Confidence Interval and Hypothesis Testing for Beta 1
Inference for the Simple-Regression Slope
Inference targets the population slope
In simple linear regression, the estimated slope b1 summarizes the sample's fitted change in Y per unit change in X. Inference concerns the population slope beta1 in the specified model. A standard two-sided test uses H0: beta1 = 0 against an alternative that beta1 differs from zero.
One calculation, two related summaries
The t statistic is (b1 - hypothesized value)/SE(b1), with simple-regression degrees of freedom based on the sample size. A confidence interval has the form b1 plus or minus t-star times SE(b1).
For example, a slope estimate of 1.18 with standard error 0.32 and a critical value of 2.03 has margin 0.6496 and interval approximately [0.53, 1.83]. At matching two-sided levels, an interval excluding zero agrees with rejecting the zero-slope null.
Interpret the interval at its scope
The slope and its interval carry response-units per predictor-unit. They are not prediction intervals for individual outcomes, and a confidence interval does not establish causation. In observational data, the modeled association may be affected by confounding, selection, measurement, or functional-form assumptions. State those boundaries rather than turning a nonzero slope into an unsupported causal claim.
A simple regression gives b1 = 0.72 with SE(b1) = 0.24. For H0: beta1 = 0, what is the t statistic?
A simple regression estimates b1 = 1.40 with SE(b1) = 0.50. A 95% t critical value for the relevant degrees of freedom is 2.10. Which interval is the 95% confidence interval for beta1?