Reading Material 1 SST/BST 3133 Regression Analysis
Introduction to Regression Analysis
In many statistical investigations from various disciplines such as marketing research, social sciences,
and medical research, the objective is to express relationships between two (or more) variables in
mathematical form. For instance, potential sales of a new product depend on its price and household
expenditure depends on the monthly income of the household. Simply, determining this relationship
between variables can be considered as ‘regression’.
Historical Origin of the term “Regression”
The Modern Interpretation of Regression
However, in reality, the true association between response variable and predictor variables is
never linear but approximation to that true relationship is considered as linear regression
Reading Material 1 SST/BST 3133 Regression Analysis
Statistical Versus Deterministic Relationships
Reading Material 1 SST/BST 3133 Regression Analysis
Correlation Analysis
Correlation analysis is used to study the strength of the relationship between two or more
variables. Correlation can be linear or nonlinear and it can also be positive or negative.
Correlation (or bivariate/simple correlation) is a measure of the relationship between the two
variables. The coefficient of correlation measures the magnitude and direction of the relationship.
Scatter Plot
The scatter plot, drawn for the identified independent and dependent variables depicts the
following.
• Whether the relationship is linear?
• Whether that linear relationship is positive, negative, or constant.
• Whether that linear relationship is strong, weak, or moderate.
Reading Material 1 SST/BST 3133 Regression Analysis
When it is only necessary to obtain a rough idea of the nature of the relationship between the
response variable and the predictor variable/s, the equation of the simple linear regression model
can be inferred by drawing a straight line freely on the scatter plot, known as the free-hand
method. Note: In this course unit, linear regression will only be taken into consideration.
Coefficient of Correlation
Reading Material 1 SST/BST 3133 Regression Analysis
Properties of a Population Correlation Coefficient (𝝆𝑿𝒀)
1. −1 +1
2. XY = 1 (perfect positive linear relationship between X and Y)
3. XY = 1 (perfect negative linear relationship between X and Y)
4. 0 XY = There is no linear relationship between X and Y
5. Zero correlation does not imply independence. However, independence implies zero correlation.
Reading Material 1 SST/BST 3133 Regression Analysis
Reading Material 1 SST/BST 3133 Regression Analysis
Spearman’s Rank Correlation Co-efficient.
Reading Material 1 SST/BST 3133 Regression Analysis
