AP Statistics Unit 9: Inference for Slope

Least Squares Regression Line (LSRL)

A method to describe the relationship between two quantitative variables using the equation $y = b0 + b1x$.

Population Parameter

The true value in a population, represented by symbols like $eta$ for slope.

Sample Statistic

An estimate derived from a sample, represented by symbols like $b$ for slope.

True Regression Model

The mathematical model representing the true relationship in the population: $y = \alpha + \beta x + \epsilon$.

Sampling Distribution of $b$

The distribution of sample slopes calculated from repeated samples, centered at the true slope $eta$.

LINER Conditions

Conditions necessary for valid inference: Linear, Independent, Normal, Equal Variance, Random.

Linear Condition

The linear relationship between $x$ and $y$ should be verified using scatter plots and residual plots.

Independent Condition

Each observation must be independent; for sampling without replacement, the population should be at least $10 imes n$.

Normal Condition

Responses ($y$) should vary normally around the true regression line, checked by histogram or normal probability plot of residuals.

Equal Variance (Homoscedasticity) Condition

Standard deviation of $y$ should remain consistent across all values of $x$, indicated by a residual plot.

Confidence Interval Formula

The formula for estimating a true slope $eta$ is $b \pm t^* SE_b$.

Standard Error of the slope ($SE_b$)

A measure of the variability of the sample slope, often provided in computer output.

Degrees of Freedom ($df$)

Calculated as $n - 2$, reflecting the loss from estimating two parameters for fitting a regression line.

Null Hypothesis ($H_0$)

States that the true slope $eta$ equals 0, indicating no linear relationship between $x$ and $y$.

Alternative Hypothesis ($H_a$)

Suggests that the true slope $eta$ is not equal to 0, indicating a linear relationship.

Test Statistic Formula

Formula for calculating the t-value: $t = \frac{b - \beta0}{SEb}$.

T-Value

The calculated t-statistic for testing the hypothesis about the slope, found in regression output.

P-Value

The probability of obtaining a sample slope as extreme as observed if the null hypothesis is true.

Rejecting Null Hypothesis

If $P < 0.05$, indicating sufficient evidence to conclude a linear relationship exists.

Extrapolation

Predicting values beyond the range of observed data; inference does not permit this.

Intercept Trap

The mistake of confusing the intercept value with the slope; always reference the correct variable row in output.

Interpreting Confidence Intervals

The correct interpretation is about the true population slope falling within a range, not stating the slope itself as a fixed value.

Significance Testing

A method to determine if evidence exists for a linear relationship by testing hypotheses about the slope.

Context in Conclusions

Always add context when concluding about slopes or confidence intervals, explaining the relationship in practical terms.

Residual Plot

A scatterplot that checks for the randomness and consistency of residuals across the range of values of $x$.