Scott Rule: How to Choose Histogram Bin Width

Data visualization plays a critical role in Lean Six Sigma analysis. Teams often start their statistical exploration by plotting a histogram. This simple chart helps engineers, analysts, and process leaders see how data spreads across a range of values. However, the histogram only works well when the bin width is chosen correctly.

If the bins are too wide, the chart hides useful variation. If they are too narrow, the histogram becomes noisy and misleading. Therefore, selecting the right bin width becomes essential.

The Scott Rule provides a reliable statistical method for determining histogram bin width. Many Six Sigma practitioners use this rule when they analyze continuous data sets such as cycle time, defect measurements, yield variation, or process temperatures.

This article explains the Scott Rule in detail. You will learn how the rule works, when to apply it, how to calculate it, and how it compares with other common histogram rules used in Six Sigma.

What Is the Scott Rule?

The Scott Rule is a statistical formula used to determine the optimal bin width for a histogram. The rule was introduced by statistician David W. Scott in 1979 as a method for minimizing information loss when visualizing continuous/variable data.

Instead of guessing how many bins to use, the Scott Rule calculates histogram bin width based on:

• The standard deviation of the data
• The number of observations in the dataset

Because it uses statistical properties of the dataset, the rule often produces smoother and more meaningful histograms.

Scott Rule Formula

The Scott Rule calculates bin width using the following equation:

Variable	Meaning
h	Histogram bin width
σ	Standard deviation of the dataset
n	Number of observations

Formula:

Bin Width (h) = 3.5 × σ ÷ n^(1/3)

After calculating the bin width, analysts divide the full data range by the bin width to determine the number of histogram bins.

Why Histograms Matter in Six Sigma

Six Sigma relies heavily on data. Teams constantly measure process performance and analyze variation. Because of this, visual tools become essential.

Histograms provide a quick snapshot of how a process behaves.

They help teams answer questions such as:

• Is the process centered?
• Does the distribution look normal?
• Are multiple process modes present?
• Do outliers exist?

However, poor bin selection can distort the picture.

For example:

Bin Choice	Result
Too few bins	Important patterns disappear
Too many bins	Random noise appears as structure
Proper bin width	True process behavior becomes visible

The Scott Rule helps avoid those problems.

Consequently, Six Sigma practitioners often apply it during the Measure and Analyze phases of DMAIC.

Understanding Histogram Bins

Before exploring the Scott Rule further, it helps to understand histogram bins.

A bin represents a range of values within the dataset. The histogram counts how many observations fall into each bin.

Consider this small dataset representing process cycle time (seconds):

45, 47, 46, 49, 48, 50, 52, 51, 47, 46

If we choose bins of width 5 seconds, we might get:

Bin Range	Count
[45–50]	8
(50–55]	2

Alternatively, if we choose bins of width 2 seconds, we get:

Bin Range	Count
[45–47]	5
(47–49]	2
(49–51]	2
(51–53]	1

The second version provides more detail. However, extremely small bins would introduce noise.

The Scott Rule helps balance detail and clarity.

The Statistical Idea Behind the Scott Rule

The Scott Rule comes from density estimation theory. The goal involves estimating the underlying probability distribution of the data.

Histograms approximate this distribution.

However, bin width strongly affects the quality of that estimate.

Scott developed the rule by minimizing the Mean Integrated Squared Error (MISE) between the histogram and the true distribution.

Although the math behind MISE becomes complex, the practical outcome remains simple.

The rule balances two competing forces:

Factor	Effect
Narrow bins	Capture more detail but increase noise
Wide bins	Reduce noise but hide structure

The Scott Rule calculates a bin width that balances both effects.

Step-by-Step Example of the Scott Rule

Let’s walk through a practical Six Sigma example.

Imagine a manufacturing process that produces metal components. Engineers measure the diameter of 200 parts.

Suppose the dataset has the following characteristics:

Metric	Value
Sample size (n)	200
Standard deviation (σ)	0.018 mm
Minimum value	9.94 mm
Maximum value	10.06 mm

Step 1: Calculate Cube Root of Sample Size

First compute:

n^(1/3)

Cube root of 200 ≈ 5.85

---

Step 2: Apply the Scott Rule Formula

h = 3.5 × σ ÷ n^(1/3)

= 3.5 × 0.018 ÷ 5.85

≈ 0.0108 mm

This result represents the recommended bin width.

---

Step 3: Calculate Data Range

Range = Max − Min

= 10.06 − 9.94

= 0.12 mm

---

Step 4: Calculate Number of Bins

Number of bins = Range ÷ Bin width

0.12 ÷ 0.0108 ≈ 11 bins

Therefore, the histogram should contain about 11 bins.

This structure should produce a balanced view of the process variation.

Why the Scott Rule Works Well for Normal Data

The Scott Rule performs best when the dataset follows a normal distribution.

Many manufacturing processes naturally produce normal distributions because variation often results from many small independent factors.

Examples include:

• Injection molding part weight
• Chemical concentration
• Cycle time
• Temperature variation

When data behaves normally, the standard deviation accurately describes spread.

Since the Scott Rule uses standard deviation, it performs well under these conditions.

However, skewed datasets may require different methods.

Scott Rule vs Other Histogram Rules

Several statistical rules help determine histogram bin sizes. Each rule uses different assumptions.

Six Sigma practitioners often compare these methods.

Below is a comparison of common approaches.

Rule	Formula Basis	Strengths	Weaknesses
Scott Rule	Standard deviation	Smooth distributions	Sensitive to outliers
Freedman-Diaconis Rule	Interquartile range	Handles skewed data	Produces more bins
Sturges’ Rule	Logarithmic sample size	Simple calculation	Underestimates bins for large datasets
Rice Rule	Sample size only	Easy to apply	Ignores data spread

Quick Comparison Example

Suppose a dataset contains 500 observations.

Rule	Approximate Bins
Sturges	10
Rice	16
Scott	18
Freedman–Diaconis	22

The Scott Rule typically falls in the middle.

This balance explains its popularity in statistical analysis.

When Six Sigma Practitioners Should Use the Scott Rule

The Scott Rule works best under certain conditions.

Six Sigma teams should apply it when the dataset meets these criteria:

Large Data Sets

The rule performs better with moderate or large sample sizes.

Small datasets may produce unstable standard deviation estimates.

Continuous Data

The rule works for measurements such as:

Example Metric	Industry
Part thickness	Manufacturing
Process temperature	Chemical processing
Cycle time	Operations
Fill volume	Pharmaceuticals

Discrete/attribute data does not benefit from this method.

Approximately Normal Distributions

The rule assumes normality. Therefore, teams should confirm this assumption.

Common checks include:

• Normal probability plots
• Skewness values
• Histogram inspection

Example: Scott Rule in a Six Sigma Project

Imagine a Six Sigma team working in a battery materials manufacturing plant.

The team analyzes powder particle size because it affects downstream electrode performance.

The engineers collect 600 measurements from a particle analyzer.

Dataset Summary

Metric	Value
Sample size	600
Standard deviation	3.2 microns
Minimum	18 microns
Maximum	40 microns

Step 1: Cube Root of Sample Size

600^(1/3) ≈ 8.43

---

Step 2: Apply Scott Rule

h = 3.5 × 3.2 ÷ 8.43

h ≈ 1.33 microns

---

Step 3: Range

40 − 18 = 22 microns

---

Step 4: Number of Bins

22 ÷ 1.33 ≈ 17 bins

The team therefore plots a histogram using 17 bins.

Immediately, the histogram reveals two peaks.

That discovery suggests two particle populations. Engineers later trace the issue to inconsistent milling conditions.

Thus, the Scott Rule helps uncover a process issue that would otherwise remain hidden.

Advantages of the Scott Rule in Six Sigma

The Scott Rule offers several benefits for practitioners.

Objective Bin Selection

The rule removes subjective guesswork.

Analysts rely on a formula rather than intuition.

Balanced Histogram Detail

The rule balances smoothness and resolution.

Therefore, histograms avoid both oversmoothing and excessive noise.

Works Well With Large Data

Six Sigma projects often involve hundreds or thousands of data points.

The Scott Rule performs well under those conditions.

Easy to Implement

Most statistical software packages include the rule.

Programs such as:

• Minitab
• Python
• R
• MATLAB
• JMP

can compute the bin width instantly.

Limitations of the Scott Rule

Despite its strengths, the Scott Rule also has limitations.

Understanding these limitations helps teams avoid mistakes.

Sensitivity to Outliers

The rule uses standard deviation.

Outliers inflate standard deviation and therefore widen bins.

Wider bins may hide important structure.

Assumes Normal Distribution

The rule works best for normal data.

Highly skewed datasets may require the Freedman-Diaconis rule instead.

Large Bins for Small Samples

Small datasets often produce overly wide bins.

Consequently, visual detail decreases.

Example Comparing Scott Rule and Freedman-Diaconis Rule

Consider a dataset representing customer service response time.

The data shows a right skew because most tickets resolve quickly while a few take much longer.

Metric	Value
Sample size	350
Standard deviation	14 minutes
Interquartile range	10 minutes

Scott Rule

h = 3.5 × σ ÷ n^(1/3)

Result ≈ 6.5 minutes

---

Freedman–Diaconis Rule

h = 2 × IQR ÷ n^(1/3)

Result ≈ 3.2 minutes

---

Outcome

Rule	Histogram Effect
Scott	Smooth but hides skew detail
Freedman–Diaconis	More bins reveal tail behavior

Therefore, the second rule works better for skewed distributions.

Scott Rule in DMAIC Projects

Six Sigma teams often apply the Scott Rule during the Measure and Analyze phases of DMAIC.

Measure Phase

During Measure, teams collect process data.

Histograms help verify baseline performance.

The Scott Rule ensures the visualization reflects actual variation.

Analyze Phase

During Analyze, teams search for root causes.

Histograms may reveal:

• Multiple process modes
• Process drift
• Outliers
• Distribution shape

Proper bin width makes these patterns easier to detect.

Using the Scott Rule in Statistical Software

Many tools automate the Scott Rule calculation.

Below are examples.

Python Example

Data scientists often use NumPy or Matplotlib.

Example code:

import numpy as np
import matplotlib.pyplot as pltdata = np.random.normal(50, 5, 500)plt.hist(data, bins='scott')
plt.show()

The program automatically calculates the Scott bin width.

R Example

R also supports the Scott Rule.

hist(data, breaks="scott")

The function calculates the optimal bins automatically.

Minitab

Minitab typically selects bins automatically. However, users can adjust bin width manually if necessary.

Many analysts compute the Scott bin width externally before building the histogram.

Practical Tips for Six Sigma Practitioners

Although the Scott Rule provides a strong starting point, practitioners should still use judgment.

Always Inspect the Histogram

After plotting the histogram, check whether the distribution makes sense.

If the chart looks overly smooth or noisy, adjust bin width slightly.

Remove Extreme Outliers

Outliers distort standard deviation.

Therefore, investigate extreme values before applying the rule.

Compare Multiple Methods

Many analysts compare results from:

• Scott Rule
• Freedman-Diaconis Rule
• Sturges’ Rule
• Rice Rule
• Square Root Rule

The best histogram usually appears quickly.

Consider Process Knowledge

Statistical rules cannot replace engineering judgment.

Process knowledge often explains unusual patterns.

Real Manufacturing Example

Consider a packaging line that fills bottles with liquid detergent.

Quality engineers measure fill weight to ensure compliance.

Data Summary

Metric	Value
Sample size	400
Standard deviation	2.4 grams
Minimum	497 g
Maximum	508 g

Scott Rule Calculation

Cube root of 400 ≈ 7.37

h = 3.5 × 2.4 ÷ 7.37

h ≈ 1.14 grams

Range

508 − 497 = 11 grams

Number of Bins

11 ÷ 1.14 ≈ 10 bins

The resulting histogram reveals a slight left skew.

Further investigation shows that a filling valve occasionally underfills during startup.

Without the correct bin width, the skew might remain hidden.

Common Mistakes When Using the Scott Rule

Even experienced analysts sometimes misuse histogram rules.

Below are common mistakes.

Using Too Few Data Points

Datasets with fewer than 30 observations produce unreliable bin widths.

Small samples rarely produce stable standard deviation estimates.

Ignoring Distribution Shape

Blindly applying the Scott Rule can hide skew or heavy tails.

Always examine data shape first.

Forgetting Units

Bin width carries the same units as the data.

For example:

Metric	Bin Width Unit
Temperature	Degrees
Cycle time	Seconds
Diameter	Millimeters

Incorrect units create confusing histograms.

How Scott Rule Supports Data-Driven Decision Making

Six Sigma focuses on evidence-based decisions.

Visualization plays a central role in that philosophy.

A well-constructed histogram helps teams:

• Detect process instability
• Identify variation sources
• Validate improvements
• Communicate results clearly

The Scott Rule strengthens these analyses by improving visualization accuracy.

Better visualizations lead to better insights.

Better insights lead to stronger process improvements.

Conclusion

The Scott Rule provides a statistically grounded method for selecting histogram bin width. Six Sigma practitioners often rely on this rule when they analyze continuous process data.

Unlike guesswork approaches, the Scott Rule uses standard deviation and sample size to determine bin width. As a result, the method balances detail and smoothness in the histogram.

However, analysts should remember that the rule assumes a roughly normal distribution. Highly skewed datasets may require alternative methods such as the Freedman-Diaconis rule.

Despite these limitations, the Scott Rule remains one of the most useful histogram binning techniques in data analysis.

When applied correctly, it helps Six Sigma teams visualize variation more clearly, identify hidden patterns, and make smarter data-driven decisions.