Key points at a glance
- Plot residuals against fitted values.
- Check competing explanations.
- Choose defensible uncertainty estimates.
Locate the assumption correctly
In classical linear regression, constant variance concerns error spread at different predicted values, not identical spread for every raw variable. A fan shape in residuals versus fitted values may indicate unequal error variance.
Inspect the pattern and data
Example: Residuals are tight at low predicted study time and spread widely at high predictions. First check outliers, typing errors and whether a nonlinear relation explains the pattern. One extreme point is not itself a general fan shape.
Discuss consequences and remedies
Unequal variance can affect conventional standard errors and tests. Robust standard errors, a justified transformation or another model may be suitable, depending on the question. Choose with methodological reasoning, not solely because software flags a test.
Report diagnosis and adjustment
Name the model, residual plot, affected range and uncertainty method used. Where useful, compare conclusions with and without adjustment. Keep unequal variance distinct from normality and independence assumptions.
Final submission context
Formal details can feel like a separate writing task, but they matter just as much for a printed submission. Anything missing, misplaced or inconsistently formatted in the document will also appear in the bound copy.
Use this guide together with your cover page, table of contents, page numbers, source notes and appendices. Prepare the final PDF for printing and binding only after the complete file has been checked.
Practical check before PDF export
- Do headings, chapter structure and page numbers match?
- Are sources, figures and tables included completely?
- Are required elements such as declarations, appendices or the cover page included where required?
- Did you open and check the final PDF after exporting it?
Checklist
- Assumption located.
- Residual plot inspected.
- Alternative pattern checked.
- Adjustment recorded.
Common mistakes
- Confusing raw spread with error variance.
- Calling every outlier heteroscedasticity.
- Treating robust errors as causal proof.
Prepare the finished thesis for submission
Once the content and university requirements are checked, review your PDF and the available printing and binding options.
Review printing optionsRelated content
Explore related guidance on thesis structure, sources, formatting and final submission.
Frequently asked questions
Must my outcome have equal spread everywhere?
The relevant question is conditional error spread in the particular model.
Is a formal test enough?
No. Inspect plots and model context too.
Is the regression worthless?
Not automatically; evaluate consequences and robust or alternative methods.