Claims Based on Slope Intervals

When we collect data and fit a least-squares regression line, the slope of that line is not a fixed truth — it is an estimate calculated from one particular sample. Because samples vary, so does the slope. Regression inference lets us move beyond a single number and construct a confidence interval for the true population slope β₁. Once we have that interval, we can make meaningful, context-grounded claims about the real relationship between two variables. A confidence interval for the slope takes the form b₁ ± t · SE(b₁), where b₁ is the sample slope, t is the critical value from a t-distribution with n − 2 degrees of freedom, and SE(b₁) is the standard error of the slope. The resulting interval gives a range of plausible values for the true slope. Interpreting that range in context — rather than just reporting numbers — is the central skill this concept addresses. The practical power of a slope interval comes from what it tells us about direction, size, and significance. If the entire interval lies above zero, we have evidence of a positive association. If it includes zero, we cannot rule out no linear relationship. If both endpoints have meaningful magnitude, we can make quantitative claims about how much the response variable typically changes per unit increase in the explanatory variable. Learning to translate these mathematical results into clear, careful statements is essential for any data-based argument.

Key takeaways
  • Identify the interval and its confidence level
  • Translate units into context
  • Assess direction of the relationship
  • Assess practical significance
  • Make a claim with appropriate hedging

Identify the interval and its confidence level

State the confidence level (typically 90%, 95%, or 99%) and both endpoints of the interval. For example: 'We are 95% confident that the true slope β₁ lies between 2.1 and 4.7.' The confidence level tells you how reliable the procedure is: if we repeated the sampling process many times, about 95% of the intervals constructed would capture the true β₁.

Translate units into context

A slope always carries units: (units of y) per (unit of x). After building an interval, restate both endpoints using the actual variable names and units. Example: 'For each additional hour of weekly exercise, we estimate the true average change in resting heart rate is between −4.7 and −2.1 beats per minute.' This makes the mathematics meaningful and verifiable against common sense.

Assess direction of the relationship

Check whether the interval lies entirely above zero, entirely below zero, or straddles zero. Entirely positive interval → evidence of a positive linear association. Entirely negative interval → evidence of a negative linear association. Interval containing zero → insufficient evidence of any linear relationship; zero slope (no association) is plausible. This is logically equivalent to a two-sided hypothesis test of H₀: β₁ = 0 at the corresponding significance level.

Assess practical significance

Statistical significance (the interval excludes zero) does not automatically mean practical significance. Examine the magnitude of both endpoints. If the interval is (0.001, 0.003) for a variable measured in dollars, the relationship may be statistically real but economically trivial. Conversely, a very wide interval signals high uncertainty even if the midpoint is large. Always ask: 'Are both endpoints large enough to matter in the real world?'

Make a claim with appropriate hedging

A well-formed claim based on a slope interval includes: (a) the confidence level, (b) the direction of the relationship, (c) the numerical range with units, and (d) an acknowledgement that this is an estimate from sample data. Avoid saying 'the slope IS between a and b'; instead say 'we are 95% confident the slope is between a and b.' Avoid implying causation unless the study design supports it — a confidence interval describes association, not cause-and-effect, unless the data come from a randomized experiment.

The Key Equation

b1±t∗⋅SE(b1),SE(b1)=ssxn−1b_1 \pm t^* \cdot SE(b_1), \quad SE(b_1) = \frac{s}{s_x\sqrt{n-1}}
  • b_1 — Sample slope (units of y / units of x): The estimated slope from the least-squares regression line fitted to the sample data.
  • t^* — Critical value (dimensionless): The value from a t-distribution with n − 2 degrees of freedom corresponding to the desired confidence level.
  • SE(b_1) — Standard error of slope (units of y / units of x): Measures how much b₁ is expected to vary from sample to sample.
  • s — Residual standard deviation (units of y): Estimates the standard deviation of the errors around the regression line; s = √[Σ(yᵢ − ŷᵢ)² / (n−2)].
  • s_x — Sample standard deviation of x (units of x): Measures the spread of the explanatory variable values in the sample.
  • n — Sample size (count): Number of data points; degrees of freedom for the t-distribution are n − 2.

Worked Example

Problem. A researcher studies the relationship between daily screen time (hours) and sleep duration (hours) for a random sample of 22 teenagers. The least-squares regression gives a slope of b₁ = −0.38 hours of sleep per hour of screen time, with SE(b₁) = 0.14. Construct and interpret a 95% confidence interval for the true slope.

(−0.672,  −0.088) hours of sleep per hour of screen time(-0.672,\; -0.088) \text{ hours of sleep per hour of screen time}

Answer. We are 95% confident that the true slope is between −0.672 and −0.088 hours of sleep per hour of screen time. Because the interval lies entirely below zero, there is convincing evidence that increased screen time is associated with decreased sleep duration among teenagers.

Figure
Diagram
Slope Confidence Intervals: Three Scenariosβ₁ = 0 (no association)Scenario A: Entirely Positive (Positive Association)0.83.2b₁=2.0Excludes 0 ✓Scenario B: Contains Zero (No Convincing Evidence)-0.52.1b₁=0.8Includes 0 ✗Scenario C: Entirely Negative (Negative Association)-4.1-1.3b₁=-2.7Excludes 0 ✓← Negative slopes | Zero | Positive slopes →
Origins · 1 minFrancis Galton and how this was discovered
FG
Francis Galton
1822–1911 · British

Galton coined the term 'regression' in the 1880s while studying the relationship between the heights of parents and their children. He observed that children of tall parents tended to be shorter than their parents, and children of short parents tended to be taller — a 'regression toward mediocrity' (what we now call regression to the mean). His work laid the conceptual groundwork for the slope and the regression line, ideas later formalized mathematically by Karl Pearson. Galton was so obsessed with measurement that he once tried to create a 'beauty map' of Britain by secretly recording the attractiveness of people he passed on the street — a reminder that being a mathematical pioneer and being ethically sound are very different things.

  • Least-Squares Regression Line — The slope b₁ whose confidence interval we build comes directly from the least-squares regression line; you must understand what b₁ measures before interpreting an interval for it.
  • Hypothesis Test for Slope — A two-sided hypothesis test H₀: β₁ = 0 and a 95% confidence interval for β₁ are mathematically equivalent; if zero falls outside the interval, the test rejects H₀ at the 5% significance level.
  • Confidence Intervals for Means — The logic and structure of a slope confidence interval mirrors that of a confidence interval for a mean: estimate ± t* × standard error. Mastering the mean interval makes the slope interval much easier to understand.
  • Conditions for Regression Inference — A slope confidence interval is only valid when four conditions hold (linearity, independence, normal residuals, equal variance). Checking these conditions is a required part of any inference procedure.
  • Residuals and Residual Plots — The standard error SE(b₁) depends on the residual standard deviation s, so understanding residuals and how to plot them is essential for interpreting where the width of a slope interval comes from.