Explore how one or two inputs change a formula while keeping possibilities, predictions and approved decisions clearly separate
Use a question–baseline–formula–variable-count–value-range–row-input–column-input–unit–calculation-state–pattern–threshold–assumption–decision-owner–release chain. Build the smallest table that answers the question, label it fully, test corner values and communicate sensitivity as conditional evidence.
Why this English matters now
Reading Excel’s Data Table results before presenting sensitivity analysis is practical English for students and planners searching how to create a one-variable or two-variable data table, run What-If Analysis or compare Excel outcomes.
Microsoft’s current guidance says a Data Table substitutes sets of values into one or two input cells and displays formula results in one worksheet table. A one-variable table can show one or more formulas; a two-variable table uses one formula and two input lists.
A grid of outcomes is not automatically a forecast, probability distribution or recommendation. Strong readers verify the formula, row and column input cells, units, value ranges, recalculation state and fixed assumptions before explaining what the table does—and does not—show.
Checked against the linked first-party sources on 30 October 2026. Interfaces, managed-device rules and school procedures can change, so reopen the current source before a live assessment, disclosure or account decision.
The practical reading method
Use a question–baseline–formula–variable-count–value-range–row-input–column-input–unit–calculation-state–pattern–threshold–assumption–decision-owner–release chain. Build the smallest table that answers the question, label it fully, test corner values and communicate sensitivity as conditional evidence.
- Microsoft describes Data Tables as What-If Analysis tools that show outcomes from sets of input values.
- A one-variable table changes one input and may display multiple formulas; a two-variable table changes two inputs for one formula.
- A Data Table cannot accommodate more than two variables, although each variable can have many test values.
- Data Tables recalculate with the worksheet unless calculation options are changed, so saved and displayed results depend on calculation state.
Official factual floor
- Microsoft Support: Calculate multiple results by using a data table
- Microsoft Support: Introduction to What-If Analysis
- Microsoft Support: Change formula recalculation, iteration or precision
- MOE English Language Syllabus 2020
- IMDA Digital Skills for Life
- Microsoft says Data Tables focus on one or two variables and show possible outcomes in tabular form.
- Microsoft says one-variable tables can evaluate multiple formulas sharing one input, while two-variable tables evaluate one formula across two input lists.
- Microsoft says Data Tables recalculate when the worksheet recalculates and documents an Automatic Except for Data Tables calculation option.
Did You Know?
A two-variable Data Table can contain many rows and columns, yet it still varies only two input cells. The table may look large and sophisticated while every other assumption remains fixed and invisible.
Full chapter index
- 1. Frame one sensitivity question
- 2. Preserve the baseline and calculation state
- 3. Verify the formula before multiplying results
- 4. Choose one-variable structure when one input changes
- 5. Choose two-variable structure for one formula
- 6. Assign row and column input cells correctly
- 7. Choose meaningful value ranges
- 8. Label units and fixed assumptions
- 9. Read patterns, not isolated attractive cells
- 10. Inspect thresholds and discontinuities
- 11. Control recalculation deliberately
- 12. Distinguish sensitivity from forecast and probability
- 13. Verify the published table and decision note
The mechanism
The table should answer a defined question about how inputs affect an output.
Microsoft describes Data Tables as What-If Analysis tools that show outcomes from sets of input values. In this chapter, the decisive phrase is frame one sensitivity question. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: A student creates fifty combinations before deciding what decision the results should inform.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. name the output
- 2. name the candidate inputs
- 3. state the decision owner
- 4. define the time horizon
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Question framing supports analysis. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
Data Tables can recalculate and change displayed results as inputs or formulas change.
A one-variable table changes one input and may display multiple formulas; a two-variable table changes two inputs for one formula. In this chapter, the decisive phrase is preserve the baseline and calculation state. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: The original assumptions are lost after new test values are pasted over them.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. save a controlled copy
- 2. record baseline values
- 3. record calculation mode
- 4. protect source data
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Version control supports reproducibility. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
Every cell in the table depends on the formula being correct.
A Data Table cannot accommodate more than two variables, although each variable can have many test values. In this chapter, the decisive phrase is verify the formula before multiplying results. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: A sign error is repeated across a polished twenty-by-twenty grid.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. inspect formula references
- 2. check units
- 3. calculate one case manually
- 4. trace key precedents
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Formula literacy supports evidence. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
A one-variable table tests many values for one cell and can show several formulas.
Data Tables recalculate with the worksheet unless calculation options are changed, so saved and displayed results depend on calculation state. In this chapter, the decisive phrase is choose one-variable structure when one input changes. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: Interest rate changes, but the table is described as varying both rate and term.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. name the single changing cell
- 2. choose row or column orientation
- 3. place related formulas correctly
- 4. hold other inputs explicit
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Scope language supports modelling. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
A two-variable table combines row and column value lists for one result formula.
Microsoft describes Data Tables as What-If Analysis tools that show outcomes from sets of input values. In this chapter, the decisive phrase is choose two-variable structure for one formula. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: Two outputs are placed in a two-variable layout and the grid becomes ambiguous.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. select one output formula
- 2. name both variables
- 3. place the row values
- 4. place the column values
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Matrix language supports comparison. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
The dialog maps each value list to a real model input cell.
A one-variable table changes one input and may display multiple formulas; a two-variable table changes two inputs for one formula. In this chapter, the decisive phrase is assign row and column input cells correctly. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: Rate values are sent to the term cell and months are sent to the percentage cell.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. read the table orientation
- 2. identify the row variable
- 3. identify the column variable
- 4. check cells and units
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Directional language supports Excel. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
Sensitivity depends on the tested domain and spacing.
A Data Table cannot accommodate more than two variables, although each variable can have many test values. In this chapter, the decisive phrase is choose meaningful value ranges. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: A budget table tests only tiny changes and hides the point where the plan fails.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. cite a range source
- 2. include the baseline
- 3. test plausible extremes
- 4. use finer steps near thresholds
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Range language supports planning. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
The grid alone does not show what every number means or what stayed constant.
Data Tables recalculate with the worksheet unless calculation options are changed, so saved and displayed results depend on calculation state. In this chapter, the decisive phrase is label units and fixed assumptions. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: A result of 250 could mean dollars, students, hours or percent.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. put units in headings
- 2. write fixed assumptions nearby
- 3. date external inputs
- 4. define rounding
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Measurement language supports communication. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
The surface can reveal direction, curvature, interaction and thresholds.
Microsoft describes Data Tables as What-If Analysis tools that show outcomes from sets of input values. In this chapter, the decisive phrase is read patterns, not isolated attractive cells. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: The lowest cost cell is highlighted without reading the workload that produces it.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. scan by row
- 2. scan by column
- 3. compare corners
- 4. identify reversals or steep changes
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Comparative language supports reasoning. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
Sudden jumps may reflect formulas, rounding, bands or genuine model behaviour.
A one-variable table changes one input and may display multiple formulas; a two-variable table changes two inputs for one formula. In this chapter, the decisive phrase is inspect thresholds and discontinuities. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: A scholarship eligibility result changes abruptly and is presented as a smooth trend.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. find the first changed cell
- 2. inspect the underlying rule
- 3. test neighbouring values
- 4. explain the threshold
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Boundary language supports decisions. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
Microsoft documents calculation options that can include or exclude Data Tables from automatic recalculation.
A Data Table cannot accommodate more than two variables, although each variable can have many test values. In this chapter, the decisive phrase is control recalculation deliberately. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: A stale table is copied after the source assumptions changed.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. record calculation mode
- 2. force a controlled recalculation
- 3. compare the baseline cell
- 4. save after verification
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: State language supports spreadsheet QA. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
A What-If table shows conditional outcomes, not how likely each input is.
Data Tables recalculate with the worksheet unless calculation options are changed, so saved and displayed results depend on calculation state. In this chapter, the decisive phrase is distinguish sensitivity from forecast and probability. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: Every row is described as an equally likely future scenario.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. use if–then wording
- 2. avoid probability claims
- 3. separate forecast methods
- 4. state unmodelled conditions
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Modal language supports honest reporting. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
The mechanism
Export, copy-paste or later edits can separate the table from its assumptions.
Microsoft describes Data Tables as What-If Analysis tools that show outcomes from sets of input values. In this chapter, the decisive phrase is verify the published table and decision note. It identifies the state or boundary to check before a click, upload, disclosure or connection turns an assumption into a real consequence.
A practical school or study situation
Example: The chart is updated but the shared PDF contains the earlier grid.
A Data Table reports formula outputs under selected input combinations and current fixed assumptions. It does not show likelihood, causal certainty, missing variables, model validity, affordability, fairness or authority to act.
- 1. record the final workbook
- 2. copy labels with the table
- 3. open the exported file
- 4. verify the shared object
Save the baseline; inspect the formula and units; choose one or two genuine input cells; write meaningful test values; assign row and column input cells correctly; calculate; test corners and the baseline intersection; inspect errors and thresholds; record fixed assumptions and calculation mode; compare a manual case; verify the exact workbook and exported table. Then name the remaining uncertainty. A successful click is evidence about one operation at one time; it is not an unlimited promise about safety, authorship, privacy, delivery or future behaviour.
Build a reproducible note
Record workbook version, business or study question, formula cell and unit, baseline, variable names and units, test-value sources, row input cell, column input cell, calculation mode, corner and threshold results, fixed assumptions, anomalies, reviewer, decision and final file identity.
Transfer: Release verification supports careers. The transferable habit is to preserve exact nouns and verbs, match them to authority and test the resulting state. That is a career skill as much as an English skill.
Four useful discoveries
Frame one sensitivity question
The table should answer a defined question about how inputs affect an output.
Boundary: A clear question can still use a weak model.
Preserve the baseline and calculation state
Data Tables can recalculate and change displayed results as inputs or formulas change.
Boundary: A baseline is not automatically a best estimate.
Verify the formula before multiplying results
Every cell in the table depends on the formula being correct.
Boundary: A correct formula can still represent an incomplete model.
Choose one-variable structure when one input changes
A one-variable table tests many values for one cell and can show several formulas.
Boundary: One-variable sensitivity cannot describe interaction between changing inputs.
Practice with realistic decisions
Choose two-variable structure for one formula
Two outputs are placed in a two-variable layout and the grid becomes ambiguous.
- select one output formula
- name both variables
- place the row values
- place the column values
Assign row and column input cells correctly
Rate values are sent to the term cell and months are sent to the percentage cell.
- read the table orientation
- identify the row variable
- identify the column variable
- check cells and units
Choose meaningful value ranges
A budget table tests only tiny changes and hides the point where the plan fails.
- cite a range source
- include the baseline
- test plausible extremes
- use finer steps near thresholds
Advice for students, families and educators
For students
- Include the baseline in the table and calculate one combination manually.
- Use if–then sentences instead of calling every result a prediction.
For families
- Ask what changed and what stayed fixed before discussing the most attractive cell.
- Help check units and practical bounds without choosing the learner’s conclusion.
For educators
- Require students to identify row and column input cells in words.
- Assess the limitations paragraph alongside the table and chart.
Frequently asked questions
What is an Excel Data Table?
A What-If Analysis range showing formula outcomes for sets of input values.
How many variables can it change?
One or two.
Can a one-variable table show several formulas?
Yes, when they share the same input cell.
Can a two-variable table show several formulas?
Its standard structure evaluates one formula across two input lists.
What is the row input cell?
The model input replaced by values arranged across the table’s row.
What is the column input cell?
The model input replaced by values arranged down the table’s column.
Does a Data Table predict the future?
No; it calculates conditional model results.
Why might results be stale?
Calculation settings or an unrecalculated workbook may preserve earlier values.
What should I verify?
Formula, variables, ranges, units, mapping, calculation state, assumptions and exact file.
Why English?
Because variable, input, output, row, column, scenario, sensitivity and forecast describe different claims.
Useful next reading
- Excel Goal Seek — work backward from one desired formula result
- Excel Scenario Manager — compare named sets with more changing cells
- Excel Solver — solve a constrained multi-variable model
Closing thought
Sensitivity analysis is useful when it makes assumptions visible. Choose meaningful ranges, label every unit, read the whole surface rather than one attractive cell and explain each result as conditional on the model you actually tested.