What ‘interpreting SPSS output’ really means
SPSS does the calculations; your job in an assignment is to read the output and explain what it means. Running the test is a few clicks — the marks are in interpreting the tables correctly, deciding whether your hypothesis is supported, and reporting the result in the conventional format. Students lose marks not because SPSS gave the wrong numbers but because they misread the tables, fixate on the wrong value, or report significance without effect size.
Across almost every test, the workflow is the same: state your hypotheses, find the significance (p) value and compare it with your chosen alpha level, report the test statistic and degrees of freedom, add an effect size, and then write a sentence in plain English about what it means for your research question. Master that pattern and SPSS output stops being intimidating.
The anatomy of SPSS output and the p-value
SPSS produces a series of labelled tables in the Output Viewer. The value you look for first in most inferential tests is in the ‘Sig.’ column — the p-value. The p-value is the probability of obtaining your result (or a more extreme one) if the null hypothesis were true. The convention is to compare it with an alpha (α) of .05: if Sig. is less than .05, you reject the null hypothesis and call the result statistically significant; if it is .05 or above, you fail to reject the null.
Two cautions. SPSS sometimes displays ‘.000’ — this does not mean zero; report it as p < .001. And statistical significance is not the same as importance: a tiny, trivial effect can be significant with a large sample, which is why effect size matters alongside the p-value. Note the test statistic (t, F, etc.) and the degrees of freedom (df) as well, because you report all three.
Descriptive statistics and reliability
Before any inferential test, you usually report descriptive statistics: the mean, standard deviation, minimum, maximum and sample size for each variable. SPSS gives these in the Descriptives or Frequencies output. They describe your data and let the reader judge its spread before you test anything.
For questionnaire studies you often report reliability using Cronbach’s alpha (Analyze → Scale → Reliability Analysis). Alpha ranges from 0 to 1; a value of .70 or above is generally considered acceptable internal consistency, .80–.90 good. You would write: the scale showed good internal consistency (α = .84). A low alpha suggests the items are not measuring a single coherent construct.
Reading t-test and ANOVA output
For an independent-samples t-test, SPSS first gives Levene’s Test for Equality of Variances: if Levene’s Sig. is above .05, read the ‘equal variances assumed’ row; if below, read ‘equal variances not assumed’. Then take t, df and the two-tailed Sig. You would report: t(48) = 2.31, p = .025, and state which group mean was higher.
For a one-way ANOVA, the key table is ‘ANOVA’, where you read the F statistic and its Sig. A significant F (p < .05) tells you the group means differ somewhere, but not where — so you then read the post-hoc tests (such as Tukey) to see which specific pairs of groups differ. Report as F(2, 87) = 4.56, p = .013, then describe the significant pairwise differences.
Reading correlation and regression output
For correlation (Pearson’s r), SPSS gives a matrix with the correlation coefficient and its Sig. The coefficient runs from −1 to +1: the sign shows direction, the size shows strength (roughly .1 weak, .3 moderate, .5+ strong). Report as r(98) = .42, p < .001, noting that correlation does not imply causation.
For regression, read three tables in order. The Model Summary gives R Square — the proportion of variance in the outcome explained by the predictors (R² = .35 means 35%). The ANOVA table tests whether the model as a whole is significant. The Coefficients table is where the action is: for each predictor, the standardised coefficient (Beta) shows its relative strength and the Sig. shows whether it is a significant predictor. You would report the model (F, R²) and then each significant predictor’s β and p-value.
Reading chi-square output
For categorical data you often use the chi-square test of independence. SPSS gives a crosstabulation and a ‘Chi-Square Tests’ table; read the Pearson Chi-Square row for the value, df and Asymptotic Sig. A significant result (p < .05) means the two categorical variables are associated. Check the footnote on expected cell counts — if too many cells have an expected count below 5, the test may be unreliable and you should note this. Report as χ²(1, N = 120) = 6.71, p = .010.
Reporting SPSS results in APA style
Results sections follow a conventional APA format that markers expect. State the test, the statistic with its degrees of freedom, the exact p-value and an effect size, then a plain sentence interpreting it. For example: An independent-samples t-test showed that exam scores were significantly higher for the revision-app group (M = 72.4, SD = 8.1) than the control group (M = 65.2, SD = 9.3), t(48) = 2.31, p = .025, d = 0.65. Italicise statistical symbols (t, F, r, p, M, SD), report p to three decimals (or p < .001), and round most values to two decimals. Present supporting tables cleanly and refer to them in the text rather than dumping raw SPSS tables into the report. See our APA referencing guide for the wider style rules.
Why effect size matters as much as significance
A result can be statistically significant yet practically trivial, especially with a large sample, so good analysis always reports an effect size alongside the p-value. Common measures are Cohen’s d for t-tests (roughly 0.2 small, 0.5 medium, 0.8 large), eta squared (η²) or partial eta squared for ANOVA, r itself for correlation, and R² for regression. Effect size answers the question significance cannot: not just ‘is there an effect?’ but ‘how big is it?’
SPSS does not always give effect sizes by default for every test, so you may need to request them or calculate them — and markers increasingly expect them. Reporting significance without effect size is one of the most common ways students lose marks in a results section, because it leaves the reader unable to judge whether a significant finding actually matters.
The most common SPSS interpretation mistakes
- Reading ‘.000’ as zero. Report it as p < .001, never p = .000.
- Confusing significance with importance. Always report an effect size to show how big the effect is.
- Ignoring Levene’s test in a t-test and reading the wrong row.
- Stopping at a significant ANOVA without running post-hoc tests to find which groups differ.
- Claiming causation from correlation or cross-sectional data.
- Dumping raw SPSS tables into the report instead of reporting results in APA format.
Checking assumptions before you trust the output
Before interpreting a parametric test in SPSS, check that your data actually meet its assumptions, because markers expect it and a violated assumption can invalidate your conclusion. For normality, inspect histograms and Q–Q plots and run the Shapiro–Wilk test (Analyze → Descriptive Statistics → Explore): a non-significant Shapiro–Wilk (Sig. > .05) suggests the data are approximately normal. For homogeneity of variance, use Levene’s test, which SPSS reports automatically inside t-test and ANOVA output.
Report briefly what you checked and found — ‘Shapiro–Wilk indicated the assumption of normality was met (p = .21), and Levene’s test showed equal variances (p = .64)’. This short paragraph signals methodological care and tells the reader your subsequent interpretation rests on solid ground. Skipping assumption checks and interpreting the test regardless is one of the quickest ways to lose marks in a statistics assignment.
When to switch to a non-parametric test
If your data seriously violate the assumptions — for example a badly skewed outcome or ordinal data — SPSS offers non-parametric alternatives that do not assume normality. Use the Mann–Whitney U test instead of an independent t-test, the Wilcoxon signed-rank test instead of a paired t-test, the Kruskal–Wallis test instead of a one-way ANOVA, and Spearman’s rho instead of Pearson’s correlation.
These tests work on ranks rather than raw values and are interpreted the same way — find the Sig. value and compare with .05 — though the statistic you report differs (for example U for Mann–Whitney). State clearly why you chose a non-parametric test (‘because the outcome was not normally distributed’), as the justification itself earns credit and shows you understand the link between data, assumptions and test choice.
Using SPSS syntax for accuracy and reproducibility
SPSS can be driven entirely through its menus, but every analysis can also be saved as syntax — SPSS’s command language. Clicking ‘Paste’ in any dialog box writes the equivalent syntax to a Syntax Editor window, which you can save and re-run. This matters for two reasons. First, reproducibility: a saved syntax file documents exactly what you did, so you (or a marker) can re-run the identical analysis and get identical output — valuable for a dissertation. Second, accuracy: re-running from syntax avoids the click-by-click errors that creep into menu-driven analysis when you repeat a test on several variables.
Including your syntax in an appendix is good practice for larger projects and is sometimes required. It demonstrates a transparent, repeatable method — the same principle behind sharing code in any data analysis — and protects you if a marker queries how a result was produced.
Stuck interpreting SPSS output? Our statistics specialists run your analysis in SPSS, interpret every table and write the results section in APA style.
Related guides
Frequently asked questions
Which value in SPSS output tells me if my result is significant?
The significance (Sig.) column — the p-value. If it is below your alpha level (usually .05) the result is statistically significant and you reject the null hypothesis. If it is .05 or above, you fail to reject the null.
What does ‘Sig. = .000’ mean in SPSS?
It does not mean exactly zero — SPSS has rounded a very small p-value. Report it as p < .001, never as p = .000.
What is a good Cronbach’s alpha?
Cronbach’s alpha measures internal consistency from 0 to 1. A value of .70 or above is generally acceptable, .80–.90 is good. A low alpha suggests the items are not measuring a single coherent construct.
Why do I need an effect size as well as a p-value?
Because a significant result can still be trivially small, especially with a large sample. Effect sizes such as Cohen’s d, eta squared or R² tell the reader how big the effect is, not just whether it exists — and markers increasingly expect them.
After a significant ANOVA, what do I do next?
A significant F tells you the group means differ somewhere but not where. Read the post-hoc tests (such as Tukey) to identify which specific pairs of groups differ significantly, and report those comparisons.
Can someone interpret my SPSS output and write the results section?
Yes — our statistics specialists run the analysis, interpret every table and write the results section in APA style. See our SPSS assignment help page or place an order.
Need your SPSS output run, interpreted and written up in APA style by a statistics specialist? Place an order or explore our SPSS assignment help — rated 4.4/5 across 871 verified Trustpilot and Sitejabber reviews.
