Scatter Diagrams and Correlation Analysis
What is covered in this Article
Scatter diagrams and correlation analysis provide essential bivariate evaluation techniques for quality engineering and process optimization. Within the Measure and Analyze Phases of the quality management framework, understanding whether an independent input variable directly impacts a dependent process output is critical. By plotting paired numerical observations on a 2D Cartesian grid, quality professionals can visually inspect relationships, calculate mathematical correlation coefficients, and isolate assignable causes before implementing corrective actions.
Fundamentals of Bivariate Data Plotting
A Scatter Diagram plots paired numerical data () with the independent input variable (
) on the horizontal axis and the dependent output response (
) on the vertical axis. Unlike univariate displays such as histograms or check sheets, scatter diagrams reveal directional associations, curvature, scatter dispersion, and potential outliers between two process parameters.
| Scatter Pattern Type | Operational Interpretation |
|---|---|
| Strong Positive Linear | As input |
| Strong Negative Linear | As input |
| No Correlation (Cloud) | Points form a dispersed cloud; variables are independent. |
| Non-Linear Curve | Variables exhibit curvilinear or polynomial dependency. |

Mathematical Mechanics of Pearson Correlation
To quantify the strength and direction of linear association between two variables, quality engineers calculate Pearson’s correlation coefficient (). The mathematical formula is defined as:
Where:
= Individual paired data observations
= Arithmetic means of the respective variables
= Total sample size of paired observations
The resulting coefficient ranges strictly from
(perfect negative linear correlation) to
(perfect positive linear correlation), with
indicating zero linear association.
Worked Numerical Example: Curing Temperature versus Bond Strength
A quality team analyzes 5 paired observations of curing temperature ( in °C) and adhesive bond strength (
in MPa): (
), (
), (
), (
), (
).
| Temp |
Strength |
Product | ||
|---|---|---|---|---|
| 100 | 12 | -20 | -4 | 80 |
| 110 | 14 | -10 | -2 | 20 |
| 120 | 15 | 0 | -1 | 0 |
| 130 | 18 | 10 | 2 | 20 |
| 140 | 21 | 20 | 5 | 100 |

Elaborate Interpretation of Scatter Diagrams
Interpreting scatter plots requires moving beyond visual inspection to evaluate operational reality:
- Correlation Versus Causation: A high correlation coefficient (
) proves statistical association but does not automatically confirm physical causation. Controlled experimentation is required to verify operational mechanisms.
- Identifying Outliers: Isolated data points falling far outside the main cluster often indicate measurement errors, transcription mistakes, or special assignable operational conditions.
- Stratified Scatter Analysis: Grouping scatter points by operator, machine, or raw material batch frequently resolves ambiguous clouds into distinct, highly correlated sub-trends.
Graphical Best Practices for Quality Engineering
- Independent Axis Placement: Always place the causal input variable (
) on the horizontal axis and the output response (
) on the vertical axis.
- Zero Baseline Avoidance: Unlike histograms or bar charts, scatter axes do not strictly require zero origins if data ranges occupy a narrow numerical band, provided scales are clearly labeled.
- Regression Trendline Inclusion: Overlay a linear regression trendline and coefficient of determination (
) to summarize bivariate relationships effectively.
Transitioning into Process Control and Improvement
Mastering scatter diagrams and correlation analysis equips quality teams to transition seamlessly from baseline measurement into advanced regression modeling and process control planning.
This content aligns with professional quality engineering body-of-knowledge standards.
Written by Ravi Prakash—Quality Expert (38+ yrs exp). Connect on LinkedIn or Contact Us.
Frequently Asked Questions (FAQ)
Q1: What is the primary operational purpose of a Scatter Diagram in quality engineering?
A Scatter Diagram visually displays paired numerical variables to evaluate whether a directional relationship or correlation exists between an independent input and a dependent process output.
Q2: Does a high correlation coefficient () prove physical causation between variables?
No. Correlation proves statistical association, but physical causation must be validated through engineering experimentation, domain knowledge, or Design of Experiments (DOE).
Q3: How is the Pearson correlation coefficient interpreted?
The coefficient ranges from
to
. Values near
indicate strong linear association, while values near
indicate little to no linear relationship.
Q4: What should an engineer do when a scatter plot displays an outlier point?
The outlier should be investigated to determine if it stems from a measurement error, data entry mistake, or a significant assignable operational event before deciding to exclude it.
Six Sigma Practice Exam Questions
- A quality engineer plots machine spindle speed against part surface roughness on a scatter diagram and calculates a Pearson correlation coefficient of
. What does this indicate?
- A) No relationship exists between speed and roughness.
- B) A strong negative linear relationship exists, meaning higher speed corresponds to lower roughness.
- C) Speed causes surface roughness to increase exponentially.
- D) The measurement system is unstable.
Correct Answer: B) A strong negative linear relationship exists, meaning higher speed corresponds to lower roughness.
Explanation: Anvalue close to
signifies a strong inverse linear correlation.
- When constructing a scatter diagram to evaluate process variables, where should the independent input variable (
) be plotted?
- A) On the vertical Y-axis
- B) On the horizontal X-axis
- C) Along the diagonal reference line
- D) It does not matter which axis is chosen
Correct Answer: B) On the horizontal X-axis
Explanation: Standard graphical convention requires plotting the independent causal variable on the horizontal X-axis. - What fundamental distinction must quality engineers maintain when interpreting high correlation coefficients in scatter analysis?
- A) Correlation implies causation only in manufacturing environments.
- B) Correlation proves statistical association, whereas causation requires physical verification.
- C) Causation can be calculated without plotting scatter points.
- D) Correlation coefficients greater than
eliminate the need for control charts.
Correct Answer: B) Correlation proves statistical association, whereas causation requires physical verification.
Explanation: Statistical correlation does not automatically equate to physical cause-and-effect mechanisms. - If a scatter plot of paired observations forms a completely random circular cloud with a calculated
, what conclusion should be drawn?
- A) A strong positive linear relationship is present.
- B) There is no linear relationship between the two variables.
- C) The data must be stratified by machine shift immediately.
- D) The sample size
is too large.
Correct Answer: B) There is no linear relationship between the two variables.
Explanation: Anvalue near zero combined with a dispersed cloud indicates no linear association.
- Which statistical metric represents the square of the Pearson correlation coefficient and measures the proportion of variance explained by a linear model?
- A) Standard deviation
- B) Process capability
- C) Coefficient of determination (
)
- D) Student’s t-statistic
Correct Answer: C) Coefficient of determination (
)
Explanation: Squaring Pearson’syields
, quantifying the proportion of variance in the dependent variable predictable from the independent variable.
- A) Standard deviation
Continue Following the Masterclass
If you would like to follow this masterclass series on quality engineering and process statistics, look at the
Previous Post: 1027. Cause and Effect Diagrams and Root Cause Analysis for Quality Engineering and the
Next Post: 1029. Process Flowcharts and Stratification Methods for Quality Engineering Excellence
Support Our Website
Your support helps us continue delivering good content and useful data. If you’d like to contribute, here’s how:
- International Contributions (PayPal): https://www.paypal.com/paypalme/rpbehara – India Contributions (UPI ID): speakingdata@ybl
- Take a look at our special picks at Amazon – https://link.amazon/B03eHjgbl
- Subscribe to our Youtube channel Excellence and Eminence LLP: https://www.youtube.com/@ExcellenceandEminenceLLP
*As an Amazon Associate, I earn from qualifying purchases.