What Is an Interval Plot and How Do You Use It in Quality Analysis?

If you have ever needed to compare group means at a glance—without drowning in boxplots or hypothesis-test output—the interval plot is one of the simplest, most visual tools in your quality toolkit. It shows the mean of each group with a confidence interval around it, letting you see at once whether groups are statistically different or just randomly noisy.

What It Is

An interval plot (also called an interval plot of means) is a statistical chart that displays the mean of each group as a point, with a vertical line (or "whisker") extending above and below it to represent the confidence interval (CI) of that mean. In the Minitab convention, the default is typically a 95% confidence interval for the mean, but you can choose other confidence levels depending on your analysis.

The key purpose is visual comparison: if two intervals overlap substantially, the means are likely not statistically different. If they do not overlap, that is strong visual evidence of a real difference.

How It Works / Formula or Steps

The interval plot is built from three simple quantities for each group:

1. Mean (\(\bar{x}\)): the average of the group's measurements.
2. Standard error of the mean: \(SE = \frac{s}{\sqrt{n}}\), where \(s\) is the sample standard deviation and \(n\) is the sample size.
3. Confidence interval: \(\bar{x} \pm t^ \cdot SE\), where \(t^\) is the critical value from the t-distribution for your chosen confidence level and degrees of freedom (\(n-1\)). For a 95% confidence interval with a large sample, \(t^\) is approximately 1.96 (the well-known z-value), but for small samples you should use the exact t-value.

Steps to create an interval plot:

1. Collect your data in groups (e.g., batches, machines, shifts, suppliers).
2. Calculate the mean, standard deviation, and sample size for each group.
3. Choose a confidence level (commonly 95%).
4. Compute the lower and upper bounds: \(\bar{x} \pm t^ \cdot \frac{s}{\sqrt{n}}\).
5. Plot each mean as a point and draw the interval as a vertical line through the point.

A Worked Illustrative Example

Example data (illustrative only): Suppose you are comparing the tensile strength of plastic parts from three injection-molding machines. You sample 10 parts from each machine and record the following summary statistics:

Machine | Mean (MPa) | Std Dev (MPa) | n

• A | 52.0 | 3.0 | 10
• B | 55.0 | 3.5 | 10
• C | 53.5 | 2.8 | 10

For a 95% confidence interval with \(n-1 = 9\) degrees of freedom, the t-critical value is \(t^ \approx 2.262\).

• Machine A: \(SE = 3.0/\sqrt{10} = 0.949\). CI = \(52.0 \pm 2.262 \times 0.949\) = \(52.0 \pm 2.15\), i.e., [49.85, 54.15].
• Machine B: \(SE = 3.5/\sqrt{10} = 1.107\). CI = \(55.0 \pm 2.262 \times 1.107\) = \(55.0 \pm 2.50\), i.e., [52.50, 57.50].
• Machine C: \(SE = 2.8/\sqrt{10} = 0.885\). CI = \(53.5 \pm 2.262 \times 0.885\) = \(53.5 \pm 2.00\), i.e., [51.50, 55.50].

Interpretation: The interval for Machine A [49.85, 54.15] overlaps with Machine C [51.50, 55.50], but it does not* overlap with Machine B [52.50, 57.50]. This visual result suggests Machine A is statistically different from Machine B, while Machine C is not clearly different from either. (For a formal conclusion, you would still run an ANOVA or t-test.)

Common Pitfalls

• Do not confuse interval plots with control charts. A control chart tracks a process over time against control limits; an interval plot compares group means at a single point in time.
• Overlapping intervals do not always mean "no difference." For a formal test, overlap of 95% CIs is a conservative check—sometimes means can be significantly different even with slight overlap. Use hypothesis tests for definitive answers.
• Watch the sample size. With very small \(n\), the t-value makes intervals wide, and the plot may hide real differences. Always report \(n\) alongside the plot.
• Do not use the raw data range (min–max) instead of the confidence interval—that would show the spread of individual values, not the precision of the mean.

---

An interval plot is a fast, honest way to communicate uncertainty in your means to non-statisticians and managers. To create one instantly from your own data, try the free interval plot tool at https://www.6sq.com/tools/interval/ — just paste your grouped data and the chart is ready in seconds.