Wize Business Analytics Textbook > Correlation and Regression

Inferences for Slope and Intercept, and Mean Response

Inference for the Slope, the Intercept, and the Mean Response

Regression inference targets parameters and mean responses

In simple linear regression, the fitted intercept and slope estimate population parameters. Their standard errors quantify sampling variation under the regression conditions: an approximately linear mean relationship, independent observations, roughly constant residual spread, and residuals suitable for the chosen small-sample inference.


A confidence interval describes plausible values of the population parameter, not a range containing a stated fraction of individual outcomes.

Slope and intercept use different standard errors

The slope standard error is small when residual variation is small and the observed explanatory values are well spread out. The intercept standard error also depends on how far zero lies from the observed explanatory values.


An intercept can therefore be imprecise even when the slope is estimated well. Interpret an intercept cautiously when zero is outside the observed range or lacks a meaningful real-world interpretation.


For a slope hypothesis test, subtract the null slope from the estimated slope and divide by its reported standard error. The reference distribution has residual degrees of freedom, not the number of observations.


A confidence interval and the corresponding two-sided test agree: exclusion of the null value at a chosen confidence level corresponds to rejection at the associated significance level, provided both use the same assumptions and convention.

Mean response is not an individual prediction

At a selected explanatory value, the fitted line estimates the mean response of the population of units with that value.


Its standard error is smallest at the sample mean of the explanatory variable and grows as the target point moves away from that center.


A confidence interval for the mean response reflects uncertainty in the fitted mean. A prediction interval for an individual response is wider because it must also include the unexplained residual variation of a new individual.

Connection between the intercept and the line

The intercept is the fitted mean response when the explanatory value is zero. Consequently, the confidence interval for the mean response at zero is algebraically identical to the confidence interval for the intercept.


This identity is a useful formula check, but it does not make extrapolation to zero scientifically defensible if zero was not represented in the study.


Inference about a slope is association under the stated model and sampling conditions. It does not on its own establish that changing the explanatory variable would change the response. Causal interpretation needs a design that addresses confounding, such as random assignment or a justified identification strategy.

A regression of weekly grocery spending (dollars) on household size (people) for a random sample of households gives a 95% confidence interval for the slope of (18.4, 31.0). Which interpretation is correct?

Choose an answer

Create a free account to get started

Unlock full course notes by signing up

  • Video lessons taught by top educators
  • Practice questions with guided solutions
  • Written review notes for reference
  • Expert answers anytime from course tutors

Regression output from n = 20 observations reports a slope estimate b1 = 2.40 with standard error SE(b1) = 0.80. For testing H0: β1 = 0, what are the test statistic and its degrees of freedom?

Choose an answer

Create a free account to get started

Unlock full course notes by signing up

  • Video lessons taught by top educators
  • Practice questions with guided solutions
  • Written review notes for reference
  • Expert answers anytime from course tutors

Mark Yourself Question

  1. Grab a piece of paper and try this problem yourself.
  2. When you're done, check the "I have answered this question" box below.
  3. View the solution and report whether you got it right or wrong.

For 14 used cars of the same model, a regression of resale price (thousands of dollars) on age (years) gives b1 = −1.25 with SE(b1) = 0.40. Assuming the regression conditions hold, construct a 95% confidence interval for the population slope β1, giving endpoints to two decimal places, and state in one sentence what it says about price and age. The 97.5th percentiles of the t distribution are 2.179 for 12 df and 2.160 for 13 df.

Create a free account to get started

Unlock full course notes by signing up

  • Video lessons taught by top educators
  • Practice questions with guided solutions
  • Written review notes for reference
  • Expert answers anytime from course tutors