Section 38 Simple Linear Regression Model: Issues


38.0.1 Regression Model

\[ \Huge y_{i} = \beta_0 + \beta_1 x_{i} + \epsilon_{i}, \space i=1,...,n \]


38.1 Response & Predictor

  • How to identify response and predictor variable(s)?

  • Example: Preparing a Standard Curve in laboratory works, e.g. RT-PCR study.

  • Example: Chemical composition and spectral data


38.2 Regressing X on Y & Regressing Y on X

  • Regressing X on Y is not the same as Y on X

  • Regression coefficient X on Y

  • Regression coefficient Y on X


38.3 Regression & Correlation

  • Correlation quantifies the degree to which two variables are related.

  • Correlation does not fit a line through the data points. It provides information regarding how much one variable tends to change when the other one does.

  • Correlation assumes linear relationhsip.

  • The estimate of correlation ranges between -1 to +1

  • Correlation is not causation.

  • In regression, we predict one variable (reponse) based on the other variable (predictor).

  • The strength of the relationship is explicit in the regression analysis, and uncertainty can be seen clearly from confidence intervals or prediction intervals.

  • Regression can capture polynomial or non-linear relationship.

  • However, interpretation of regression coefficient as the causal relationship should be avoided.


38.4 Zero intercept

  • The meaning of zero intercept.

  • The fitting of zero intercept should be considered cautiously.

  • The \(R^2\) for models with and without intercept cannot be compared.


38.5 Interpreting intercept

  • Check if the interpretation of intercept is meaningful

  • Centering of data


38.6 Interpretation of regression coefficient

  • Standard interpretation

  • Effects and association

  • Regression coefficient does not answer causality


38.7 Centering and scaling X variable

  • Estimates of \(\beta_0\) and \(\beta_1\)


38.8 Centering and scaling Y variable

  • Estimates of \(\beta_0\) and \(\beta_1\)


38.9 Prediction of Y

  • Confidence Interval & Prediction Interval

  • Avoid extrapolation


38.10 Regression toward the mean

  • Regression towards the mean is a mathematical phenomenon; it is not a causal phenomenon.

  • If a variable is extreme on its first measurement, it will tend to be closer to the average on its second measurement.

  • Consider carefully the regression toward the mean while planning an experimental design and interpreting the data.