How do you run a multiple regression in SPSS?

Open Analyze, then Regression, then Linear. Put your continuous outcome in Dependent and your predictors in Independents, leaving the method on Enter. Under Statistics and Plots request the collinearity diagnostics, Durbin-Watson, and residual plots, then click OK and read the Model Summary, ANOVA, and Coefficients tables.

How do you check for multicollinearity in SPSS?

Tick Collinearity diagnostics under Statistics in the Linear Regression dialog. In the Coefficients table, read tolerance and the variance inflation factor for each predictor. Tolerance below 0.1 or a variance inflation factor above about 10 signals multicollinearity serious enough to make the individual coefficients unstable.

What is the difference between B and Beta in SPSS regression?

B is the unstandardised coefficient: the change in the outcome for a one-unit change in the predictor, in the original units, used to describe a concrete effect. Beta is the standardised coefficient, which puts all predictors on the same scale so you can compare their relative importance. Report both in a dissertation.

Should you use stepwise regression in SPSS?

Avoid stepwise methods for confirmatory research. Letting the software select predictors by p-value capitalises on chance and produces models that rarely replicate in a new sample. For a dissertation, enter the predictors your theory specifies using the Enter method, so the model tests a hypothesis rather than mining the data.

How to run a multiple regression in SPSS

To run a multiple regression in SPSS, you open Analyze > Regression > Linear, enter one continuous outcome and two or more predictors, request the diagnostic statistics, and read the coefficients table to see which predictors matter. The model that survives examination depends on checking several assumptions, watching for multicollinearity, and reporting standardised coefficients alongside the raw ones. This guide covers the full procedure, the diagnostics, and the reporting.

What multiple regression answers

Multiple linear regression predicts a single continuous dependent variable from a set of predictors, and tells you the unique contribution of each predictor while holding the others constant. That phrase, holding the others constant, is the whole reason to use it rather than a series of separate correlations: it separates overlapping influences. When your outcome is binary rather than continuous, the matching method is logistic regression in SPSS instead, because linear regression cannot model a yes-or-no result well.

The assumptions, and how to check each in SPSS

Multiple regression carries a stack of assumptions, and SPSS produces a check for every one if you ask. Request them in the dialog: under Statistics, tick Collinearity diagnostics and Durbin-Watson; under Plots, plot the standardised residuals (ZRESID) against the standardised predicted values (ZPRED) and add the histogram and normal probability plot; under Save, store Cook's distance.

Assumption	How to check it in SPSS
Linearity and equal variance of residuals	The ZRESID against ZPRED plot shows a shapeless band
Normality of residuals	Histogram and normal P-P plot of residuals
Independence of errors	Durbin-Watson near 2 in the Model Summary
No multicollinearity	Tolerance above 0.1 and VIF below about 10
No overly influential cases	Cook's distance below 1

The full reasoning behind these is in linear regression assumptions, which is worth reading before you commit to a model.

Running the model and the syntax to keep

Open Analyze > Regression > Linear, put the outcome in Dependent and the predictors in Independent(s), and leave the method on Enter unless you have a specific reason otherwise. Convert any categorical predictor into dummy variables first, because regression needs numeric contrasts. The reproducible syntax is:

REGRESSION
  /STATISTICS COEFF OUTS CI(95) R ANOVA COLLIN TOL CHANGE
  /DEPENDENT achievement
  /METHOD=ENTER motivation hours_study prior_gpa
  /RESIDUALS DURBIN
  /CASEWISE PLOT(ZRESID) OUTLIERS(3)
  /SAVE COOK.

Avoid the Stepwise method for confirmatory work: letting the software pick predictors by p-value capitalises on chance and produces models that rarely replicate. Enter the predictors your theory specifies.

Reading the regression output

Read three tables in order. The Model Summary gives R, R squared (the proportion of variance the model explains), the adjusted R squared (which penalises extra predictors and is the honest figure to quote), and the Durbin-Watson statistic. The ANOVA table gives the overall F test of whether the model as a whole predicts better than the mean. The Coefficients table is where the findings live: the unstandardised B and its confidence interval, the standardised Beta, the t test and significance for each predictor, and the tolerance and VIF for multicollinearity. Reading this carefully is the core of interpreting SPSS output.

Unstandardised against standardised coefficients

The unstandardised coefficient (B) tells you how many units the outcome changes for a one-unit rise in a predictor, in the original measurement units, which is what you use to describe a concrete effect or build a prediction. The standardised coefficient (Beta) puts every predictor on the same scale so you can compare their relative importance, since predictors measured in different units are otherwise impossible to rank. Report both: B for meaning, Beta for comparison. A predictor can be statistically significant yet have a small Beta, which is exactly the nuance an effect size is meant to surface.

Reporting multiple regression in APA style

Report the overall model, then the individual predictors, usually in a table. A clean passage reads: the model significantly predicted achievement, F(3, 116) = 18.4, p < .001, and explained 32 percent of the variance (adjusted R squared = .32). Motivation was the strongest unique predictor, B = 0.41, 95 percent confidence interval [0.26, 0.56], Beta = .38, p < .001, while prior study hours did not contribute once the other predictors were accounted for, p = .21. State that you used the Enter method and that the assumptions were checked. The conventions are in the APA rules for reporting statistics, feeding the wider results chapter.