Confidence Intervals for Slopes of Regression Lines

Population vs. Sample Slope: In simple linear regression, the population model is represented as $y = \beta_0 + \beta_1 x + \epsilon$, where $\beta_1$ is the true population slope. Since we rarely observe the entire population, we use the sample slope $b_1$ (or $\hat{\beta}_1$) as a point estimate for $\beta_1$.

The Purpose of the Interval: A confidence interval for the slope provides a range of values that is likely to contain the true population slope $\beta_1$ at a specified confidence level (e.g., 95%). It answers the question: 'Based on our sample, what is the range of possible values for the average change in $y$ for every one-unit increase in $x$?'

Standard Error of the Slope ($SE_{b_1}$): This value measures the estimated standard deviation of the sampling distribution of $b_1$. It quantifies how much the sample slope is expected to vary from one random sample to another.

Sampling Distribution of $b_1$: If the regression assumptions are met, the sampling distribution of the sample slope $b_1$ is approximately normal. It is centered at the true population slope $\beta_1$ with a standard deviation that decreases as the sample size increases or as the spread of the $x$-values increases.

The t-Distribution: Because we must estimate the population standard deviation of the residuals using the sample standard deviation ($s$), we use the $t$-distribution rather than the standard normal ($z$) distribution. This accounts for the extra variability introduced by using an estimate for the standard error.

Degrees of Freedom: For slope inference, the degrees of freedom are calculated as $df = n - 2$. This is because we have estimated two parameters from the data: the $y$-intercept ($\beta_0$) and the slope ($\beta_1$).

The Confidence Interval Formula: The interval is constructed using the point estimate plus or minus the margin of error: $b_1 \pm t^* \cdot SE_{b_1}$ where $b_1$ is the sample slope, $t^*$ is the critical value for the desired confidence level with $df = n - 2$, and $SE_{b_1}$ is the standard error of the slope.

Calculating the Standard Error: The standard error of the slope is calculated as: $SE_{b_1} = \frac{s}{s_x \sqrt{n-1}}$ where $s$ is the standard deviation of the residuals and $s_x$ is the standard deviation of the $x$-values. This formula shows that the slope estimate becomes more precise as the sample size increases or as the $x$-values become more spread out.

Step-by-Step Construction: First, verify the LINE conditions. Second, calculate the sample slope and its standard error (often provided by software). Third, find the critical value $t^*$ using a $t$-table or calculator. Finally, compute the interval and interpret it in the context of the variables.

Slope vs. Correlation: While both measure the relationship between variables, the slope CI provides the specific rate of change in the units of the variables, whereas correlation is a unitless measure of strength and direction.

Standard Deviation of Residuals ($s$) vs. Standard Error of Slope ($SE_{b_1}$): $s$ measures the typical distance that the observed $y$-values fall from the regression line, while $SE_{b_1}$ measures the typical distance that the sample slope $b_1$ falls from the population slope $\beta_1$.

Always Check Degrees of Freedom: A common mistake is using $n-1$ for the $t^*$ critical value. Remember that for regression, you lose two degrees of freedom ($n-2$) because you are estimating both the intercept and the slope.

Interpret the Slope Correctly: Ensure your interpretation mentions 'for every 1 unit increase in $x$' and 'the predicted/average change in $y$'. Avoid saying that $y$ will change by that amount for every individual case.

Zero in the Interval: If the confidence interval for the slope contains zero (e.g., -0.2 to 0.5), it implies that there is no statistically significant linear relationship between the variables at that confidence level, as a slope of zero is a plausible value.

Units Matter: Always include the units of the response variable per unit of the explanatory variable in your final conclusion to demonstrate a full understanding of the context.