Unit 11 · Advanced regression

How to check regression assumptions in SPSS

Check linearity, homoscedasticity, normality of residuals, independence (Durbin-Watson) and influential cases (Cook’s distance) in SPSS.

Free tutorialPro practiceAbout 5 minutes6 practice questions

Quick answer: How do you check regression assumptions in SPSS?

  1. In Analyze › Regression › Linear, click Plots… and put *ZRESID on Y and *ZPRED on X.
  2. Tick Histogram and Normal probability plot.
  3. Under Statistics… tick Durbin-Watson and Collinearity diagnostics; under Save… tick Cook’s distance.
  4. Check for a shapeless residual cloud, Durbin-Watson near 2 and Cook’s distance below 1.

Last updated November 2026 · Not sure this is the right test? Try the Which statistical test? tool.

Regression assumptions

Regression results are only trustworthy if certain assumptions hold. Most are checked through the residuals: the differences between each person’s actual and predicted score.

When to use it

  • After every linear or multiple regression, before interpreting the results.

How it works

Linearity and homoscedasticity: plot standardised residuals against standardised predicted values; you want a shapeless cloud of even width. A funnel means unequal spread; a curve means a non-linear relationship. Normality of residuals: histogram and P-P plot. Independence: Durbin-Watson around 2 (1.5–2.5 is generally fine). Outliers: standardised residuals beyond ±3. Influential cases: Cook’s distance above 1. Multicollinearity: VIF above 10 (or 5).

Worked example

Residuals were evenly spread and roughly normal, Durbin-Watson = 1.94, no standardised residuals beyond ±3, maximum Cook’s distance = .09 and all VIFs below 1.1, so assumptions were met.

Key terms

Homoscedasticity
Equal spread of residuals at all predicted values.
Durbin-Watson
Tests whether residuals are independent (autocorrelation).
Cook’s distance
How much a case changes the regression results.
Leverage
How unusual a case’s predictor values are.

Common mistakes

  • Checking normality of the raw variables instead of the residuals.
  • Deleting influential cases without investigating or reporting them.

How to do it in SPSS

Requesting assumption checks

Most regression assumptions are checked through the residuals: the gaps between the actual and predicted scores.

Linear Regression: Plots
Variables
DEPENDNT
*ZPRED
*ZRESID
*DRESID
*ADJPRED
*SRESID
▶
Y:1
*ZRESID
▶
X:2
*ZPRED
Standardized Residual Plots3
✓Histogram
✓Normal probability plot
ContinueCancelHelp
  1. 1
    Y: *ZRESIDStandardized residuals go on the vertical axis.
  2. 2
    X: *ZPREDStandardized predicted values go on the horizontal axis. Together they check linearity and equal spread (homoscedasticity).
  3. 3
    Residual plotsThe histogram and P-P plot check that residuals are roughly normal.

While you’re in the dialog: Statistics… › Durbin-Watson and Collinearity diagnostics, and Save… › Cook’s distance.

Reading the residuals

Four quick checks cover most of what markers look for.

Output1 [Document1] – SPSS Statistics Viewer
FileEditViewDataTransformAnalyzeGraphsUtilitiesExtensionsWindowHelp
Output
Log
Regression
Model Summary
Residuals Statistics
Scatterplot
Model Summary
ModelRR
Square
Adjusted
R Square
Std. Error of
the Estimate
Durbin-
Watson
1.565.319.3029.641.941
Residuals Statistics
MinimumMaximumMeanStd. DeviationN
Predicted Value48.2184.6566.646.85120
Std. Residual−2.612.872.000.987120
Cook's Distance.000.0873.009.014120
Centered Leverage Value.002.081.025.016120
Scatterplot · Dependent Variable: score4
Regression Standardized Predicted ValueZRESID
  1. 1
    Durbin-WatsonChecks that residuals are independent. Values between about 1.5 and 2.5 are fine.
  2. 2
    Std. ResidualValues beyond ±3 suggest outliers. Here the range is −2.61 to 2.87, so none.
  3. 3
    Cook's DistanceShows whether any single person pulls the line. Maximum .087 is far below the usual cut-off of 1.
  4. 4
    Residual plotA shapeless cloud around zero is what you want. A funnel shape means unequal spread; a curve means the relationship isn’t linear.

The SPSS syntax

Prefer typing to clicking? Open File › New › Syntax, paste this in, select it and choose Run › Selection. Every dialog's Paste button writes the same kind of code.

REGRESSION
  /STATISTICS COEFF OUTS R ANOVA COLLIN TOL
  /DEPENDENT score
  /METHOD=ENTER hours attendance anxiety
  /SCATTERPLOT=(*ZRESID ,*ZPRED)
  /RESIDUALS DURBIN HISTOGRAM(ZRESID) NORMPROB(ZRESID)
  /SAVE COOK.

Practice data

Download the file used in this lesson and run it yourself in SPSS. Your results will closely match the screens above, though a few numbers may differ slightly.

  • exam_multi.savRevision hours, attendance, anxiety and exam score (N = 120)

Now practise it interactively →

Reading is only half of it. Tap through the real SPSS menus, read the output and pick the right write-up in 6 quick questions, with instant feedback on every answer.

Practice for this lesson is part of Pro, with every unit, unlimited hearts and saved progress.

More in Advanced regression