Linear regression models the relationship between numerical variables using a straight-line equation. The core aim of Science mastery is not to make students press a calculator button and copy an equation. It is to help them understand what the slope, intercept, residuals and predictions mean scientifically.
For students and parents searching for linear regression, regression line, best fit line, slope and intercept, linear model or how to interpret linear regression, the most useful principle is this: linear regression turns a visual trend into a quantitative model.
The equation is useful only if the straight-line model actually fits the scientific relationship well enough.
The 60-Second Linear Model
A simple linear regression equation is:
y = a + bx
where:
- a = intercept;
- b = slope;
- x = predictor variable;
- y = predicted outcome.
The line is chosen to fit the data according to a defined criterion, commonly least squares.
Wait, What? Regression Is More Than Drawing a Line by Eye?
Yes.
A hand-drawn line of best fit is a useful school-level approximation.
Regression calculates a specific line using an objective rule.
In ordinary least squares, the line minimises the sum of squared vertical residuals.
See Line of Best Fit.
The Slope
The slope tells how much predicted Y changes for each one-unit increase in X.
Example:
y = 5 + 2x
Slope = 2.
For every 1-unit increase in X, predicted Y increases by 2 units.
The scientific interpretation depends on the variables and units.
The Intercept
The intercept is the predicted Y value when X = 0.
But it may not always have scientific meaning.
If X = 0 lies outside the observed range, the intercept may be an extrapolation rather than a real observed condition.
Always interpret it in context.
A Worked Example
Suppose a regression equation is:
temperature increase = 1.2 + 0.8 × power
If power rises by 1 unit, predicted temperature increase rises by 0.8 units.
If power = 5:
predicted temperature increase = 1.2 + 0.8 × 5 = 5.2.
The model provides an estimate, not a guaranteed exact outcome.
Residuals
A residual is:
observed value − predicted value.
Residuals show how far each data point lies from the regression line.
Small residuals suggest the line fits those observations closely.
Large residuals may indicate:
- outliers;
- poor model fit;
- nonlinearity;
- measurement variation.
Residual Plots
A residual plot helps check whether a linear model is appropriate.
A good linear fit often produces residuals scattered around zero without a strong pattern.
A curved residual pattern suggests the relationship may not be linear.
Regression should be diagnosed, not merely calculated.
Regression and Correlation
Correlation asks:
How strongly do X and Y move together linearly?
Regression asks:
What line best predicts Y from X under the model?
Regression Does Not Prove Causation
A regression slope can be highly significant even when:
- a confounder drives both variables;
- reverse causation exists;
- selection bias creates the relationship.
The equation models association.
Causation depends on design.
Interpolation
Using the regression line to predict within the observed X range is interpolation.
This is generally safer because the model is supported by nearby data.
See Interpolation and Extrapolation.
Extrapolation
Predicting beyond the observed range is extrapolation.
This is riskier because the relationship may change outside the measured region.
A line that fits from 20°C to 60°C may not continue indefinitely.
R-Squared
At more advanced levels, R² describes the proportion of variation in Y explained by the linear model with X in a simple regression context.
R² close to 1 means the line accounts for a large proportion of observed variation.
But high R² does not prove:
- causation;
- correct model structure;
- absence of bias.
Regression and Outliers
Outliers can strongly influence:
- slope;
- intercept;
- correlation;
- R².
Always inspect the scatter plot before trusting the equation.
Regression and Units
The slope carries units.
If Y is measured in °C and X in watts, slope units are:
°C per watt.
This makes the coefficient scientifically interpretable.
Primary Science Foundations
Primary learners can begin with:
- does Y generally rise as X rises?
- does the pattern look roughly straight?
- can we estimate a value between measured points?
Secondary Science Linear Regression
Secondary students should increasingly understand:
- slope;
- intercept;
- residuals;
- R²;
- interpolation;
- extrapolation;
- causal limits.
How to Practise
Given a scatter plot and regression equation:
- interpret the slope;
- interpret the intercept carefully;
- predict one value inside the data range;
- identify one risky extrapolation;
- inspect whether residuals support a linear model.
Common Linear-Regression Mistakes
- assuming regression proves causation;
- interpreting an impossible intercept literally;
- extrapolating far beyond the data;
- ignoring curved residual patterns;
- trusting R² without examining the graph;
- forgetting units on the slope.
Frequently Asked Questions
What is linear regression?
Linear regression fits a straight-line model describing how one numerical variable changes with another.
What does the slope mean?
It is the predicted change in Y for a one-unit increase in X.
What is a residual?
The difference between an observed value and the value predicted by the regression line.
Does a high R² prove causation?
No. It only describes how much variation the fitted model accounts for under the data and model.
Why is extrapolation risky?
Because the relationship observed inside the data range may not continue outside it.
Useful eduKateSG Routes
The Core Aim
Linear regression turns a trend into an equation.
Interpret the slope. Question the intercept. Inspect residuals. Predict carefully. Never confuse a fitted line with proof of cause.
That is the core aim: model scientific relationships quantitatively while respecting the limits of the model.
Properly taught kids shine a bright light into the future.