Statistics
Guess the Regression Line
Can you fit the paper-bridge data by eye?
Adjust the slope and intercept until your line seems to fit the points. Turn on residuals to see where it misses.
Paper bridges under load
Each trial uses a stack of paper spanning the same gap. The point records the number of layers and how many washers the bridge held. These are fictional classroom trials.
Solid green line: your guess. Square point: movable trial.
Change one trial
Move the square to 11 layers and 6 washers. Predict which way the least-squares line will turn. Compare with a trial at 4 layers and 6 washers.
You can also drag the square. The other six observations stay fixed.
Trial values and residuals
| Layers | Washers | Predicted | Residual |
|---|
Teaching notes and definitions
A predicted load is intercept + slope × layers. A residual is observed load minus predicted load, measured vertically. Least squares minimizes the sum of squared vertical residuals (SSE), measured here in washers squared.
A point far from the average predictor value has high leverage. To assess influence, compare the fitted line with and without the movable trial. An extreme layer count can have high leverage even when the point lies close to the pattern and barely changes the fit.