1. Read the model in context
Suppose a linear model predicts a practice score, y, from the number of practice sessions, x. The fitted model is ŷ = 2x + 5, and the observed data cover 1 to 6 sessions. This is an invented teaching example, not evidence about real students.
Gradient: the model predicts an increase of 2 score points for each additional practice session.
Intercept: the model predicts 5 points at zero sessions. Because zero is outside the observed range, that interpretation should be treated cautiously.
The hat in ŷ indicates a predicted value. A prediction is not a guarantee that every observation will lie on the line.
2. Calculate and interpret a residual
A student completes 4 sessions and scores 16 points. The model predicts:
Residual = observed − predicted
= 16 − 13 = 3 points
The positive residual means the observed score is 3 points above the prediction. A negative residual would mean the observation is below it. Always keep the subtraction order consistent.
3. Know where the evidence ends
Predicting the score for 4 sessions is interpolation because 4 lies within the observed range of 1 to 6. Predicting for 12 sessions is extrapolation.
The arithmetic is correct. The prediction may still be unreliable because the relationship has not been observed in that range. If the practice test has a maximum of 20 points, 29 is also impossible in this context.
Keep mathematical correctness and model suitability separate. A formula can be evaluated correctly while producing an unrealistic prediction.
4. Avoid turning association into causation
A relationship between practice sessions and scores does not, by itself, establish that the sessions caused the score difference. Prior knowledge, motivation and the difficulty of the tasks could also matter.
A stronger conclusion identifies what the data show, limits the claim to the evidence and acknowledges relevant assumptions. For example: “Within the observed range, the model predicts higher scores with more sessions; this relationship alone does not establish causation.”
Your turn: interpret the error
Using ŷ = 2x + 5, an observation at x = 3 has an actual score of 9. Find the prediction and residual, then explain the sign.
Reveal the worked solution
Residual = 9 − 11 = −2 points
The observed score is 2 points below the model prediction. The model overpredicts this observation by 2 points.
Before finishing a modelling answer, check: Did I name the variables, include units, explain the result and recognise the model’s limitations?
These are original teaching examples. Your school’s syllabus and task-specific instructions take priority. Ask about personalised support.