A scatter diagram plots paired observations on x and y axes. Scales, units, horizon, missing data, and potential third variables belong in interpretation; correlation does not establish causality.

1. 1

   Define variables

   Define variables

   

   Goal

   Produce variable contract so the core question “What pattern exists between X and Y, and which observations or groups need explanation?” can be answered with stronger support.

   Inputs for this step

   • paired quantitative data
   • definitions and units
   • segment and quality attributes

   Actions

   • Specify X, Y, unit, population, and horizon.
   • Record source, rationale, and open uncertainty directly in variable contract.

   Output

   variable contract

   variable contract · source · uncertainty · next decision

   Quality check

   Are observations correctly paired?

   Pitfall

   An aggregate trend hides segments or correlation is reported as causation. Mark segments separately, inspect raw data, disclose scaling, and test causal hypotheses separately.

   Timing: 10–20 minutes

2. 2

   Check data

   Check data

   

   Goal

   Produce clean dataset so the core question “What pattern exists between X and Y, and which observations or groups need explanation?” can be answered with stronger support.

   Inputs for this step

   • variable contract
   • definitions and units
   • segment and quality attributes

   Actions

   • Check missing data, duplicates, measurement errors, and range.
   • Record source, rationale, and open uncertainty directly in clean dataset.

   Output

   clean dataset

   clean dataset · source · uncertainty · next decision

   Quality check

   Are exclusions justified and counted?

   Pitfall

   An aggregate trend hides segments or correlation is reported as causation. Mark segments separately, inspect raw data, disclose scaling, and test causal hypotheses separately.

   Timing: 10–20 minutes

3. 3

   Diagram

   Diagram

   

   Goal

   Produce scatter diagram so the core question “What pattern exists between X and Y, and which observations or groups need explanation?” can be answered with stronger support.

   Inputs for this step

   • clean dataset
   • definitions and units
   • segment and quality attributes

   Actions

   • Plot points with suitable scale and relevant segment encoding.
   • Record source, rationale, and open uncertainty directly in scatter diagram.

   Output

   scatter diagram

   scatter diagram · source · uncertainty · next decision

   Quality check

   Are overlap and groups visible?

   Pitfall

   An aggregate trend hides segments or correlation is reported as causation. Mark segments separately, inspect raw data, disclose scaling, and test causal hypotheses separately.

   Timing: 10–20 minutes

4. 4

   Interpret

   Interpret

   

   Goal

   Produce analysis note so the core question “What pattern exists between X and Y, and which observations or groups need explanation?” can be answered with stronger support.

   Inputs for this step

   • scatter diagram
   • definitions and units
   • segment and quality attributes

   Actions

   • Describe form, direction, strength, outliers, and alternatives.
   • Record source, rationale, and open uncertainty directly in analysis note.

   Output

   analysis note

   analysis note · source · uncertainty · next decision

   Quality check

   Does the statement avoid unsupported causality?

   Pitfall

   An aggregate trend hides segments or correlation is reported as causation. Mark segments separately, inspect raw data, disclose scaling, and test causal hypotheses separately.

   Timing: 10–20 minutes