STAT 101: Statistical Methods

MODULE 4.1 1st Semester 2023-2024

Inference on Single Population

In this module as well as in succeeding modules, our focus will now be on inferential statistics. Recall that
inferential statistics uses inductive reasoning to draw inferences about the characteristics of a population based
on a random sample. We will be addressing two areas of concern in inferential statistics: estimation of an
unknown true value of a parameter of interest and test of hypothesis about the value of the unknown parameter
or the distribution of the population under study. For simplicity of analyses, we will assume from this point on
that samples were taken using simple random sampling.

Before you start the module, you may listen to this song titled A Statistically Significant Love Song by Henrik
Widegren.

LEARNING OBJECTIVES
At the end of this module, you must be able to:
1. use the most appropriate statistical method(s) for inference on a single population;
2. interpret the output of a statistical software in the inference of a single population; and
3. recommend appropriate actions based on statistical results.

TOPIC 4A. ESTIMATION AND TEST OF HYPOTHESIS ON ONE POPULATION MEAN

ILLUSTRATION

We might want to know

• the mean length of student’s stay (in years) in the university
• the mean starting salary (in pesos) of graduates
• the mean percentage of students who found work within 3 months of graduation for the past 10 years

Let us say you want to compute the population mean, , for the illustrations above. To do this, we need to do a
census and get data from all units under study, then compute  from the population data. This can be easily
done if the population is small. However, if the population is large, it is not practical nor efficient to do a census
to obtain the needed data.

THINK ABOUT THIS!

So how can we obtain information about the population mean, , without taking a census?

We can take a sample from the population then use inferential statistics (estimation and hypothesis testing) to
provide information on the population mean,  .

Recall in the Normal Distribution in Module 2, there are two parameters for this distribution, namely the mean
 and the variance  2. For this topic, our focus will be on the mean. We will begin by examining methods for
estimating the mean of the population.

DEFINITION

A point estimator of the population mean is the sample mean, denoted as X . It is defined as the sum of
the data values divided by the total number of sampled observations. The sample mean is computed using
the formula,
n

x i
X= i =1

where xi is the ith value of the variable X and n is the sample size.

Institute of Statistics, CAS, UPLB [1]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

Sampling distribution of X

Recall that a statistic is a random variable computed from sample values. Being a random variable, it follows
some probability distribution based on the samples of a given size from a population. This probability distribution
is specifically called the sampling distribution. All statistics such as sample mean, sample median, and sample
standard deviation have their own sampling distributions. For this topic, we will be talking about the sampling
distribution of the sample mean X .

Several theorems relate the sampling distribution of the sample mean to the distribution of the parent population.
The two most important theorems are as follows:

Theorem 1. Let a random variable X be normally distributed with mean  and variance  . The sampling
2

distribution of the sample mean X obtained from a random sample of size n is normal with mean
x =  and variance  x2 =  .
2

n
Theorem 2. (CENTRAL LIMIT THEOREM) Let a continuous random variable X have a distribution, say G, that
is non-normal with mean  and variance  2. The sampling distribution of the sample mean X
2
approaches the normal distribution with mean x =  and variance  2 = as the sample size
x n
n is increased without bounds.

DEFINITION

To summarize, Theorem 1 states that the sampling distribution of the sample mean is normally distributed
for any normally distributed parent population.

Theorem 1. If X ~ N(, 2), then X ~ N(, 2/n)

Theorem 2 states that even if the parent distribution is not normal, the sampling distribution of the sample
mean approximates a normal distribution as the sample size gets larger. The sampling distribution of the
sample mean will be approximately normal for sufficiently large n. In literature, n is considered sufficiently
large when n  25.
Theorem 2. If X ~ G(, 2), then X ≈ N(, 2/n) as n→∞
where G is any non-normal distribution

Given the theorems on the sampling distribution of X , you have learned the mean of the sampling distribution
of X is equal to the population mean ( x =  ). Therefore, X is called an unbiased estimator of the population
mean. In addition, the standard deviation of the sampling distribution of X is also known as the standard error,
which is a measure of precision of X .

Confidence Interval Estimator

Aside from getting a point estimate, we can also obtain an interval estimate of the population mean. A
confidence interval estimator indicates the accuracy of that estimate, and its length indicates the precision
of the estimate (the shorter the interval, the more precise the estimate). The length of the confidence interval is
determined by the margin of error, where the interval is expressed in the form X  Margin of Error.

The confidence interval estimator for  differs depending on whether  2 is known or unknown.

Institute of Statistics, CAS, UPLB [2]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

DEFINITION

For a population with known variance or standard deviation, the confidence interval estimator for  is given
by:
  
X Z /2  
 n

When using this estimator, the variable should be normally distributed, or that the sample size is large
enough for the Central Limit Theorem to apply.

The confidence level is in the form 1 – α, where α is the level of significance with a value between 0 and 1. To
find α, simply subtract the confidence level from 1. For example, if the confidence level is 90%, then α = 1 –
0.90 = 0.10.

The symbol Z denotes the Z-score that has area α to its right under the standard curve. The following are
common Z-score values used in obtaining interval estimators.

Table 4.1. Common Z-scores values with their

corresponding level of confidence.
Level of Confidence Z /2
90% 1.645
95% 1.96
98% 2.33
99% 2.575

DEFINITION

If the population variance or standard deviation is unknown, we cannot use the confidence interval given
above. In such cases, the population standard deviation σ is estimated by the sample standard deviation s.
The confidence interval estimator for  is given by:

 s 
X t /2( n −1)  
 n

The underlying distribution for this case is the Student’s t- distribution. Hence, we use t /2( n −1) instead of
Z /2 . Like the normal distribution, the Student’s t-distribution is symmetric at 0, asymptotic, and the total
area under the curve is 1. A t-distribution is identified by its degrees of freedom (df = n–1), and as the df
increases, the distribution resembles the standard normal curve. When using this estimator, the variable
should be normally distributed, or that the sample size is large enough for the Central Limit Theorem to
apply.

For a video discussion of the two confidence intervals above, you may watch these videos:
Confidence Interval for the Mean (σ Known)
Confidence Interval for the Mean (σ Unknown)

Institute of Statistics, CAS, UPLB [3]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

EXAMPLE 4.1 Chips Ahoy! Challenge*

*Example lifted from Weiss (2012)

Nabisco challenged students across the nation to confirm their claim that there are 1000
chocolate chips in every 18-ounce bag of Chips Ahoy! cookies, on average. It was reported
that one such group that participated in the challenge was an introductory statistics class at
the US Air Force Academy. Friends and families of the cadets sent 275 bags of Chips Ahoy!
cookies from all over the country. From the 275 bags, 39 were randomly selected for the
study. For each of the 39 bags selected for the study, the cadets dissolved the cookies in
water to separate the chips, and then counted the chips.
Icon made by Freepik from www.flaticon.com

In this example, the variable of interest is the number of chocolate chips in every 18-ounce bag of Chips Ahoy!
which is measured in the ratio scale. Suppose we want to describe the average of this variable using the mean,
 Given the p-value from the Shapiro-Wilk test equal to 0.4717, the assumption of normality of the data was
satisfied.

Parameter of interest:
 = mean number of chocolate chips in every 18-ounce bag of Chips Ahoy!

R COMMANDER OUTPUT R COMMANDER OUTPUT

mean sd se(mean) n Shapiro-Wilk normality test

1240.538 92.42546 14.79992 39
95 percent confidence interval: data: chips
1210.578 1270.499 W = 0.97332, p-value = 0.4717

Based on the sample of 39 bags, a point estimate of the mean number of chocolate chips in every 18-ounce
bag of Chips Ahoy! is 1240.538 with measure of precision 14.7999 (standard error). In addition, there is 95%
confidence that the true mean number of chocolate chips in every 18-ounce bag of Chips Ahoy! is between
1210.578 and 1270.499

Hypothesis Testing (Parametric Test)

Aside from estimation, testing of hypothesis is another major concern of inferential statistics. Recall in Module
1 that hypothesis testing is a technique used to determine whether a specific conjecture about the parameter(s)
or distribution of the population under study will be rejected or failed to be rejected. Results of a test of
hypothesis provide sound basis for decision-making in most applications.

For this topic, we will be making decisions given a hypothesized value of a population mean . And in doing so,
we will consider different test procedures and decide which test is the most appropriate to use given a problem.

To proceed with a parametric test, we need to check first if the data came from a normal distribution (recall
the normality tests in Module 3). The appropriate parametric test to use depends whether  is known or
unknown.

If  is known, the test procedure to be used is called the Z-test. It is given by the test statistic:
X − 0
Zc =

n
where X is the sample mean, 0 is the hypothesized value of the mean,  is the population standard deviation,
and n is the sample size.

Institute of Statistics, CAS, UPLB [4]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

If  is unknown, the test procedure to be used is called the t-test. It is given by the test statistic:
X − 0
tc =
s
n
where X is the sample mean, 0 is the hypothesized value of the mean, s is the sample standard deviation, and
n is the sample size.

By getting the p-value associated with these test statistics, we can now compare it with the desired level of
significance to decide whether we have evidence to reject the null hypothesis or not.

EXAMPLE 4.1 Chips Ahoy! Challenge (continued…)

Nabisco challenged students across the nation to confirm their claim that there are 1000 chocolate chips in
every 18-ounce bag of Chips Ahoy! cookies, on average. It was reported that one such group that participated
in the challenge was an introductory statistics class at the US Air Force Academy. Friends and families of the
cadets sent 275 bags of Chips Ahoy! cookies from all over the country. From the 275 bags, 39 were randomly
selected for the study. For each of the 39 bags selected for the study, the cadets dissolved the cookies in water
to separate the chips, and then counted the chips.

We have confirmed previously that the assumption on normality was satisfied for this data. Hence a parametric
test can be used to test the claim that there are indeed 1000 chocolate chips, on average, in every 18-ounce
bag of Chips Ahoy! cookies. The appropriate null and alternative hypotheses are:

Ho:  = 1000; The mean number of chocolate chips in every 18-ounce bag of Chips Ahoy! is 1000.
Ha:  ≠ 1000; The mean number of chocolate chips in every 18-ounce bag of Chips Ahoy! is not 1000.

R COMMANDER OUTPUT
Since  is unknown, we will use the t-test to test for the
Shapiro-Wilk normality test population mean.
data: chips
W = 0.97332, p-value = 0.4717

R COMMANDER OUTPUT

One Sample t-test

data: chips
t = 16.253, df = 38, p-value < 2.2e-16
alternative hypothesis: true mean is not equal to 1000

Given the p-value is less than 2.2 x 10-16, we reject the null hypothesis (Ho). Therefore, there is sufficient
evidence that the mean number of chocolate chips in every 18-ounce bag of Chips Ahoy! is not 1000.

Relationship between Hypothesis Testing and Confidence Interval

DEFINITION

We can also use the confidence interval for  to come up with conclusion for the population mean. Given a
two-tailed hypothesis test (alternative hypothesis is ≠), Ho will be rejected if and only if  0 (hypothesized
mean) lies outside the constructed confidence interval for .

Institute of Statistics, CAS, UPLB [5]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

Following the example from earlier, recall that the set of hypotheses is Ho:  = 1000 vs Ha:  ≠ 1000. Also, the
95% confidence interval for μ is [1210.578, 1270.499]. Since the hypothesized mean, which is 1000, is outside
the confidence interval, we reject Ho at 5% level of significance.

Hypothesis Testing (Nonparametric Test)

THINK ABOUT THIS!

What if the assumption(s) of the Z-test and t-test is(are) violated?

Suppose we wanted to use a Z-test or t-test but the normality assumption is not satisfied, or the variable of
interest is in ordinal scale (not quantitative). Under these conditions, parametric tests are not appropriate to use.
Instead, we can use nonparametric methods. Recall in Module 3 that nonparametric methods are usually
resistant to outliers and deviations from normality.

The Wilcoxon Signed-Ranks test is a nonparametric test which assumes that the variable has a symmetric
distribution. Since the mean and median of a symmetric distribution are equal, the Wilcoxon Signed-Ranks test
can be used to perform a hypothesis test for both population median and population mean (unless the variable
of interest is in ordinal scale, then only median can be used).

To get the Wilcoxon Signed-Rank test statistic, W = sum of positive ranks, the hypothesized value of the median
m0 should first be subtracted from each observation . This difference will then be made positive by taking its
absolute value. Then, we rank the absolute differences from smallest to largest. Give each rank the same sign
as the sign in the Difference column. W is obtained by taking the sum of the positive ranks. The procedure is
summarized in Table 4.2.

Table 4.2. Steps for obtaining the Wilcoxon Signed-Rank test statistic.
Difference Signed
Obs x |D| Rank of |D|
D = x - m0 Rank R

: : : : :

EXAMPLE 4.2 LAKE PH LEVEL

A lake is classified as non-acidic if its pH level is greater than 6. A group of UPLB researchers
measured the pH of 15 randomly selected lakes in CALABARZON. Do the data provide
sufficient evidence to conclude that, on the average, CALABARZON lakes are non-acidic?

We are interested in obtaining information on the average pH level of CALABARZON lakes,

which can be measured by either the mean or median. We first check the assumption of
normality to determine which test procedure is more appropriate to use.
Icon made by Freepik from www.flaticon.com

Institute of Statistics, CAS, UPLB [6]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

R COMMANDER OUTPUT

Shapiro-Wilk normality test Based on the Shapiro-Wilk test, the p-value is equal to
data: phLevel 0.03757. The assumption of normality is NOT satisfied, so
W = 0.87317, p-value = 0.03757 we cannot use a parametric test.

To check the distribution of the data, a histogram was

generated. The graph shows that the data is relatively
symmetric, therefore, we can proceed with the Wilcoxon
Signed-Ranks test.

R COMMANDER OUTPUT
Wilcoxon signed rank test with continuity correction
data: phLevel
V = 82.5, p-value = 0.1054
alternative hypothesis: true location is greater than 6

Given the p-value equal to 0.1054, we fail to reject Ho. Therefore, there is NO sufficient evidence that the
average pH level of CALABARZON lakes is greater than 6 (non-acidic).

ACTIVITY 4.1 CLASSROOM MOTIVATION PLAN

Instruction: Provide the missing information in the analysis to make a recommendation on the given problem.

A group of educators in a certain institution conducted a study about their student’s motivation
in learning. The study aims to craft a classroom motivation plan in improving student’s
learning outcomes. To do this, a simple random sample consisting of 169 students were
chosen and were asked to answer questions on perception of learning, classroom and school
community inventory, academic motivation, alienation, and academic self-concept. The data
can be found in the MS Excel file Activity 4.1 Motivation.xlsx.
Icon made by Freepik from www.flaticon.com

1. Some learning activities must be performed in a group. However, this is not good for students who do not
consider themselves as “team players”. In this regard, the students’ alienation, which is the state of being
isolated from a group/activity which they are involved with, will be analyzed to know the potentials of
grouping activities as part of the motivation plan.
i. Provide a point estimate (with measure of precision) of the true mean alienation score of the students. If
possible, construct a 95% CI of the parameter.
ii. An average score of at least 100 means a high degree of alienation. Is there enough evidence to say that
the students have a high degree of alienation? Help the group of educators decide by providing necessary
statistical results.

Institute of Statistics, CAS, UPLB [7]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

R COMMANDER OUTPUT R COMMANDER OUTPUT

Shapiro-Wilk normality test mean sd se(mean) n
data: alienation 67.13609 11.24136 0.8647203 169
W = 0.9892, p-value = 0.2253 95 percent confidence interval:
65.42898 68.84321

In this problem, the variable of interest is the alienation score

of the student which is measured in the ________ scale. The
assumption of normality for this score is (satisfied/not
satisfied) given p-value equal to ________. A point estimate
of the true mean alienation score of the students is ________
with measure of precision _________. There is 95%
confidence that the true mean alienation score of the students
is between ________ and ________.

R COMMANDER OUTPUT R COMMANDER OUTPUT

One Sample t-test Wilcoxon signed rank test with

data: alienation continuity correction
t = -38.005, df = 168, p-value < 2.2e-16 data: alienation
V = 0, p-value < 2.2e-16

It is of interest to test the hypotheses:

Ho: The average alienation score of the students is at least 100.
Ha: The average alienation score of the students is _________.

To determine which test to use for this set of hypotheses, the assumption of normality was tested. Given the
software output, this assumption is (satisfied/not satisfied) given p-value equal to ________. Thus, the (name
of the test) should be used to answer the objective of the researchers.

Using (name of the test) with p-value equal to ________, the data (does/does not) provide sufficient evidence
that ____________________________________________. Therefore, the educators (should/should not)
include learning activities which are done in groups.

2. To further develop the classroom motivation plan, the group thought of incorporating
extrinsic type of motivation in the plan. Extrinsic motivation refers to the behavior that
is driven by external rewards such as money, fame, grades, and praise. Based on
past studies using the same instrument, the ideal average score for extrinsic
motivations is 55. If the average score of the students meets the ideal score, extrinsic
type of motivation will be included in the plan. If not, they will consider a different
motivation approach.
Icon made by Freepik from www.flaticon.com

Institute of Statistics, CAS, UPLB [8]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

R COMMANDER OUTPUT
R COMMANDER OUTPUT
Shapiro-Wilk normality test One Sample t-test
data: extr_mot data: extr_mot
W = 0.97609, p-value = 0.005112 t = 8.682,df = 168, p-value = 3.282e-15

R COMMANDER OUTPUT
Wilcoxon signed rank test with
continuity correction
data: extr_mot
V = 11462, p-value = 2.707e-13

In this problem, the variable of interest is the extrinsic motivation score of the student which is measured in the
________ scale. The assumption of normality for this score is (satisfied/not satisfied) given p-value equal to
________. Hence, to test the hypotheses,

Ho: The average score for extrinsic motivation of the students is 55.
Ha: The average score for extrinsic motivation of the students is _________.

the (name of the test) should be used. Given the p-value equal to ________, data (does/does not) provide
sufficient evidence that ____________________________________________. Therefore, the educators
_________________________________________.

Institute of Statistics, CAS, UPLB [9]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

Level of Measurement Interval/Ratio Scale Ordinal Scale

of the Variable

Is the
Yes No
assumption of
NORMALITY
satisfied?

Yes No
Known Symmetric
σ? distribution?

Yes

Z-test t-test Wilcoxon Signed-Ranks Test

Figure 4.1 Summary of statistical tests for Topic 4A.

Institute of Statistics, CAS, UPLB [ 10 ]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

TOPIC 4B. ESTIMATION AND TEST OF HYPOTHESIS ON ONE POPULATION VARIANCE

ILLUSTRATION

With queues on several windows, a bank manager finds that the standard deviation of the normally
distributed waiting times on Friday afternoons is 6.2 minutes. The bank tried a single main queue and finds
that for a random sample of 25 customers, the waiting times have a standard deviation of 3.8 minutes. At
5% level of significance, test the claim that a single queue gives a lower variation in waiting time.

For this illustration, we are now interested to describe the spread or variability of the values of a variable.
Recall in the previous topic that there are two parameters for the Normal distribution, namely the mean  and
the variance  2. For this topic, our focus will be on the variance. We will begin by examining methods for
estimating the variance of the population.

DEFINITION

A point estimator of the population variance is the sample variance, denoted as s2. It is defined as:

( X )
n 2
i −X
s2 = i =1

n −1
where xi is the ith value of the variable X, X is the sample mean, and n is the sample size.

The sampling distribution of s2 is Chi-square distribution with n-1 degrees of freedom.

Where xi is the ith value of the variable X and n is the sample size.
Confidence Interval Estimator

We can also obtain an interval estimate of the population variance.

DEFINITION

The confidence interval estimator for  2 is given by:

 ( n − 1) s 2 ( n − 1) s 2 
 2 , 2 
  /2( n −1) 1− /2( n −1) 

By using this estimator, the variable should be normally distributed, or that the sample size is large enough
for the Central Limit Theorem to apply.

Hypothesis Testing (Parametric Test)

For this topic, we will be making decisions given a hypothesized value of a population variance  2. Before we
proceed with a parametric test, we need to check if the data came from a normal distribution (recall the
normality tests in Module 3). If the assumptions are satisfied, we can use the test procedure Chi-square test
for one population variance with the following test statistic:

c2 =
( n − 1) s 2
 02
where s2 is the sample variance,  02 is the hypothesized value of the variance, and n is the sample size.

Institute of Statistics, CAS, UPLB [ 11 ]

Module 4.1 [STAT 101 MODULE HANDOUT]
Inference on Single Population

EXAMPLE 4.3 CEREAL BOXES

At a cereal filling plant, quality control engineers do not want the variance of weights of 750-
gram cereal boxes to exceed 100 grams 2. A random sample of seven boxes with a nominal
weight of 750 grams gave the following weights: 775, 780, 781, 795, 803, 810, 823. Verify if
the plant is working within the specified variance.

Icon made by Freepik from www.flaticon.com

In this example, the variable of interest is the weight of a 750-gram cereal box which is measured in the ratio
scale. Suppose we want to describe the spread of this variable using the variance,  2 Given the p-value from
the Shapiro-Wilk test equal to 0.6055, the assumption of normality of the data was satisfied.

Parameter of interest:
 2 = variance of weights of 750-gram cereal boxes

R COMMANDER OUTPUT R COMMANDER OUTPUT

var_estimate lower_CI upper_CI Shapiro-Wilk normality test

315.5714 131.0388 1530.2359 data: weight
W = 0.93629, p-value = 0.6055

Based on the sample, a point estimate of the variance of weights of 750-gram cereal boxes is 315.5714. There
is 95% confidence that the true variance of weights of 750-gram cereal boxes is between 131.0388 and
1530.2359

We use the Chi-square test to test the hypotheses:

Ho: σ2 ≤ 100; The variance of weights of 750-gram cereal boxes is at most 100 grams 2.
Ha: σ2 > 100; The variance of weights of 750-gram cereal boxes is greater than 100 grams2.

R COMMANDER OUTPUT

One sample Chi-squared test for variance

data: weight
X-squared = 18.934, df = 6, p-value = 0.004276
alternative hypothesis: true variance is greater than 100

Given the p-value equal to 0.004276, we reject the null hypothesis (Ho). There is sufficient evidence that
variance of weights of 750-gram cereal boxes is greater than 100 grams2. Thus, the cereal filling plant is not
working within the specified variance and may cause irregularities in their products.

REFERENCE
WEISS, N. A. (2012) Introductory Statistics. 9th Ed. Pearson Education, Inc.

Institute of Statistics, CAS, UPLB [ 12 ]