Skip to content
LeanSuite - Lean Manufacturing Software

Free template

Updated

Scatter diagram template

A scatter diagram plots pairs of measurements, one on each axis, to show whether one changes with the other: for example feed rate against surface roughness. The correlation coefficient r, from −1 to +1, puts a number on how closely the dots follow a straight line, and the least squares line helps read the trend. A pattern shows that two things move together, not that one causes the other. It is one of the seven basic quality tools. Page 1 of the PDF has the header fields, a table for 50 pairs with a group code, a plot grid and a box for the quadrant (median) count. Page 2 shows five typical patterns, a rough guide to reading r, the formulas for r, the slope and the intercept, the quadrant method, the warnings about cause and stratification, and a worked example plotted to scale. The Excel version takes up to 50 pairs, flags and leaves out rows without both values, and works out n, the sums, r, R², the slope and intercept of the least squares line, a rough strength label, the medians and the quadrant count, and r, slope and intercept for up to 4 groups. It plots each group with its own marker, draws the line and has a worked example sheet.

Free to use: print it, copy it and edit it for your team. Enter your name and work email once to download.

Page 1 of the scatter diagram template: fields for x, y, process, dates and group codes, a numbered table for 50 pairs of x, y and group, a plot grid with axes for x and y, and a quadrant count box for the medians and the dots in each quarter.
50 pairs with group codesPlot gridQuadrant (median) countCorrelation r and R²Least squares liner by groupWorked example

When to use it

When to use a scatter diagram

  • To test a suspected cause from a fishbone or a 5 Whys: whether the defect or the measurement changes when the input changes.
  • To find the setting range that keeps an output in specification, such as a temperature, pressure, speed or feed.
  • To check whether a quick test can stand in for a slow one: a hardness reading against a tensile test, an in-process gauge against the lab.
  • When data from several machines, shifts or lots is mixed, to see whether splitting it by group changes the picture.

How to fill it in

  1. 1

    Choose the pair

    x is the suspected cause or input, y the effect or output. Each pair comes from the same part, batch or moment.

  2. 2

    Collect the pairs

    Aim for 30 or more, over the widest range of x you can safely run, with a group code for the machine, shift or lot.

  3. 3

    Plot

    x across, y up, with scales that fill the grid. One dot per pair, and a different mark for each group.

  4. 4

    Count the quarters

    Draw lines at the median x and the median y. Count the dots upper right plus lower left against upper left plus lower right, leaving out dots on a line.

  5. 5

    Work out r and the line

    From the sums on page 2, or let the Excel version do it. Read the line only inside the range of x you tested.

  6. 6

    Split by group, then confirm

    Work out r for each group. Confirm a suspected cause with a trial that changes only x before you change the process.

Reading r

What the correlation coefficient r says

A rough guide only: cut-offs vary by field, a curved pattern can give r near 0, and a few pairs can give a high r by chance. Always look at the plot.

  • 0.7 to 1

    Reads as
    Strong positive: y rises as x rises
    What to do
    Likely related: confirm the cause with a trial
  • 0.3 to 0.7

    Reads as
    Moderate positive
    What to do
    Related, with other factors at work: stratify and look for them
  • −0.3 to 0.3

    Reads as
    Little or no straight-line relationship
    What to do
    Check the plot for a curve or for groups pulling apart
  • −0.7 to −0.3

    Reads as
    Moderate negative
    What to do
    Related, with other factors at work: stratify and look for them
  • −1 to −0.7

    Reads as
    Strong negative: y falls as x rises
    What to do
    Likely related: confirm the cause with a trial

A filled-in example

Illustrative, not a benchmark

An example: feed rate against surface roughness Ra on two lathes, 24 pairs, 12 from each (illustrative numbers). The Excel version's Worked example sheet has all 24 pairs.

  • All 24 pairs: r = 0.80, R² = 0.64, least squares line Ra = 1.072 + 6.00 × feed.
  • Quadrant count: medians 0.22 mm/rev and 2.455 µm. Upper right 9, lower left 9, upper left 2, lower right 2, and 2 dots on the median line left out: 18 against 4, a positive relationship.
  • Split by lathe: r = 0.96 for lathe A (Ra = 0.373 + 7.66 × feed) and 0.97 for lathe B (Ra = 0.993 + 7.91 × feed).
  • Over the feeds both lathes ran, 0.14 to 0.30 mm/rev, lathe B leaves a surface 0.65 to 0.69 µm rougher at the same feed.

Mixing the lathes made the relationship look weaker than it is and hid a second cause: the machine. Within each lathe, roughness follows feed closely, and lathe B went on the list for a check at the machine before anyone changed the feed.

Common mistakes

  • Reading correlation as cause

    Two things can move together because both follow a third, such as temperature or time of day. Confirm with a trial that changes only x.

  • Mixing groups

    Data from different machines, shifts or lots can hide a strong pattern or create a false one. Give each pair a group code and work out r per group.

  • Too narrow a range of x

    If x hardly varies, no relationship can show, however strong it is. Run x over the widest range you can safely test.

  • Stretching the line

    The least squares line describes the range you tested. Outside it the relationship may bend or break.

Download the template

Free to use: print it, copy it and edit it for your team. Enter your name and work email once to download.

Run this template in LeanSuite

In LeanSuite defects are captured as they happen and traced to their root causes with 5 Whys, 4M1D and fishbone, and KPI Builder tracks quality KPIs against targets and shows where defects happen.

FAQ

Scatter diagram template: common questions

More free templates

All templates

Can't find the template you need?

Tell us what you're looking for. We'll make it and send it to your inbox, free.

A shift handover sheet, a red tag log, a changeover checklist for one machine: if your team uses it on the floor, ask.

By sending this form you agree to our Terms of Service and Privacy Policy.