Econometrics (ECO 2400)

Fundamentals of Mathematical Statistics

23 January 2026

Tanya Jain Lecture II 23 January 2026 1 / 41

Population and sample

A population refers to the entire set of individuals, objects, or data points

that you want to study.
A sample is a subset of the population that is selected for analysis. It’s
used when studying the entire population is impractical or impossible.
To use sample data to make inferences about an entire population, it
is necessary to make some assumptions about the relationship
between the two.
If X1 , ..., Xn are independent random variables having a common
distribution F, then we say that they constitute a sample (sometimes
called a random sample) from the distribution F.
Random samples can be drawn from any distribution. Most commonly
used: normal distribution (2 parameter: mean and variance)
The quantities µ and σ 2 are called the population mean and the
population variance, respectively.

Tanya Jain Lecture II 23 January 2026 2 / 41

Sampling distribution

Let X1 , X2 , ..., Xn be a sample of independent values from the

population.
The sample mean is defined by:
X1 + X2 + ....Xn
X̄ =
n
Since the value of the sample mean X̄ is determined by the values of
the random variables in the sample, it follows that X̄ is also a random
variable.  
X1 + ....Xn
E [X̄ ] = E =µ
N
σ2
 
¯ X1 + ....Xn
Var (X ) = Var =
N n
Hence, the expected value of the sample mean is the population mean
µ whereas its variance is 1/n times the population variance.

Tanya Jain Lecture II 23 January 2026 3 / 41

Properties of estimators

Let X = (X1 , X2 , ...Xn ) be a sample from a population whose

distribution is specified up to an unknown parameter θ.
Also, θ̂ be an estimator of θ.
The question is how to determine whether θ̂ is worth an estimator of
θ?
▶ Unbiased
▶ Efficient

Tanya Jain Lecture II 23 January 2026 4 / 41

Unbiased Estimator

A point estimator θ̂ is said to be unbiased estimator, if E (θ̂) = θ.

If an estimator is not unbiased, then the
bias = {E (θ̂) − θ}

Example 1
Suppose X ∼ N (µ, σx2 ) and we draw a random sample of size n from this
population.
The following are two estimates of µ:
P
Xi
(i) X =
n
P
∗ X
(ii) X = .
n+1
Check, if the above estimators are unbiased.

Tanya Jain Lecture II 23 January 2026 5 / 41

Efficient Estimator

The estimator which has lower variance comparatively is considered

more efficient than the others.
Example 2
If X1 , X2 and X3 constitute a sample of size 3 from normal population with
the mean µ and the variance σ 2 , find the efficiency of:
X1 + 2X2 + X3 X1 + X2 + X3
relative to
4 3

Tanya Jain Lecture II 23 January 2026 6 / 41

Mean square error

Sometimes, one estimator is unbiased and other has minimum

variance.
In that case, we require a tool that combines minimum variance and
unbiased principles.
This combined is achieved minimising MSE (mean square error).

MSE = E (θ̂ − θ)2

MSE = Variance of the estimator + (bias)2

Example 3
Given two estimators t1 and t2 of the population mean (µ = 10)
E (t1 )=10, E (t2 )= 11, V (t1 ) = 4 and V (t2 ) = 1.
Which estimator is better and why?

Tanya Jain Lecture II 23 January 2026 7 / 41

Large Samples

The Law of Large Numbers asserts that as the number of trials or samples
increases, the observed outcomes tend to converge closer to the expected
value and the variability reduces.

But they are costly!!

Tanya Jain Lecture II 23 January 2026 8 / 41

Point estimation

A point estimate is a single value used to approximate an unknown

population parameter.

This may not always be accurate as it relies on a single value.

Hence, interval estimation is better.

It rovides a range of values to account for sampling variability.

Tanya Jain Lecture II 23 January 2026 9 / 41

Interval estimation

Suppose, X1 , X2 , ..., Xn is a sample from a normal population having

unknown mean µ and known variance σ 2 .

Now, we can’t expect that the prediction of sample mean will be

exact true to our estimate, rather it will ”be close”.

Hence, it is sometimes valuable to be able to specify an interval for

which we have a certain degree of confidence that µ lies within.

Tanya Jain Lecture II 23 January 2026 10 / 41

Confidence interval for mean

Confidence Interval for a Normal Mean When the Variance Is

Known
σ2
X̄ is normal with mean µ and variance , it follows that:
n
X̄ − µ
√
σ/ n
has a standard normal distribution.
Therefore,
√ x̄ − µ
 
P −1.96 < n < 1.96 = 0.95
σ
After solving this, “with 95 percent confidence” we assert that the
√
true mean lies within 1.96 σ/ n of the observed sample mean.
The interval -  
σ σ
x̄ − 1.96 √ , x̄ + 1.96 √
n n
is called a 95% confidence interval estimate of µ.
Tanya Jain Lecture II 23 January 2026 11 / 41
Confidence interval for mean

Confidence Interval for a Normal Mean When the Variance Is

Unknown

X1 , ..., Xn is a sample from a normal distribution with unknown mean

µ and unknown variance σ 2 .
Since σ is unknown, we can no longer base our interval on the fact
√ X̄ − µ
that n is a standard normal random variable.
σ
So, for constructing a 100(1 − α) percent confidence interval for µ,
we will use the fact that :
√ X̄ − µ
n
S
is a t-random variable with (n − 1) degrees of freedom.
n
(Xi − X̄ )2
P

where, S = i=1
(n − 1)
Tanya Jain Lecture II 23 January 2026 12 / 41
Confidence interval for mean

√ X̄ − µ
 
P −tα/2,n−1 < n < tα/2,n−1 = 1 − α
S
 
s s
µ ∈ x̄ − tα/2,n−1 √ , x̄ + tα/2,n−1 √
n n

It tells us that the probability that the random interval contains the
population mean is 100(1 − α)%.

Tanya Jain Lecture II 23 January 2026 13 / 41

Confidence interval for mean

Example 4
Suppose that when a signal having value µ is transmitted from location A
the value received at location B is normally distributed with mean µ and
variance 4. That is, if µ is sent, then the value received is µ + N where N,
representing noise, is normal with mean 0 and variance 4. To reduce error,
suppose the same value is sent 9 times. If the successive values received
are 5, 8.5, 12, 15, 7, 9, 7.5, 6.5, 10.5, construct a 95 percent confidence
interval for µ.

Tanya Jain Lecture II 23 January 2026 14 / 41

Central limit theorem

What if the underlying population is non-normal

Here CLT becomes useful.

As the sample size increases and its variance is finite, then the distribution
of the sample mean approaches normal distribution irrespective of the
shape of the population distribution.

Tanya Jain Lecture II 23 January 2026 15 / 41

Hypothesis Testing

A test of hypotheses is a method that uses sample data to decide

between two competing claims (hypotheses) about a population
characteristic.
Null hypothesis: A claim about a population characteristic that is
initially assumed to be true. The null hypothesis is denoted by H0 .
Alternative hypothesis: A competing claim about a population
characteristic. The alternative hypothesis is denoted by Ha .
In carrying out a test of H0 versus Ha , the null hypothesis H0 will be
rejected in favor of Ha only if the sample provides convincing evidence
that H0 is false.
If the sample does not provide such evidence, H0 will not be rejected.

Tanya Jain Lecture II 23 January 2026 16 / 41

Hypothesis testing

The form of a null hypothesis is

H0 : population characteristic = hypothesized value
where the hypothesized value is a specific number determined by the
problem context.

The alternative hypothesis will have one of the following three forms:
Ha : population characteristic > hypothesized value
Ha : population characteristic < hypothesized value
Ha : population characteristic = hypothesized value

Tanya Jain Lecture II 23 January 2026 17 / 41

Testing hypothesis

Steps in a hypothesis test:

1 Describe the population characteristic of interest.
2 State the null hypothesis H0 .
3 State the alternative hypothesis Ha .
4 Select the significance level a for the test.
5 Display the test statistic to be used, with substitution of the
hypothesized value identified in Step 2.
6 Calculate all quantities appearing in the test statistic and the value of
the test statistic itself.
7 Determine the P-value associated with the observed value of the test
statistic.
8 State the conclusion (which is reject H0 or fail to reject H0 otherwise).
The conclusion should then be stated in the context of the problem.

Tanya Jain Lecture II 23 January 2026 18 / 41

Hypothesis Testing

METHOD I
Using critical region.

Tanya Jain Lecture II 23 January 2026 19 / 41

Hypothesis Testing

METHOD II
Using p-value.
▶ p- value is the probability value: area/ region/ size of test statistical
value.
▶ In right tail test: p-value is the area to the right of test statistical value.

Tanya Jain Lecture II 23 January 2026 20 / 41

Hypothesis Testing

In left tail test: p-value is the area to the left of test statistical value.

Tanya Jain Lecture II 23 January 2026 21 / 41

Hypothesis Testing
In 2-tail test:
p-value = 2(area to right of test statistical value) iff T.S value is
positive)
p-value = 2(area to left of test statistical value) iff T.S value is
negative)

Conclusion
p-value > α: do not reject H0
p-value < α : reject H0
Tanya Jain Lecture II 23 January 2026 22 / 41
Test for population mean

CASE I: Variance is known.

x̄ − hypothesised value
Test statistic: z = σ
√
n
p-value: compound as an area under the z-curve.

Tanya Jain Lecture II 23 January 2026 23 / 41

Test for population mean

Example 5
All cigarettes presently in the market have an average nicotine content of
at least 1.6 mg per cigarette. A firm that produces cigarettes claims that
it has discovered a new way to cure tobacco leaves that will result in the
average nicotine content of a cigarette being less than 1.6 mg. To test this
claim, a sample of 20 of the firm’s cigarettes were analyzed. If it is known
that the standard deviation of a cigarette’s nicotine content is .8 mg, what
conclusions can be drawn, at the 5 percent level of significance, if the
average nicotine content of the 20 cigarettes is 1.54?

Tanya Jain Lecture II 23 January 2026 24 / 41

Test for population mean
CASE II: Variance is unknown.
x̄ − hypothesised value
Test statistic: t = s
√
n
p-value: compound as an area under the t-curve with (n-1) degrees of
freedom.

Tanya Jain Lecture II 23 January 2026 25 / 41

Test for population mean

Example 6
It is assumed that the mean systolic blood pressure is 120 mm Hg. In a
Heart Study, a sample of 100 people had an average systolic blood
pressure of 130.1 mm Hg with a standard deviation of 21.21 mm Hg. Is
the group significantly different (with respect to systolic blood pressure)
from the regular population?
(t0.025,99 = 1.960)

Tanya Jain Lecture II 23 January 2026 26 / 41

Testing the equality of means of two populations

CASE I: Variances are known.

Suppose that X1 , ..., Xn and Y1 , ..., Ym are independent samples from
normal populations having unknown means µx and µy but known
variances σx2 and σy2 .
Consider the problem of testing the hypothesis:
H0 : µ x = µ y
Ha : µx ̸= µy
Since X̄ is an estimate of µx and Ȳ of µy , it follows that (X̄ − Ȳ )
can be used to estimate µx − µy .
Because the null hypothesis can be written as H0 : µx − µy = 0, it
seems reasonable to reject it when (X̄ − Ȳ ) is far from zero.

Tanya Jain Lecture II 23 January 2026 27 / 41

Testing the equality of means of two populations

σ 2 σy2
 
X̄ − Ȳ ∼ N µx − µy , x +
n m
(X̄ − Ȳ ) − (µx − µy )
r ∼ N (0, 1)
σx2 σy2
+
n m
Since, the hypothesised value for difference in means is zero.
The test statistic is calculated as follows:
(X̄ − Ȳ )
z= r
σx2 σy2
+
n m

Tanya Jain Lecture II 23 January 2026 28 / 41

Testing the equality of means of two populations

Example 7
A consumer advocacy group wants to study consumer satisfaction with
their shopping experience at the country’s two biggest retailers. The group
surveyed consumers and asked them to rate one of the retailers in a
number of different categories. An overall satisfaction score out of 100
summarized the responses for each consumer sampled. In a sample of 35
consumers for retailer A, the average overall satisfaction score was 79. In a
sample of 30 consumers for retailer B, the average overall satisfaction
score was 71. Based on prior experience with the satisfaction rating scale,
the population standard deviation for retailer A is assumed to be 10 and
the population standard deviation for retailer B is assumed to be 12.
Is there evidence to suggest that the mean satisfaction score for retailer A
is different than the mean satisfaction score for retailer B?

Tanya Jain Lecture II 23 January 2026 29 / 41

Testing the equality of means of two populations

CASE II: Variances are unknown, but equal. (σx2 = σy2 )

(X̄ − Ȳ ) − (µx − µy )
q ∼ tn+m−2
SP2 (1/n + 1/m)

where, Sp2 , is the pooled estimator of the common variance σ 2 , given by:

(n − 1)Sx2 + (m − 1)Sy2
Sp2 =
n+m−2
When H0 is true, and so µX − µy = 0, the test statistic is:

(X̄ − Ȳ )
t= q with (n+m-2) degrees of freedom.
SP2 (1/n + 1/m)

Tanya Jain Lecture II 23 January 2026 30 / 41

Testing the equality of means of two populations

Example 8
A psychologist was interested in exploring whether or not male and female
college students have different driving behaviors. She opted to focus on
the fastest speed ever driven by an individual. She conducted a survey of a
random male college students and a random female college students. Here
is a descriptive summary of the results of her survey:

Is there sufficient evidence at the α = 0.05 level to conclude that the

mean fastest speed driven by male college students differs from the mean
fastest speed driven by female college students? (Assume, that the
population variances are equal) (t0.025,61 = 1.960)

Tanya Jain Lecture II 23 January 2026 31 / 41

Testing the equality of means of two populations

CASE III: Variances are unknown and also unequal. (σx2 ̸= σy2 )
Since, Sx2 is the natural estimator of σx2 and Sy2 of σy2 .
The test statistic will be:
X̄ − Ȳ
z= r
Sx2 Sy2
+
n m

Example 9
Repeat example 8 assuming that the population variances are not equal.

Tanya Jain Lecture II 23 January 2026 32 / 41

Tests About a Population Proportion

Let p denote the proportion of individuals or objects in a population

who possess a specified property.
. A random sample of n individuals is selected from the population.
The sample proportion-
number in the sample that possess the property
p̂ =
n
serves as the basis for testing hypotheses about p.

Tanya Jain Lecture II 23 January 2026 33 / 41

Tests About a Population Proportion

The large-sample test procedure is based on the same properties of

the sampling distribution of p, that were used to obtain a confidence
interval for p:
µp = p
r
p(1 − p)
σp =
n
When n is large, the sampling distribution of p is approximately
normal.
The standardized variable-
p̂ − p
z= r
p(1 − p)
n
has approximately a standard normal distribution when n is large.

Tanya Jain Lecture II 23 January 2026 34 / 41

Tests About a Population Proportion

Tanya Jain Lecture II 23 January 2026 35 / 41

Tests About a Population Proportion

Example 10
Obesity is an increasing problem in America among all age groups. The
article “Factors Affecting Obesity and Waist Circumference Among U.S.
Adults” reported that 686 individuals in a sample of 2014 adult men were
found to be obese. An earlier survey based on people’s own assessment
revealed that 20% of adult Americans considered themselves obese. Does
the recent data suggest that the true proportion of men who are obese is
more than 1.5 times the percentage from the self-assessment survey? Let’s
carry out a test of hypotheses using a significance level of .10.

Tanya Jain Lecture II 23 January 2026 36 / 41

Errors in hypothesis testing

Just as a jury may reach the wrong verdict in a trial, there is some
chance that using a test procedure with sample data may lead us to
the wrong conclusion about a population characteristic.
One incorrect conclusion in a criminal trial is for a jury to convict an
innocent person. Another is for a guilty person to be set free.
Similarly, there are two different types of errors that might be made
when making a decision in a hypothesis test.
Type I error: The error of rejecting H0 when H0 is true.
The probability of type I error is denoted by α and is also called the
significance level of the test.
Type II error: The error of failing to reject H0 when H0 is false. The
probability of type II error is denoted by β

Risk of error is the price paid for basing the decision on sample data.

Tanya Jain Lecture II 23 January 2026 37 / 41

Errors in hypothesis testing

Tanya Jain Lecture II 23 January 2026 38 / 41

Errors in hypothesis testing

Example 11
The U.S. Bureau of Transportation Statistics reports that for 2015, 79.9%
of all domestic passenger flights arrived within 15 minutes of the scheduled
arrival time (Air Travel Consumer Reports, February 2016). Suppose that
an airline with a poor on-time record decides to offer its employees a
bonus if the airline’s proportion of on-time flights exceeds the overall
industry rate of 0.799 in an upcoming month. We can use p to represent
the actual proportion of the airline’s flights that are on time during the
month of interest.

Tanya Jain Lecture II 23 January 2026 39 / 41

Errors in hypothesis testing

−→H0 : p = 0.799 and Ha : p > 0.799

In this context, a Type I error (rejecting a true H0 ) is concluding that the
airline on-time rate exceeds the overall industry rate, when in fact the
airline does not have a better record. This Type I error would result in the
airline rewarding its employees when the actual proportion of on-time
flights was not actually greater than 0.799.
A Type II error (not rejecting a false H0 ) is not concluding that the
airline’s on-time proportion is greater than the industry proportion when
the airline really did have a better on-time record. A type II error would
result in the airline employees not receiving a reward that they deserved.

Tanya Jain Lecture II 23 January 2026 40 / 41

Power of the test

The power of a hypothesis test is the probability of making the

correct decision if the alternative hypothesis is true.
That is, the power of a hypothesis test is the probability of rejecting
the null hypothesis H0 when the alternative hypothesis Ha is the
hypothesis that is true.

Power of the test = 1 − β

Tanya Jain Lecture II 23 January 2026 41 / 41