How to Report Non Significant Results in SPSS: Interpreting and Writing Up a p > .05 Finding
I will tell you exactly how to report non significant results in SPSS just gave you, I already know what your screen looks like. A Sig. value sitting somewhere north of .05, a blinking cursor in your Chapter 4, and a small voice asking if your whole dissertation just collapsed. It did not. Let me walk you through exactly what that number means and exactly what to type next.
I have read this exact SPSS output on hundreds of student laptops over the last 12 years. The reaction is always the same: panic first, question second. So let’s skip the panic and go straight to the part that actually gets you through your viva.
Quick definition: A non significant result means your sample data did not give you strong enough evidence to reject the null hypothesis at your chosen alpha level, usually .05. It does NOT mean you proved the null hypothesis true. It means your test simply did not find enough evidence either way.
What a Non-Significant Result Actually Means (and What It Doesn’t)
A non significant hypothesis SPSS test doesn’t pass judgment on your research skills or your research design by itself. It just means your data didn’t clear the bar you set for evidence. Here is what that actually implies, test by test, and what it clearly does not.
p > .05 Does Not Mean “No Effect Exists”
A p-value only tells you how likely your observed data would be if the null hypothesis were true. It is not a measure of how big or how real your effect is. A result of p = .08 could still reflect a genuine small effect that your sample size was just too small to detect clearly.
This is the single biggest misunderstanding I see when students don’t know how to interpret non significant results on their own. They jump from “not significant” straight to “there is no relationship at all,” which is a much bigger and often wrong claim.
Should I Accept or Reject the Null Hypothesis?
You fail to reject the null hypothesis. You do not accept it. The alternative hypothesis is the one claiming a real difference or relationship exists, and failing to reject the null only means your data didn’t give you strong enough evidence for that alternative, not that the alternative is false. This distinction is not just academic nitpicking, examiners genuinely check for this exact phrasing.
Saying “we accept the null hypothesis” tells your committee you don’t fully understand hypothesis testing logic. Saying “we fail to reject the null hypothesis, suggesting insufficient evidence for the alternative hypothesis” tells them you do. One sentence, one word swap, a very different impression.
Why Your SPSS Output Might Be Showing No Significance
Before you write a single word of your discussion chapter, figure out WHY the result came out this way. There are usually three reasons, and your write-up should reflect which one applies to you.
Small Sample Size and Low Statistical Power
If your n is small, your test simply doesn’t have enough power to pick up anything but a large effect. This risk has a specific name, a Type II error, which happens when a real effect exists but your study isn’t powerful enough to detect it.
This is the most common reason I see in student dissertations, especially with convenience samples under 100 respondents. Running a quick sample size calculation before data collection usually catches it early. If you’ve already collected your data, this is still worth naming directly in your limitations section.
High Variability or Noisy Data
Sometimes the effect is there but your data is too scattered around it. Wide standard deviations, outliers, or a poorly controlled variable can all hide a real pattern. Check your assumptions properly using something like the Shapiro-Wilk normality guide before you conclude the effect simply isn’t there.
A Genuinely Weak or Absent Real-World Effect
Sometimes, and this is completely fine academically, there just isn’t a meaningful relationship between your variables. This is still a valid, publishable finding. A dissertation with hypothesis not supported is not automatically a weaker dissertation, it just needs a more careful discussion section, which I’ll cover below.
How to Report Non-Significant Results in SPSS: Reading the Output Tables
Here is where most online guides give up and just say “check your Sig. column.” That’s not enough on its own. IBM’s own SPSS documentation confirms the Sig. column is the exact two-tailed significance value you should read first in every one of these four tables, so let’s go table by table.
- Independent Samples t-Test Table: Look at the “Sig. (2-tailed)” row under either “Equal variances assumed” or “not assumed,” depending on your Levene’s test result. If this value is above .05, your groups did not differ significantly.
- ANOVA Output Table: Check the Sig. column in the main ANOVA summary table, next to your F value. A non significant ANOVA dissertation result means none of your groups differed meaningfully on the outcome variable. If you were planning post-hoc comparisons anyway, it’s worth understanding when to use Tukey HSD versus Games-Howell first, since running the wrong follow-up test on an already non-significant result is a common mistake.
- Regression Coefficients Table: Look at the Sig. column next to each predictor’s B value in the Coefficients table. A non significant regression results SPSS output simply means that specific predictor did not meaningfully explain variation in your dependent variable, even if your overall model R² looks decent.
- Chi-Square Output: Check the Asymp. Sig. (2-sided) value in the Chi-Square Tests table. Above .05 means no significant association between your categorical variables.
Reading the table correctly is step one. Writing it up correctly is step two, and that’s where marks are actually won or lost.
How to Report a Non-Significant p-Value in APA Format
This is the part students usually search for directly: how to report p value greater than 0.05 without sounding unsure of yourself.
APA Rules for Reporting Exact p-Values
The APA Publication Manual, 7th edition, asks you to report the exact p-value to two or three decimal places, not just “p > .05,” even when the result is non significant. If SPSS shows .000, you report it as p < .001, since SPSS rounds down and never actually displays a true zero.
Copy-Ready APA Sentences
Here are templates you can adapt directly, test by test:
- t-test: “There was no statistically significant difference in scores between Group A (M = 24.3, SD = 4.1) and Group B (M = 23.8, SD = 4.6); t(58) = 0.52, p = .605.”
- ANOVA: “A one-way ANOVA revealed no statistically significant effect of group on outcome scores, F(2, 87) = 1.44, p = .242.”
- Regression: “Predictor X was not a statistically significant predictor of the outcome variable, B = 0.12, SE = 0.09, p = .178.”
- Chi-Square: “There was no statistically significant association between the two categorical variables, χ²(1, N = 150) = 1.76, p = .185.”
Adapt these to your own numbers, don’t just copy them in blindly. Examiners do notice mismatched sample sizes.
How to Discuss Non-Significant Findings in Your Dissertation
Reporting the number is easy. Learning how to discuss non significant findings dissertation committees actually accept, that takes a bit more thought.
Structuring Chapter 5 Around a Null Finding
Start by restating what you expected and what you found. Then explain the possible reasons, sample size, measurement issues, or a genuinely absent effect. Finally, connect it back to existing literature instead of treating it as an isolated failure.
Phrases to Use vs Phrases to Avoid
Avoid phrases like “almost significant,” “approaching significance,” or “trending towards significance.” These are considered weak academic language and most examiners will flag them. Instead use “the result did not reach statistical significance” or “there was insufficient evidence to support the hypothesis.”
Tying the Result Back to Your Literature Review
If your result contradicts an earlier study, say so directly. Maybe their sample, context, or measurement tool differed from yours, and that’s worth naming. If it aligns with a body of mixed findings in your field, mention that too. This is where a hypothesis not supported dissertation actually becomes more interesting to examine, not less.
Short case study: One postgraduate student I worked with, let’s call her Anjali (a composite example reflecting a pattern I see often, not one specific person), ran a regression testing whether social media usage predicted academic procrastination. Her result came back non significant, B = 0.09, p = .310. Instead of panicking, she wrote a tight paragraph connecting it to two other studies with similarly weak effects in Indian university samples, and flagged her convenience sample as a likely reason. Her supervisor’s only comment was “good, honest discussion,” no resubmission needed.
Should You Still Report Effect Size and Confidence Intervals?
Yes, always. A non significant p-value paired with a reasonably sized effect and a wide confidence interval tells a very different story than one paired with a tiny effect size near zero. Unlike a p-value, which only tells you whether your result crossed a threshold, a confidence interval shows you the plausible range for the true effect size, which is often more useful than the p-value alone when your result lands above .05.
Reporting Cohen’s d, eta-squared, or R² alongside your p-value shows your examiner you understand the difference between statistical significance and practical significance. That single detail is often exactly what separates a pass from a distinction.
What to Do Next: Rerun, Reframe, or Report As-Is
So, what if results are not statistically significant and you’re staring down a submission deadline? You generally have three honest paths: rerun only if there’s a real error, reframe the discussion around why it happened, or report it as-is with a careful explanation. You don’t need to invent a brand new hypothesis either. Most committees expect you to discuss why the original one wasn’t supported rather than retrofitting a different one after the fact.
Should I Rerun My Test If It’s Not Significant?
It is completely fine to rerun a test if you spot a genuine coding error, a wrongly entered variable, or a missed assumption check. It is not fine to keep swapping tests or trimming data points until something crosses .05. That second behaviour is called p-hacking, and experienced examiners can usually spot it from a mismatched methodology section alone.
How to Talk to Your Supervisor About a Null Result
Most supervisors have seen this a hundred times. Send them your output, your interpretation, and one line on what you plan to write in the discussion. This takes the panic out of the email and makes you look organised rather than stuck.
My Take After 12 Years of Reading Dissertations With Non-Significant Chapters
Most of the guidance floating around online falls into two extremes. One camp is purely reassuring, “it’s fine, don’t worry,” with zero actual SPSS screenshots, zero APA sentence templates, and nothing you can actually type into your document. The other camp is purely technical and textbook-style, correct but written for professors defending a point, not for a student who needs to submit in six hours.
My honest opinion, after sitting through more than a few viva defences on exactly this topic, is that a non significant result handled well in your write-up often reads as more mature research than a significant one handled carelessly. Examiners are not looking for perfect p-values. They want to see that you understand what your data is actually telling you.
If you want hands-on help interpreting non significant SPSS results before you submit, that’s exactly the kind of review I offer through SPSS tutoring support. A second pair of eyes on your output and your write-up usually catches small wording issues before your supervisor does.
If you’re further along and unsure if you need deeper support, it’s worth checking if a dissertation expert fits where you are right now. These templates are often enough on their own, so that check is just to be sure. And if your non-significant result came from a regression model specifically, double-check your regression diagnostics before you write anything up. A violated assumption can sometimes be the real reason behind a weak result.
Frequently Asked Questions
Can I accept the null hypothesis if not significant?
No. You fail to reject the null hypothesis. Accepting it implies you’ve proven no effect exists, which a non-significant result cannot do on its own.
How do you know whether to accept or reject the null hypothesis?
You reject the null hypothesis when your p-value falls below your alpha level, usually .05. You fail to reject it when the p-value is above that threshold, but either way, you are never technically “accepting” anything, only deciding whether your evidence was strong enough to reject.
Why don’t we ever accept the null hypothesis?
Because a non-significant result only shows an absence of strong evidence against the null, not proof that the null is true. Researchers almost always try to reject the null rather than accept it, since accepting it would require ruling out every possible real effect, which no single study can do.
Will a non-significant result fail my dissertation?
No, as long as you discuss it properly with reasoning tied to your methodology and literature. Examiners fail weak discussions, not non-significant numbers.
How do I report a p-value greater than 0.05 in APA format?
Report the exact value to two or three decimal places, for example p = .342, along with your test statistic and degrees of freedom, rather than writing “not significant” alone.
Is “marginally significant” an acceptable term to use?
Most examiners and journals discourage it. Report the exact p-value and let the number speak instead of softening it with vague language.
Does a non-significant result mean there’s no effect at all?
Not necessarily. It means your sample didn’t provide strong enough evidence to detect one, which can happen even when a real, smaller effect exists.
Should I rerun my analysis if the first test isn’t significant?
Only if you find a genuine error in your data or test selection. Rerunning repeatedly until something turns significant is considered poor research practice.
Written by Siddharth Gupta, 12+ years in applied statistics and dissertation consulting (R, Python, SPSS, Stata). Connect on LinkedIn.