Basic Terminology of Statistical Inference

Inference

Inference is the process of drawing a conclusion based on evidence, observations, or known

facts.

Example: You see a boy walking fast with a backpack in the morning.
Inference: He is probably going to school.

Statistical Inference

Statistical inference is the process of using sample data to make conclusions or predictions about
a larger population.

Example: Suppose you want to know the average height of all students in your school.
You can’t measure every student, so you measure the heights of 50 randomly chosen students.
You find that the average height of your sample is 5.4 feet. You then infer that the average height
of all students in the school is also about 5.4 feet.

Types of Statistical Inference

i) Deductive inference

ii) Inductive inference

Deductive inference
Deductive inference is a type of reasoning where you start with general rules or facts, and use them
to reach a specific, logical conclusion. In fact this type of inference has no statistical inference.

Example: Premise: All humans are mortal. Rahim is human being. So, Rahim is Mortal. This
conclusion is 100% logically true if premises are true. This is deductive inference.

Inductive Inference
The inference which is obtained by the basis of empirical or experimental activities is called
inductive inference.

Example: You observe:

1. Observation 1: The first mango you ate was sweet.

2. Observation 2: The second mango was also sweet.

3. Observation 3: The third mango was sweet too.

So the inductive inference is All mangoes are sweet.

Distinguish between deductive inference and inductive inference

Aspect Deductive Inference Inductive Inference

Drawing a specific and certain Drawing a general and probable

Definition conclusion from general premises or conclusion based on specific observations
facts. or examples.

Reasoning
General → Specific Specific → General
Direction

Certainty of Conclusion is logically certain if

Conclusion is probable, not guaranteed.
Conclusion premises are true.

Starting Point Begins with a general rule or law. Begins with specific observations or data.

The sun has risen in the east every day

All humans are mortal. Socrates is a
Example observed. Therefore, the sun always rises
human. So, Socrates is mortal.
in the east.

Mathematics, formal logic, proofs, Science, everyday reasoning, hypothesis

Used in
law. formation.

Nature of
Definitive and necessary. Tentative and based on likelihood.
Conclusion

Estimator: The statistic which is used to estimate some unknown parameter of population is
known as estimator.

Example: You randomly pick 10 apples from the basket. You weigh those 10 apples and find their
average weight.

 The average weight of those 10 apples is called the estimate.

 The formula or method you use to calculate this average is the estimator.

 You use this estimate to guess the average weight of all apples in the basket.
Estimate: An estimate is the actual number or value you get when you use an estimator (a formula
or method) on your sample data to guess a population parameter.
Example: Imagine you want to know the average number of hours students study per day at your
school. You ask 20 students and find their average study time is 3.5 hours. Here, 3.5 hours is the
estimate of the average study time for all students in the school.
Estimation: The procedure of estimating the population parameter on the basis of sample
observation by some technique or any statistical device is called estimation.

Example: A teacher wants to know the average score of her entire class on a recent exam, but she
doesn’t have time to check every student’s paper. Instead, she randomly selects 10 papers,
calculates the average score of those, and uses this average to estimate the class’s overall
performance. This process of using a small sample to guess the whole class’s average score is
called estimation.

Types of Estimation:

i) Point estimation
ii) Interval estimation

Point estimation: Point estimation is a method of estimating a population parameter by using a

single value calculated from sample data.
Example: A researcher wants to find the average weight of dogs in a city. She weighs 40 randomly
selected dogs and calculates their average weight to be 25 kg. This single value, 25 kg, is called
the point estimate of the average weight of all dogs in the city.

Interval estimation: Interval estimation gives a range of values within which the true population
parameter is expected to lie, along with a level of confidence (like 95%).

Example: A nutritionist wants to estimate the average calorie intake of adults in a city. After
sampling 100 people, she calculates that the average calorie intake is 2,200 calories, and she is
95% confident that the true average lies between 2,100 and 2,300 calories. This range is called a
confidence interval, and the process of finding it is interval estimation.

Mean Square Error (MSE): Mean Square Error (MSE) measures the average of the squares of
the errors—that is, the average squared difference between estimated values and the actual value.

MSE is defined as M.S.E= E(t − θ)

Example: A weather forecaster predicts the temperature for five days, and the actual temperatures
are known. The forecaster’s predictions are compared to the real temperatures, and the squared
differences are calculated and averaged. This average of squared errors is called the Mean Square
Error (MSE), showing how accurate the predictions are—the smaller the MSE, the better the
predictions.
Distinguish between Point Estimation vs. Interval Estimation

Feature Point Estimation Interval Estimation

A single value used to estimate a A range of values used to estimate a

Definition
population parameter population parameter

Output One number (e.g., 5.6) A range (e.g., 5.4 to 5.8)

Certainty No measure of confidence Provides a confidence level (e.g., 95%)

Sample mean = 5.6 → estimate of

Example 95% confidence interval: 5.4 to 5.8
population mean

Precision More precise, but less reliable Less precise, but more reliable

Better for decision-making under

Usefulness Quick, simple estimate
uncertainty

There are two methods of finding point estimator

1) Classical method

 Method of maximum Likelihood.

 Method of moments.
 Method of least squares.
 Method of chi-square.
 Method of modified chi-square.
2) Non- Classical method

 Method of bayes estimate.

 Method of minimax estimate.

Properties of a Good Estimator

A good estimator should have certain desirable properties to ensure it provides reliable and
accurate estimates of population parameters.

 Unbiasedness
 Consistency
 Efficiency
 Sufficiency