample design each sampled item Example: Express Deliveries
provides a pair of data values. A Chennai-based firm has documents that must
This design often leads to a smaller sampling error be quickly distributed to state offices throughout
than the independent-sample design because the India. The firm must decide between two
variation between sampled items is eliminated as a delivery services, Speed Post (India Postal Service)
source of sampling error. and Blue Dart (DHL Express), to transport its
documents.
49 50
Inferences About the Difference Between Inferences About the Difference Between
Two Population Means: Matched Samples Two Population Means: Matched Samples
Example: Express Deliveries
Delivery Time (Hours)
In testing the delivery times of the two services,
District Office BD SP Difference
the firm sent two reports to a random sample of its
Lucknow 32 25 7
state offices with one report carried by Speed Post Bhopal 30 24 6
and the other report carried by Blue Dart. Do the Bengaluru 19 15 4
data on the next slide indicate a difference in mean Kochi 16 15 1
delivery times for the two services? Use a .05 level of Mumbai 15 13 2
significance. Kolkata 18 15 3
Bhubaneswar 14 15 -1
Hyderabad 10 8 2
New Delhi 7 9 -2
Goa 16 11 5
51 52
Inferences About the Difference Between Inferences About the Difference Between
Two Population Means: Matched Samples Two Population Means: Matched Samples
p –Value and Critical Value Approaches p –Value and Critical Value Approaches
1. Develop the hypotheses. 2. Specify the level of significance. a = .05
H0: d = 0
3. Compute the value of the test statistic.
Ha: d
Let d = the mean of the difference values for the d i ( 7 6... 5)
d 2. 7
two delivery services for the population n 10
of district offices 2
( di d ) 76.1
sd 2. 9
n 1 9
d d 2.7 0
t 2.94
sd n 2.9 10
53 54
9
Inferences About the Difference Between Inferences About the Difference Between
Two Population Means: Matched Samples Two Population Means: Matched Samples
p –Value Approach Critical Value Approach
4. Compute the p –value. 4. Determine the critical value and rejection rule.
For t = 2.94 and df = 9, the p–value is between For a = .05 and df = 9, t.025 = 2.262.
.02 and .01. (This is a two-tailed test, so we double
Reject H0 if t > 2.262
the upper-tail areas of .01 and .005.)
5. Determine whether to reject H0.
5. Determine whether to reject H0.
Because t = 2.94 > 2.262, we reject H0.
Because p–value < a = .05, we reject H0.
We are at least 95% confident that there is a
We are at least 95% confident that there is a
difference in mean delivery times for the two
difference in mean delivery times for the two
services?
services?
55 56
SUMMARY
IFERENCIAL
STATISTICS
57 58
10