Statistics fundamentals

Confidence Interval Explained: What It Means and How to Report It in APA

Q: Does a wider confidence interval mean my result is not significant?

Not necessarily. A wide CI means low precision, but significance depends on whether the CI includes the null value (zero for differences, one for odds ratios). A wide CI that still excludes zero is both significant and imprecise - which means you have found an effect but cannot estimate its magnitude precisely.

Q: Should I report 95% or 99% confidence intervals in my thesis?

95% is the standard for most social science research. Use 99% only if you have a specific reason (e.g., your field convention or a very large sample where the extra precision matters). Be consistent - do not mix 95% and 99% CIs in the same analysis without explanation.

Q: Can a confidence interval be just a single number (not a range)?

No. A CI is always an interval, not a point. If your CI appears to collapse to a point, it is a display rounding issue. A CI can theoretically be very narrow (with very large samples and low variability), but it always has width.

Q: Why does SPSS show a negative lower bound in my CI?

A negative lower bound for a mean difference CI is completely normal - it means zero (or no difference) is within the plausible range, which corresponds to a non-significant result. For example, a CI of [−3.2, 8.5] for a group difference includes zero, so you cannot conclude the groups truly differ.

Q: Does a non-overlapping confidence interval always mean a significant difference?

Not necessarily. When CIs for two group means do not overlap, the difference is almost certainly significant at p < .05. But overlapping CIs do not automatically mean non-significance - the correct test is whether the CI for the difference between means includes zero. Always test the difference directly rather than visually comparing individual CIs.

Q: What is the difference between a confidence interval and a prediction interval?

A confidence interval estimates where the true population mean (or parameter) lies. A prediction interval estimates where a single new observation from the same population will fall. Prediction intervals are always wider than confidence intervals because individual observations are more variable than means. In thesis work, CIs are almost always what you need to report.

April 20266 min read

Almost every published study reports confidence intervals - and almost every student misinterprets them. Saying 'there is a 95% probability the true value lies within this interval' is technically wrong and will lose you marks in your thesis defense. Here is the correct interpretation of a 95% confidence interval, how it relates to your p-value, and the exact APA format for reporting CIs across t-tests, ANOVA, regression, and correlation.

Free sample chapter

Data Analysis From Survey to Results

Step-by-step guidance for choosing the right test, running it, and writing up APA results - in plain language, not theory. Get the free sample chapter when you join the waitlist.

Statistical Analysis Step by Step book cover

What a 95% Confidence Interval Actually Means

A 95% confidence interval (CI) does NOT mean "there is a 95% probability that the true value lies within this interval." The true population value is fixed - it either is or is not within the interval. Probability does not apply to a single computed interval.

The correct interpretation: if you repeated your study 100 times and computed a 95% CI each time, approximately 95 of those intervals would contain the true population value.

In practice, for your thesis, report the CI alongside your estimate and interpret it as the range of plausible values for the population parameter, given your data.

How Confidence Intervals Relate to P-Values in Your Thesis

A 95% CI and a p < .05 significance test are mathematically linked.

If a 95% CI for a mean difference does not include zero, the corresponding t-test will be significant at p < .05.

If the CI includes zero (i.e., zero is a plausible value for the true difference), the test is not significant at p < .05.

This is why confidence intervals are more informative than p-values alone: they show you both whether an effect is significant AND how large (or small) it plausibly is.

What Makes a Confidence Interval Wider or Narrower?

Three things affect interval width:

1. Sample size: larger samples produce narrower intervals. Doubling your sample size roughly halves the margin of error.

2. Confidence level: a 99% CI is wider than a 95% CI (more certainty requires a wider net). A 90% CI is narrower.

3. Variability in your data: higher standard deviation = wider CI. If your data is very spread out, your estimate is less precise.

A wide CI means your estimate is imprecise - your data could be consistent with a range of true values. A narrow CI means your estimate is precise.

Confidence Intervals for Different Statistics in Your Thesis

CIs can be computed for almost any estimate:

Mean: [M − (t* × SE), M + (t* × SE)] - most common in thesis work

Mean difference (t-test): CI for the difference between two group means. If this CI excludes zero, the difference is significant.

Regression coefficient (β): CI shows the range of plausible slope values. A CI excluding zero means the predictor is a significant predictor.

Correlation (r): Fisher's z-transformation is used to compute CI for Pearson r. Wide CI around a correlation suggests low precision - usually due to small sample size.

Odds ratio (logistic regression): CI on the log scale, exponentiated back. A CI excluding 1.0 indicates a significant association.

How to Read Confidence Interval Output in SPSS

SPSS labels CI columns as "Lower Bound" and "Upper Bound" (or "95% CI Lower" and "95% CI Upper" in newer versions).

For an independent t-test: the CI is for the mean difference between groups. If [Lower Bound, Upper Bound] does not include 0, the groups differ significantly.

For linear regression: each coefficient has its own CI. A coefficient where Lower Bound and Upper Bound have the same sign (both positive or both negative) is significant.

For ANOVA post-hoc (Tukey): SPSS shows CIs for each pairwise mean difference. CIs not spanning zero indicate significant pairs.

How to Report Confidence Intervals in APA Format

APA 7th edition requires reporting CIs for all major estimates.

Format: [LL, UL] (no spaces inside brackets, comma-space between bounds)

Examples:
Mean: M = 42.3, 95% CI [38.7, 45.9]
Mean difference: The groups differed by 6.4 points, 95% CI [2.1, 10.7]
Correlation: r(48) = .54, p = .012, 95% CI [.27, .73]
Regression coefficient: β = 0.32, 95% CI [0.11, 0.53], p = .004

Always state the confidence level (95% is standard; state if different). Do not write "CI = [38.7, 45.9]" without specifying the level.

Frequently asked questions

Does a wider confidence interval mean my result is not significant?

▾

Not necessarily. A wide CI means low precision, but significance depends on whether the CI includes the null value (zero for differences, one for odds ratios). A wide CI that still excludes zero is both significant and imprecise - which means you have found an effect but cannot estimate its magnitude precisely.

Should I report 95% or 99% confidence intervals in my thesis?

▾

95% is the standard for most social science research. Use 99% only if you have a specific reason (e.g., your field convention or a very large sample where the extra precision matters). Be consistent - do not mix 95% and 99% CIs in the same analysis without explanation.

Can a confidence interval be just a single number (not a range)?

▾

No. A CI is always an interval, not a point. If your CI appears to collapse to a point, it is a display rounding issue. A CI can theoretically be very narrow (with very large samples and low variability), but it always has width.

Why does SPSS show a negative lower bound in my CI?

▾

A negative lower bound for a mean difference CI is completely normal - it means zero (or no difference) is within the plausible range, which corresponds to a non-significant result. For example, a CI of [−3.2, 8.5] for a group difference includes zero, so you cannot conclude the groups truly differ.

Does a non-overlapping confidence interval always mean a significant difference?

▾

Not necessarily. When CIs for two group means do not overlap, the difference is almost certainly significant at p < .05. But overlapping CIs do not automatically mean non-significance - the correct test is whether the CI for the difference between means includes zero. Always test the difference directly rather than visually comparing individual CIs.

What is the difference between a confidence interval and a prediction interval?

▾

A confidence interval estimates where the true population mean (or parameter) lies. A prediction interval estimates where a single new observation from the same population will fall. Prediction intervals are always wider than confidence intervals because individual observations are more variable than means. In thesis work, CIs are almost always what you need to report.

Not sure which statistical test to use?

Answer 5 quick questions about your research design and get the right test - with an explanation of why - in under two minutes.

Statoria Team

Statistics educators & software developers

We build Statoria to help bachelor and master students get through their thesis data analysis without stress. Our guides are written by researchers with experience in social science statistics and student supervision.