Class 12 Statistics Notes · GSEB

Properties of Regression Coefficient

Regression — learn the 6 key properties of regression coefficient including sign matching, product relationship, and independence from origin and scale. GSEB Class 12 Statistics notes.

Last updated: 22 Sep 2026

Notes

Properties of Regression Coefficient

Linear Regression — Chapter 5, GSEB Class 12 Statistics

Properties

1

Sign of b Matches Sign of r

If r > 0, both byx and bxy > 0. If r < 0, both byx and bxy < 0.

The regression slope always moves in the same direction as the correlation. Positive r → upward-sloping line; negative r → downward-sloping line.

2

byx × bxy = r²

The product of the two regression coefficients equals the square of the correlation coefficient.

This is always ≤ 1 since r² ≤ 1. It links regression back to correlation.

3

Not Independent of Change of Origin

Adding or subtracting a constant from X or Y changes the value of b.

Unlike r, the regression coefficient IS affected by shifting the origin. If you add 10 to all X values, b changes.

4

Not Independent of Change of Scale

Multiplying or dividing X or Y by a constant changes the value of b.

If you convert X from metres to centimetres (×100), b changes by a factor of 100. The slope depends on the units used.

5

r = √(byx × bxy)

The correlation coefficient is the geometric mean of the two regression coefficients.

This follows from byx × bxy = r². Taking square root gives r = √(byx × bxy).

6

b Can Exceed 1 in Absolute Value

Unlike r (bounded −1 to +1), b has no fixed range. |b| can be greater than 1.

b depends on the units of measurement. If Y changes by 100 when X changes by 1, b = 100. This is perfectly valid.

Comparison with r

Propertyr (Correlation)b (Regression)
Range−1 to +1 (bounded)No fixed range
Independent of origin✓ Yes✗ No
Independent of scale✓ Yes✗ No
Sign± (direction)Same as r
UnitsUnitlessHas units (Y units / X units)

Key Takeaways

Key Takeaways

  • Sign of b always matches sign of r — both positive or both negative.
  • byx × bxy = r² — product of regression coefficients equals squared correlation.
  • b is NOT independent of change of origin — shifting values changes b.
  • b is NOT independent of change of scale — changing units changes b.
  • r = √(byx × bxy) — correlation is the geometric mean of the two coefficients.
  • Unlike r (bounded −1 to +1), b can exceed 1 in absolute value.