Actuarial Statistics · Linear regression models

Simple Linear Regression Model and Assumptions

Updated 11 October 2026 · Fact-checked

The simple linear regression model is Y = α + βx + ε. Y is the response, x is the explanatory variable, and ε is a random error. You assume the errors have mean zero, constant variance and are uncorrelated. For inference you also assume they are normal. Check these assumptions before you trust any result.

Understand Simple Linear Regression Model and Assumptions

Regression models how one variable changes when another changes. You want to explain or predict a response variable Y using an explanatory variable x. In CS1 the response is random. The explanatory variable is treated as fixed and known, not random.

The simple linear regression model says Yᵢ = α + βxᵢ + εᵢ for i = 1, 2, …, n. Here α is the intercept and β is the slope. The term εᵢ is the error. It captures everything that moves Y but is not in the straight line. Without it, every point would lie exactly on the line, which never happens with real data.

The model is called linear because it is linear in the parameters α and β. The line is the mean of Y at each x: E[Yᵢ] = α + βxᵢ. So β is the average change in Y for a one-unit rise in x. And α is the average of Y when x = 0, which may have no practical meaning if x = 0 is outside the data range.

The assumptions are all about the errors. They have mean zero. They have the same variance σ² at every x, called homoscedasticity. They are uncorrelated, and often taken as independent. For confidence intervals and tests you add that they are normally distributed, so εᵢ ~ N(0, σ²). Then Yᵢ ~ N(α + βxᵢ, σ²).

The assumptions matter because the formulas for standard errors, t-tests and intervals depend on them. Least squares gives the line even without normality. But the inference built on it is only valid when the assumptions hold reasonably well.

Key rules to remember

Simple linear regression model: Yᵢ = α + βxᵢ + εᵢ, i = 1, …, n; Y is the response, x the explanatory variable (fixed), ε the random error.
Mean of the response: E[Yᵢ] = α + βxᵢ; Follows because E[εᵢ] = 0 and xᵢ is fixed.
Variance of the response: Var(Yᵢ) = σ²; Same at every xᵢ. This is constant variance.
Error assumptions: E[εᵢ] = 0, Var(εᵢ) = σ², Cov(εᵢ, εⱼ) = 0 for i ≠ j; These are the basic assumptions for least squares to behave well.
Normal error model: εᵢ ~ N(0, σ²) independent, so Yᵢ ~ N(α + βxᵢ, σ²); Needed for exact t-tests and confidence intervals.

How to solve Simple Linear Regression Model and Assumptions questions

Use this method for any question that asks you to state, interpret or check the model and its assumptions.

1Identify the response Y and the explanatory variable x from the context. The thing you want to predict or explain is Y.
2Write the model in full: Yᵢ = α + βxᵢ + εᵢ. Say what α, β and εᵢ mean in the context given.
3State the error assumptions: zero mean, constant variance σ², uncorrelated. Add normality if the question involves tests or intervals.
4Derive what is asked from these. For example, E[Yᵢ] = α + βxᵢ and Var(Yᵢ) = σ².
5Interpret the slope in context: the average change in Y per one-unit increase in x, in the units given.
6If asked to check assumptions, say which plot or feature you would look at and what a violation looks like.
7Comment on how a violation would affect results, such as unreliable standard errors or intervals.

Quickest way: Three-line recall of the model

When to use it: Use this for short MCQs and for opening a written answer under time pressure.

Write Y = α + βx + ε with x fixed and ε random.
Write the errors as mean 0, variance σ², uncorrelated, and normal for inference.
Convert to Y: mean α + βx, variance σ², so Y is random only through ε.

Common mistakes in Simple Linear Regression Model and Assumptions

Saying the explanatory variable x is random with a variance.
Students mix regression with correlation, where both variables are random.
Fix: In this model x is fixed and known. Only ε, and therefore Y, is random.
Writing the assumptions about Y instead of ε, or forgetting that the mean of Y changes with x.
Students recall 'constant variance' but attach 'constant mean' as well.
Fix: The mean of Y is α + βxᵢ, which varies. Only the variance σ² is constant.
Claiming that linear means the plot must be a straight line in x only.
Students ignore that linearity refers to the parameters.
Fix: Say the model is linear in α and β. Transforming x, such as using x², can still fit this framework.
Stating that normality is needed for least squares estimates to exist.
The assumptions are memorised as one block.
Fix: Least squares needs only the mean and variance assumptions. Normality is needed for exact t-tests and confidence intervals.
Interpreting the intercept as meaningful when x = 0 is outside the data.
Students read α mechanically.
Fix: Say α is the fitted mean at x = 0, and note it may be an extrapolation with no real meaning.
Confusing the error εᵢ with the residual.
Both are called 'error' in everyday speech.
Fix: The error is the unobservable gap from the true line. The residual is the observed gap from the fitted line.

Worked examples

Example 1

An insurer models the claim amount Y (in ₹ thousand) on the sum insured x (in ₹ lakh) as Yᵢ = α + βxᵢ + εᵢ, with εᵢ independent N(0, σ²). Given α = 5, β = 2 and σ² = 9, find the mean and standard deviation of Y when x = 10, and interpret β.

Show the solution

The mean is E[Y] = α + βx = 5 + 2 × 10 = 25.
The variance is σ² = 9, the same at every x.
The standard deviation is √9 = 3.
Because the errors are normal, Y ~ N(25, 9) at x = 10.
The slope β = 2 means the average claim rises by ₹2 thousand for each extra ₹1 lakh of sum insured.

Answer: At x = 10, Y ~ N(25, 9): mean ₹25 thousand and standard deviation ₹3 thousand. β = 2 is the average rise of ₹2 thousand per ₹1 lakh.

Example 2

List the assumptions of the simple linear regression model with normal errors. Then state which assumption is most in doubt if the plot of residuals against fitted values fans out as the fitted value increases.

Show the solution

The model is Yᵢ = α + βxᵢ + εᵢ with xᵢ fixed and known.
The errors have mean zero: E[εᵢ] = 0.
The errors have constant variance: Var(εᵢ) = σ² for all i.
The errors are uncorrelated, and independent under normality: Cov(εᵢ, εⱼ) = 0 for i ≠ j.
For inference the errors are normal: εᵢ ~ N(0, σ²).
A fan shape means the spread of residuals grows with the fitted value.
That is non-constant variance, so the constant variance assumption is in doubt.

Answer: Assumptions: fixed x, errors with mean 0, constant variance σ², uncorrelated, and normal for inference. A fanning residual plot suggests the constant variance (homoscedasticity) assumption is violated.

Exam tips

Always write the model with subscripts and say that x is fixed and ε is random. Examiners give marks for this.
State each assumption separately. Do not bundle them into one phrase, as marks are awarded per assumption.
When asked to interpret, use the units and context in the question, not just 'a unit increase'.
In written answers on diagnostics, name the plot, describe what you see and say which assumption it challenges.
In the computer-based paper, the same assumptions guide your checks of the residual plots in R, so link each plot to an assumption.

Practice questions from Linear regression models

Simple Linear Regression Model and Assumptions in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Simple Linear Regression Model and Assumptions: frequently asked questions

What are the assumptions of simple linear regression in CS1?

The model is Yᵢ = α + βxᵢ + εᵢ with x fixed. The errors have mean zero, constant variance σ² and are uncorrelated. For exact tests and intervals you also assume they are normally distributed.

Which variable is the response and which is explanatory?

The response Y is the variable you want to explain or predict, and it is random. The explanatory variable x is the one used to explain it, and it is treated as fixed.

Is normality needed to fit the least squares line?

No. You can compute the least squares estimates with only the mean and variance assumptions. Normality is needed for exact t-tests, confidence intervals and prediction intervals.

What does 'linear' mean in linear regression?

It means the model is linear in the parameters α and β. The explanatory variable can be transformed, for example by taking its logarithm, and the model can still be linear in this sense.