Skip to content
  • Open the math solver
  • Ask the AI tutor
  • Find an AP course
  • Create a quiz
  • Open the calculator
  • Graphing Calculator
  • Browse all calculators
  • Read math guides
↑ ↓Enter to open · Esc to close
Solve a problem
UnofficialAP StatisticsUnit 9 of 9

AP Statistics Unit 9: Inference for Quantitative Data: Slopes

Intervals and tests for the slope of a regression line.

Unit overview

The slope bb of a sample regression line estimates the true slope β\beta of the population relationship. Because bb varies from sample to sample, you can build a confidence interval b±t∗SEbb \pm t^* SE_b or test H0:β=0H_0: \beta = 0 with t=b/SEbt = b / SE_b and n−2n - 2 degrees of freedom.

Conditions are summarised as LINER: the true relationship is linear, observations are independent, residuals are approximately normal, the spread of residuals is roughly equal for all xx, and the data come from a random sample or randomised experiment. Almost every question gives computer output, so reading the slope and its standard error from a table is part of the skill.

Key topics

Sampling distribution of bb
Centred at β\beta with standard error SEbSE_b from computer output.
Confidence interval for slope
b±t∗SEbb \pm t^* SE_b with df=n−2df = n - 2.
Test for slope
t=(b−β0)/SEbt = (b - \beta_0)/SE_b, usually with β0=0\beta_0 = 0.
Checking conditions
Residual plot for linearity and equal spread, a histogram or dotplot of residuals for normality.

Worked example

Computer output gives slope b=1.8b = 1.8 with SEb=0.6SE_b = 0.6. Find the test statistic for H0:β=0H_0: \beta = 0.

  1. t=b−0SEbt = \dfrac{b - 0}{SE_b}.
  2. t=1.80.6=3t = \dfrac{1.8}{0.6} = 3.

Answer: t=3t = 3

Common mistakes

  • Using n−1n - 1 degrees of freedom instead of n−2n - 2.
  • Reading the standard error of the intercept instead of the slope from output.
  • Interpreting the interval as describing individual points rather than the true slope.

Exam tips

  • Interpret the slope interval as a range of plausible values for the change in the mean response per unit increase in xx.
  • If the interval contains 0, there is no convincing evidence of a linear relationship at that confidence level.

See the AP Statistics exam format and every unit on the course page.