Open Analyze > Regression > Linear, move your outcome into Dependent and your one or more predictors into Independent(s), request the residual plots and collinearity diagnostics, then run it. You read three tables in order: the Model Summary for R Square, the ANOVA table for whether the model is significant, and the Coefficients table for each predictor. This guide shows the menu path, the exact values to read, the assumption checks, and how to report the result in APA.
Regression is where SPSS gives you the most output for one click, and where students lose the most marks by reading the wrong table. Simple linear regression uses one predictor to explain one continuous outcome. Multiple regression uses two or more predictors at once. The menu path is the same for both, so the only difference on screen is how many variables you move into the Independent(s) box. This guide is part of our SPSS homework help and our wider statistics homework help. It walks the setup, the three tables you must read, the five assumptions and how to check each one, and the sentence that reports it all.
One idea to hold onto. Regression answers two separate questions. First, does the model as a whole work? That is the ANOVA table. Second, which predictors carry the model, and how strongly? That is the Coefficients table. The Model Summary sits between them and tells you how much variance the model explains. Read them in that order and the output stops being a wall of numbers.
Same menu path, same three tables. The number of predictors decides which extra columns matter.
| Your question | Your data | Model |
|---|---|---|
| Does one variable predict a continuous outcome? | One continuous predictor, one continuous outcome | Simple linear regression |
| Do several variables together predict a continuous outcome? | Two or more predictors, one continuous outcome | Multiple regression |
| Which predictor matters most once the others are held constant? | Two or more predictors on different scales | Read Standardized Beta |
| Are my predictors too closely related? | Two or more predictors that may overlap | Check Tolerance and VIF |
A categorical predictor with more than two categories needs dummy coding first. A binary outcome needs logistic regression, not linear.
Menu path, the settings that give you the assumption checks, the three tables to read, and how to report it. Example values follow a standard teaching dataset predicting exam score from revision hours and test anxiety.
The dialog is the same for one predictor or many. Add two extra requests before you run so the assumption output appears.
For simple linear regression you move one variable into Independent(s). For multiple regression you move two or more. Everything else is identical. Leave the Method as Enter, which forces all predictors into the model at once and is the default for a stated hypothesis. Stepwise methods let SPSS pick predictors and should only be used when your brief asks for them.
The constant. SPSS includes a constant, the intercept, automatically. It is the predicted outcome when every predictor equals zero, and it appears as the (Constant) row in the Coefficients table.
Tells you the overall strength of the model in one number, R Square.
What to read. R is the correlation between the observed and predicted outcome. R Square is the share of variance in the outcome that the model explains, from zero to one. Adjusted R Square corrects R Square for the number of predictors and is the figure to report for multiple regression, because adding any predictor can only push plain R Square up. Std. Error of the Estimate is the typical size of the prediction error, in the units of the outcome.
| Model | R | R Square | Adjusted R Square | Std. Error of the Estimate | Durbin-Watson |
|---|---|---|---|---|---|
| 1 | .760 | .577 | .563 | 8.14 | 1.98 |
Here R Square is .577, so the two predictors explain about 58 percent of the variance in exam score. The Durbin-Watson value of 1.98 sits close to 2, which supports independence of the residuals, covered under the assumptions below.
Tests whether the model as a whole predicts the outcome better than simply guessing the mean.
What to read. The Regression row, for F and Sig. If Sig. is below .05, the model is statistically significant: your predictors together explain a reliable amount of variance. If Sig. is .05 or above, the model does not predict the outcome and the individual coefficients are not worth interpreting.
| Sum of Squares | df | Mean Square | F | Sig. | |
|---|---|---|---|---|---|
| Regression | 5157.6 | 2 | 2578.8 | 38.90 | .000 |
| Residual | 3778.4 | 57 | 66.3 | ||
| Total | 8936.0 | 59 |
The Sig. of .000 means the model is significant. SPSS rounds tiny p-values to .000, so you report this as p < .001, never as p = .000. The degrees of freedom for the F test are 2 and 57, taken from the Regression and Residual rows.
Gives the equation and tests each predictor. This is the table with the most reporting value.
What to read. Each row is one predictor plus the (Constant). Unstandardized B is the change in the outcome for a one-unit rise in that predictor, holding the others constant. Std. Error is the precision of that estimate. Standardized Beta puts every predictor on the same scale so you can compare their relative strength; the predictor with the largest absolute Beta is the strongest. t and Sig. test whether that predictor adds significantly to the model: a Sig. below .05 means it is a significant contributor.
| B | Std. Error | Beta | t | Sig. | Tolerance | VIF | |
|---|---|---|---|---|---|---|---|
| (Constant) | 52.10 | 4.30 | 12.12 | .000 | |||
| Revision hours | 1.94 | 0.28 | .55 | 6.93 | .000 | .82 | 1.22 |
| Test anxiety | -0.76 | 0.22 | -.31 | -3.45 | .001 | .82 | 1.22 |
Both predictors are significant. Revision hours has the larger Beta, .55 against -.31, so it is the stronger predictor. The regression equation is: predicted exam score = 52.10 + 1.94 × revision hours - 0.76 × test anxiety. Each extra hour of revision adds about 1.94 points to the predicted score once anxiety is held constant. The Tolerance of .82 and VIF of 1.22 are both well inside the safe range, so multicollinearity is not a problem here.
Regression has five assumptions you can check inside SPSS. A model that breaks them can look tidy and still be wrong.
| Assumption | How to check it in SPSS | What you want to see |
|---|---|---|
| Linearity | Scatterplot of ZRESID against ZPRED, from the Plots button | Points scattered evenly around zero with no curve |
| Independence of errors | Durbin-Watson statistic in the Model Summary | A value near 2, roughly between 1.5 and 2.5 |
| Homoscedasticity | Same ZRESID against ZPRED scatterplot | A constant band of spread, not a funnel shape |
| Normality of residuals | Normal P-P plot of the residuals, from the Plots button | Points hugging the diagonal line |
| No multicollinearity | Tolerance and VIF in the Coefficients table | VIF below 10, Tolerance above 0.1 |
Note that regression checks normality on the residuals, not on the raw variables, which is a frequent source of confusion. The residual plots come from the Plots button you set earlier, and the collinearity diagnostics come from the Collinearity diagnostics checkbox. When homoscedasticity or linearity fails, transforming a variable or adding a missing predictor often fixes it; when it cannot be fixed, a resistant or non-linear model may be needed.
Residuals fanning out, a Durbin-Watson far from 2, or VIF values climbing on your own dataset? Our statisticians will diagnose it, apply the right fix and interpret the model. Get a quote →
The marks are in the sentence, not the pasted tables. Copy these patterns and swap in your own numbers.
Common mistakes
Procedures on this page were checked against the IBM SPSS Statistics documentation and the UCLA statistical computing regression output guide.
When the assumptions break, the predictors overlap, or the deadline is unforgiving, a specialist runs the model in SPSS on your file and explains the output so you can defend it.
Upload your data file, the assignment brief and your deadline, then get a free quote from support.
Approve the price and a statistician who works in SPSS every day starts right away.
Get the output, interpretation and write-up in your account, with free revisions if you need them.
"Very good and friendly customer service. Most importantly give great results on time. Im very I chose homeworkdoer to do work for me. Excellent service."
"Such great customer service and really helped out when I was in a jam for time and made awesome grades. Thank you so much."
R Square is the proportion of variance in the outcome that the model explains, on a scale from zero to one. An R Square of .577 means the predictors account for about 58 percent of the variance. For multiple regression, report Adjusted R Square, which corrects for the number of predictors in the model.
Each row is one predictor plus the Constant. Unstandardized B is the change in the outcome for a one-unit rise in that predictor, holding the others constant. Standardized Beta puts every predictor on the same scale so you can compare their strength. If the Sig. for a predictor is below .05, it is a significant contributor.
VIF, the variance inflation factor, flags multicollinearity. A common rule is that VIF below 10 is acceptable and below 5 is comfortable, while Tolerance, which is one divided by VIF, should stay above 0.1. Past those thresholds, two or more predictors are too closely related and the individual coefficients become unstable.
Simple linear regression has one predictor and multiple regression has two or more. The menu path is identical, Analyze then Regression then Linear; you just move more variables into Independent(s). With multiple predictors you also read the Standardized Beta column and check the VIF values.
Check linearity and homoscedasticity on the scatterplot of standardized residuals against standardized predicted values, independence with the Durbin-Watson statistic near 2, normality of residuals on the Normal P-P plot, and multicollinearity with Tolerance and VIF. Request the plots and collinearity diagnostics from the Statistics and Plots buttons before running the model.
It tests whether the model as a whole predicts the outcome better than the mean alone. Read F and its Sig.: if Sig. is below .05 the model is significant, meaning the predictors together explain a reliable amount of variance. It does not tell you which predictor matters; that is the Coefficients table.
Hand it to a statistician and get accurate output, checked assumptions and a clear interpretation before your deadline.
Get my free quote →