You have run your regression in SPSS. The output is significant, the coefficients look sensible, and now comes the part many students find hardest: writing it up so a reader, an examiner, or a journal reviewer can understand exactly what you did and what you found.
Reporting regression results in APA style (7th edition) is not about copying SPSS tables into your document. It is about selecting the right statistics, presenting them in the right order, and writing them in a format other researchers recognise instantly. This guide shows you what APA style requires, how to get each number from SPSS, and how to write it up for simple linear regression, multiple regression, and regression tables, with worked illustrative examples throughout.
What APA Style Requires for Regression
APA style asks for enough detail that another researcher could judge your analysis and, in principle, reproduce it. For a regression, that means reporting:
- The type of regression and the variables involved, including which variable is the outcome and which are predictors.
- The overall model fit: R-squared, and usually adjusted R-squared, plus the F test with its degrees of freedom and p-value.
- The individual predictors: unstandardised coefficients (B), standard errors, standardised coefficients (Beta), t-values, p-values, and confidence intervals where appropriate.
- The sample size, so readers can check the degrees of freedom.
- Any assumption checks you carried out, briefly, if they affect interpretation.
You do not need to report every number SPSS produces. Descriptive statistics, collinearity diagnostics, and residual checks belong in your analysis notes and are summarised briefly in text, not dumped wholesale into the results section.
Getting the Numbers from SPSS
If you have not yet run the model, here are the click-by-click steps so your output contains everything you need for APA reporting.
Step 1: Go to Analyze, then Regression, then Linear.
Step 2: Move your outcome variable into the Dependent box and your predictor or predictors into the Independent(s) box.
Step 3: Click Statistics and tick Estimates, Model fit, and Confidence intervals. Estimates gives you B, standard error, Beta, t, and Sig. Model fit gives you R-squared and the ANOVA F test. Confidence intervals give you the 95 percent confidence interval for each B, which APA style encourages. Click Continue, then OK.
Paste-Ready SPSS Syntax
Paste this into a Syntax window and run it, replacing the variable names with your own, to reproduce the same output:
REGRESSION /MISSING LISTWISE /STATISTICS COEFF OUTS CI(95) R ANOVA /CRITERIA=PIN(.05) POUT(.10) /NOORIGIN /DEPENDENT outcome /METHOD=ENTER predictor1 predictor2 predictor3.
The CI(95) keyword is what adds confidence intervals to the Coefficients table, and R and ANOVA add the model summary and overall F test.
Reading the Three Tables You Need
Model Summary: This table gives R, R-squared, Adjusted R-squared, and the standard error of the estimate. R-squared is the proportion of variance in the outcome explained by the model. Adjusted R-squared adjusts that figure for the number of predictors, and it is the fairer figure to emphasise when models have different numbers of predictors.
ANOVA: This table tests whether the model as a whole predicts the outcome significantly better than a model with no predictors. Read the F value, its two degrees of freedom (regression and residual), and the Sig. value, which is the p-value for the overall model.
Coefficients: This table gives one row per predictor plus the constant (intercept). For each predictor, read B (the unstandardised coefficient), Std. Error, Beta (the standardised coefficient), t, Sig. (the p-value), and the lower and upper bounds of the 95 percent confidence interval. The constant row is usually reported only if the intercept itself is meaningful for your research question; in many write-ups it is omitted from the text and shown in the table only.
Reporting Simple Linear Regression
Simple linear regression has one predictor. In text, report the model fit first, then the predictor. An illustrative example, with made-up numbers used only to show the format, reads:
“A simple linear regression was conducted to predict exam score from hours of study. The model was significant, F(1, 98) = 42.36, p less than .001, and explained 30% of the variance in exam score, R-squared = .30. Hours of study was a significant positive predictor of exam score, B = 2.45, SE = 0.38, Beta = .55, t(98) = 6.51, p less than .001, 95% CI [1.70, 3.20].”
Notice the pattern: F with degrees of freedom in parentheses, then the p-value, then R-squared. For the predictor: B, standard error, Beta, t with degrees of freedom, p-value, and confidence interval. Report exact p-values where SPSS gives them (for example, p = .024), and use p less than .001 only when SPSS rounds the value to .000. Never write p = .000, because a p-value is never exactly zero.
Reporting Multiple Regression
Multiple regression follows the same pattern, with one extra step: state which predictors were entered, then report the overall model, then each predictor in turn. An illustrative example reads:
“A multiple regression was conducted to predict job satisfaction from workload, supervisor support, and years of experience. The overall model was significant, F(3, 196) = 28.14, p less than .001, R-squared = .30, adjusted R-squared = .29, indicating that the three predictors together explained 30% of the variance in job satisfaction. Supervisor support was a significant positive predictor, B = 0.62, SE = 0.11, Beta = .41, t(196) = 5.64, p less than .001, 95% CI [0.40, 0.84]. Workload was a significant negative predictor, B = -0.28, SE = 0.09, Beta = -.19, t(196) = -3.11, p = .002, 95% CI [-0.46, -0.10]. Years of experience was not significant, B = 0.04, SE = 0.06, Beta = .05, t(196) = 0.67, p = .504, 95% CI [-0.08, 0.16].”
Report non-significant predictors too. Omitting them gives a misleading picture of the model you actually ran. If a predictor is non-significant, say so plainly and give its statistics, as in the example above.
Presenting Regression Results in a Table
When a model has several predictors, a table is clearer than a long paragraph. APA tables typically show, for each predictor, B, standard error, Beta, t, p, and sometimes the confidence interval, with the model-level statistics (R-squared, adjusted R-squared, and F) in a note below the table or in the text that introduces it. Keep the table to the statistics readers need: do not reproduce SPSS’s full output, with its extra columns, into your document. Give the table a clear number and title above it, such as Table 1, Multiple Regression Predicting Job Satisfaction, and refer to it in the text before it appears.
Common Mistakes to Avoid
First, do not report R without saying whether you mean R or R-squared. They are different numbers, and reviewers will notice.
Second, do not confuse B and Beta. B is in the units of your variables and is what you use to describe expected change in the outcome. Beta is standardised and is what you use to compare the relative strength of predictors measured on different scales.
Third, do not write that a result “proves” an effect. Regression shows association and prediction, not causation, unless your design supports a causal claim. Use language such as “predicted” or “was associated with.”
Fourth, do not round everything to whole numbers. Report coefficients to two decimal places as a minimum, and keep enough precision that a reader could reconstruct your claims.
Fifth, do not forget the degrees of freedom. An F or t value without degrees of freedom cannot be checked, and APA style requires them.
A Brief Note on Assumptions
A results section does not need a full assumption audit, but a sentence or two shows diligence. For example, you might write that residual plots showed no serious departures from linearity or homoscedasticity, and that variance inflation factors were within acceptable limits. If an assumption was violated and you took action, such as transforming a variable, say what you did and why. Readers trust a write-up that acknowledges its checks far more than one that pretends they were unnecessary.
Key Takeaways
- Report the overall model first (F, degrees of freedom, p, R-squared, adjusted R-squared), then each predictor.
- For each predictor, give B, standard error, Beta, t, p, and the 95% confidence interval.
- Report exact p-values except when SPSS rounds to .000, in which case write p less than .001.
- Report non-significant predictors as well as significant ones.
- Use a table for multiple regression with several predictors, and keep it to the statistics readers need.
Frequently Asked Questions
Should I report B or Beta?
Report both. B tells readers the effect in real units, and Beta lets them compare predictors on different scales. In text, B usually comes first, followed by Beta.
Do I need to report the constant (intercept)?
Only if it is meaningful for your question. In most write-ups the intercept appears in the table but is not discussed in the text, because it represents the predicted outcome when all predictors are zero, which is often not a realistic value.
What if my overall model is significant but no individual predictor is?
This can happen when predictors share variance. Report the model as significant, report each predictor honestly, and discuss the overlap rather than hiding the non-significant coefficients.
How do I report R-squared as a percentage?
Multiply by 100 in words, for example, “explained 30% of the variance,” while giving the decimal form (R-squared = .30) in the statistical notation. Do both, not one or the other, in your first mention.
Need this done on your data? Get a free quote