METHOD GUIDE
Simple Linear Regression in SPSS
The Slope, R Squared and the Residual Plots
With one predictor, SPSS prints several tables, and the slope, R squared and residual plots each answer a different question. Of the three, the slope with its interval usually carries the answer, and the residual plots decide whether it can be trusted.
IN SHORT
What the Fitted Line Estimates and What It Cannot Show
A simple linear regression models the mean of a continuous outcome as a straight-line function of one predictor and estimates the slope, the expected difference in the outcome for a one-unit difference in the predictor. It serves prediction and the description of association, and it supports causal wording only where the design does. To run simple linear regression in SPSS, open Analyze, Regression, Linear, enter the outcome as Dependent and the predictor as Independent, request confidence intervals under Statistics and a plot of standardized residuals against predicted values under Plots. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, carries out and reviews this analysis through an in-house PhD-level statistician.
ANALYSIS SUPPORT
Analysis Support for This Method
A single predictor is rarely the whole story, and the model that fits the design may need more terms. Both pages below start from the research question and not from the dialog.
Simple Linear Regression
The one-predictor model fitted within your study design, with the checks on linearity and influential points and the reporting of the slope agreed for your data.
Multiple Linear Regression
A model with several predictors, for when other variables must be held constant and the slope of one predictor is read adjusted for the rest.
BEFORE YOU RUN IT
Questions to Settle Before Fitting the Line
A line can always be fitted, so the useful checks are whether it should be and what the fitted slope will be allowed to mean.
A Line That Fits the Pattern The scatterplot should show a roughly straight trend. A curve calls for a transformed predictor or a polynomial term, and a straight line fitted to it misdescribes the data.
Units and the Meaning of Zero The slope is in outcome units per predictor unit, and the intercept is the expected outcome at predictor zero. If zero lies outside the observed range, centering the predictor gives the intercept a meaning.
One Pair per Case Each case contributes one pair of values. Repeated measures, or participants nested in classes or clinics, need a model that accounts for the dependence between observations.
Prediction or Association A model built for prediction is judged by its error on new data, while one built to describe association is judged by the slope and its interval. The aim decides what is reported.
IN SPSS
How to Run Simple Linear Regression in SPSS
Simple regression uses the same Linear dialog of IBM SPSS as a model with several predictors, so only the one-predictor settings are listed. Its labels shift a little between versions.
- Plot the outcome against the predictor and look for a straight-line trend and extreme points.
- Open Analyze, Regression, Linear, and enter the outcome under Dependent and the predictor under Independent(s).
- Under Statistics, tick Confidence intervals for the coefficients, and keep Estimates and Model fit.
- Under Plots, plot standardized residuals (*ZRESID) against standardized predicted values (*ZPRED).
- Under Save, request Cook’s distance, so that influential cases can be examined.
- Run it and read the Model Summary, the ANOVA table and the Coefficients table in turn.
READING THE OUTPUT
What Each Table and Plot of a One-Predictor Model Contributes
The SPSS model summary R squared is printed first, but the slope in the Coefficients table is what answers most research questions.
| Output element | What it tells you | What to check |
|---|---|---|
| Model Summary: R and R Square | R is the absolute correlation of predictor and outcome, and R Square is the share of outcome variance the line accounts for in this sample | That R squared is not read as the quality of prediction in new data, nor as a causal share |
| Std. Error of the Estimate | The typical distance between observed outcomes and the fitted line, in outcome units | That it is interpreted in the units of the outcome, as a measure of prediction error in this sample |
| ANOVA | An F test of the whole model, which with one predictor is the square of the slope’s t | That it is not reported as separate evidence, because it adds nothing beyond the slope’s test here |
| Coefficients: B and Std. Error | The intercept and the slope in raw units, with the standard errors | That the slope is stated in the units of both variables, with the 95.0% Confidence Interval for B |
| Coefficients: Beta | The slope in standard-deviation units, which here equals the correlation | That it is added only where the units have no natural meaning, not instead of the raw slope |
| Residual scatterplot (*ZRESID against *ZPRED) | Whether the residuals form a shapeless band around zero | That no curve, funnel or isolated point appears before the slope is interpreted |
| Cook’s distance (saved variable in Data View) | The influence of each case on the fitted line | That cases standing out from the rest are examined and their effect on the slope is reported |
REPORTING
Reporting the Slope, Its Interval and R Squared
The slope leads, in the units of both variables, and R squared follows as context rather than as the headline.
Report the slope in raw units with its interval: B = ___, 95% CI [___, ___], t(df) = ___, p = ___, then R² = ___ and the sample size. The Coefficients table prints this interval when it is requested under Statistics, whereas the Model Summary gives no interval for R squared, so any interval for it comes from another program. Describe the slope in words with both units, as the expected change in the outcome for one unit of the predictor, and call it an association unless the design supports a causal reading. Mention the residual plot and any influential case, and say whether the slope was estimated with or without it. Predictions stay within the observed range of the predictor.
COMMON MISTAKES
Misreadings of a Fitted Line That Reviewers Catch
The recurring mistakes concern how the numbers are read, and the first two come from treating R squared as more than it is.
R Squared as Quality A high R squared does not show that the model is right, and a low one does not make a slope worthless. The slope, its interval and the residual plots carry the judgment.
Slope as Cause The regression slope in SPSS from observational data describes how the outcome differs across the predictor, not what would follow if the predictor were changed. Confounders remain.
Extrapolation Predictions beyond the observed range of the predictor assume that the line simply continues. The range is stated, and predictions are kept within it.
Plot Never Opened A model summary with a respectable R squared can sit beside a curve or a funnel in the residuals. The residual plot is read together with the tables.
Outcome and Predictor Swapped Regressing Y on X and X on Y gives different slopes, and only one answers the question asked. The direction follows the research question, not the order of the columns.
Beta Without Units Reporting only the standardized beta hides what one unit of the predictor is worth. The raw slope and its interval come first, and beta is an addition.
Line already fitted? Independent Statistical Review checks the residual plots, the slope interval and the wording of the result before you submit.
FREQUENTLY ASKED QUESTIONS
Questions About One-Predictor Regression Output
Each answer below starts from a table or plot that SPSS printed for a one-predictor model, beginning with what R squared can and cannot tell you.
What does R squared mean in the SPSS model summary?
R squared is the proportion of the outcome’s variance that the fitted line accounts for in your sample. Adjusted R squared is lower, by more when the sample is small, because it corrects for the number of predictors and cases. What counts as a high value depends on the field, since human behavior is rarely predicted tightly, and neither value shows that the predictor causes the outcome.
Why is the Beta in my Coefficients table the same as my correlation?
With one predictor, the standardized slope equals the Pearson r between the predictor and the outcome. For the same reason the F in the ANOVA table equals the square of the slope’s t, and R Square equals r squared. A second predictor is what makes these quantities differ.
How do I read the residual plot SPSS produced?
Standardized residuals against standardized predicted values should form a shapeless band centered on zero. A curve points to a relationship that is not straight, a funnel to a spread that changes with the predicted value, and an isolated point to a case that deserves a closer look. In small samples the plot is read with caution, because chance patterns appear easily.
Can OnlineSPSS check my regression output before I report it?
Yes. An in-house PhD-level statistician can review the residual plots, the slope and its interval and the interpretation of R squared, or carry out the model on your data from the start. When the model is settled and only the tables need work, Results Reporting & Interpretation covers that stage. The research question and the conclusions remain with you.
How do you protect my research data and confidentiality?
Your data and project details are handled confidentially, and project information is not published. Researcher names and institutions are not used publicly without permission. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You share your research question, design and relevant materials, such as the dataset or the output you have, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
A Fitted Line and
Doubts About Reading It?
Send the predictor and the outcome with their units, together with the model summary and the residual plots if you have them. Those materials show whether a straight line suits your data and what the report should rest on, and the scope follows from them.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014