Module 08 Activity

Scenario

Fit and interpret a simple relationship without overclaiming what it shows.

What you build

A regression with its assumptions checked, its interval reported and its purpose stated.

Steps

1. Plot the relationship before quoting any correlation, and check whether one point is driving it.
2. Choose Pearson or Spearman deliberately, and treat a large gap between them as evidence of curvature.
3. Fit the model, report the slope with its confidence interval, and state the observed range of the predictor.
4. Inspect the residuals for pattern. Report the residual standard deviation as well as R-squared.
5. Decide whether a question needs a confidence interval for the mean response or a prediction interval for a new observation, and say which your figure shows.
6. State whether you are predicting or explaining, and list the confounders you could not measure.

Evidence to hand in

• The scatter plot and any influential point.
• The correlation, with the method chosen and its interval.
• The fitted model with the slope interval and the observed range.
• The residual inspection and the residual SD.
• The interval you chose, with its justification.
• The prediction-or-explanation statement with unmeasured confounders.

Review checklist

• Every relationship was plotted before its correlation was quoted.
• The slope is reported with an interval and the observed range of x.
• Residuals were inspected for pattern, not just summarised.
• The figure states whether it shows a confidence or a prediction interval.
• Causal language is used only where the predictor was randomised.

Mark complete