Control Charts, Trends, and Alert Limits
Use time-series graphics and control charts to distinguish expected variation, unusual signals, drift, and decision limits.
Use time-series graphics and control charts to distinguish expected variation, unusual signals, drift, and decision limits.
Core science
A run chart preserves values in time order and can reveal shifts, trends, cycles, and sudden changes. A control chart adds statistically derived limits based on a stable process model. Control limits describe expected process variation; specification, alert, and action limits describe requirements or decisions. They are not interchangeable. A process can be stable but consistently outside a specification, or unstable while most individual points remain inside a wide specification.
Chart choice depends on the data: continuous measurements, counts, proportions, individual values, subgroup means, ranges, or moving ranges. Sampling interval, subgroup definition, measurement resolution, censoring, missing data, seasonality, and autocorrelation affect limits and signals. Recalculating limits after every excursion can hide deterioration. Limits should be established from an appropriate reference period and changed through documented review.
Rules such as one point beyond a limit, several points on one side, or a sustained trend are prompts for investigation, not proof of a specific cause. Multiple simultaneous charts increase false alarms. Biological growth and seasonal outdoor data may have deliberate trends, so detrending, stage-specific baselines, or model-based residuals may be needed. Preserve raw data and annotate irrigation, maintenance, treatment, weather, and crop-stage events.
Why this matters in cultivation
- Use charts for stable, repeated processes such as meter checks, irrigation delivery, room conditions, drying mass, laboratory controls, and recurring quality metrics. Define who reviews signals and what action follows.
Measure and record
Record 1
Define the process, measurand, data source, sampling or subgroup rule, time base, chart type, reference period, center-line method, and control-limit calculation before interpreting signals.
Record 2
Keep specification, alert, and action limits separate from statistical control limits. Record missing-data rules, outlier handling, maintenance or process annotations, and any known changes in measurement systems.
Record 3
For each signal, document the rule triggered, investigation, suspected and confirmed causes, action taken, and whether subsequent data demonstrate recovery or a new stable process.
Common misconceptions
Evidence limits and uncertainty
Control-chart validity depends on an appropriate process model, stable measurement, meaningful subgrouping, and treatment of autocorrelation or nonstationarity.
A statistically stable process can still be biologically poor or outside specification, and an improving process can temporarily be statistically unstable.
Check your reasoning
- For "Control Charts, Trends, and Alert Limits", explain the mechanism behind this objective: Use time-series graphics and control charts to distinguish expected variation, unusual signals, drift, and decision limits. Which observation or measurement would best test whether that mechanism is operating in the real crop?
- A learner claims, "A point inside control limits is automatically acceptable." Use the lesson’s science and evidence limits to explain why that claim is unreliable, then name one observation or measurement that could separate the competing explanations.
- Applied case — Use charts for stable, repeated processes such as meter checks, irrigation delivery, room conditions, drying mass, laboratory controls, and recurring quality metrics. Define who reviews signals and what action follows. Build a verification plan using the lesson’s record set (Process and measurand; data source; sampling/subgroup rule; time base; chart type; center and limit method; reference period; specification/alert/action limits; missing/outlier rule; annotations; signal; investigation; action; effectiveness.). What would you compare before and after the action, and what result would make you revise the original interpretation?
Require lesson-specific evidence, not memorized universal targets. Open the rationales after you have written or discussed your own answer.
Answer rationale 1: Mechanism / workflow rationale
- A strong answer should connect the response to the lesson objective: Use time-series graphics and control charts to distinguish expected variation, unusual signals, drift, and decision limits.
- A run chart preserves values in time order and can reveal shifts, trends, cycles, and sudden changes. A control chart adds statistically derived limits based on a stable process model. Control limits describe expected process variation; specification, alert, and action limits describe requirements or decisions. They are not interchangeable. A process can be stable but consistently outside a specification, or unstable while most individual points remain inside a wide specification.
- Chart choice depends on the data: continuous measurements, counts, proportions, individual values, subgroup means, ranges, or moving ranges. Sampling interval, subgroup definition, measurement resolution, censoring, missing data, seasonality, and autocorrelation affect limits and signals. Recalculating limits after every excursion can hide deterioration. Limits should be established from an appropriate reference period and changed through documented review.
- The most useful verification evidence includes Define the process, measurand, data source, sampling or subgroup rule, time base, chart type, reference period, center-line method, and control-limit calculation before interpreting signals..
- Keep this limit explicit: Control-chart validity depends on an appropriate process model, stable measurement, meaningful subgrouping, and treatment of autocorrelation or nonstationarity.
Answer rationale 2: Misconception rationale
- The shortcut is unreliable because the lesson explicitly teaches a more conditional explanation.
- Representative misconception: A point inside control limits is automatically acceptable. Statistical control describes process behavior, while acceptability depends on specifications or biological requirements.
- A run chart preserves values in time order and can reveal shifts, trends, cycles, and sudden changes. A control chart adds statistically derived limits based on a stable process model. Control limits describe expected process variation; specification, alert, and action limits describe requirements or decisions. They are not interchangeable. A process can be stable but consistently outside a specification, or unstable while most individual points remain inside a wide specification.
- A useful discriminator is Keep specification, alert, and action limits separate from statistical control limits. Record missing-data rules, outlier handling, maintenance or process annotations, and any known changes in measurement systems..
- Do not overextend the conclusion beyond this limit: Control-chart validity depends on an appropriate process model, stable measurement, meaningful subgrouping, and treatment of autocorrelation or nonstationarity.
Answer rationale 3: Applied verification rationale
- In practice: Use charts for stable, repeated processes such as meter checks, irrigation delivery, room conditions, drying mass, laboratory controls, and recurring quality metrics. Define who reviews signals and what action follows.
- Record before action: Define the process, measurand, data source, sampling or subgroup rule, time base, chart type, reference period, center-line method, and control-limit calculation before interpreting signals..
- Also record: Keep specification, alert, and action limits separate from statistical control limits. Record missing-data rules, outlier handling, maintenance or process annotations, and any known changes in measurement systems..
- After the action, repeat the same measurement or observation so the comparison is valid.
- Revise the interpretation if the result conflicts with the lesson limit or the expected response: Control-chart validity depends on an appropriate process model, stable measurement, meaningful subgrouping, and treatment of autocorrelation or nonstationarity.
Related lessons
Sources and evidence
- NIST/SEMATECH e-Handbook of Statistical MethodsV21-SRC-009
Primary reference for experimental design, measurement process characterization, control charts, uncertainty, regression, and statistical graphics.
- ISO 5725 series — Accuracy (trueness and precision) of measurement methods and resultsV21-SRC-018
Repeatability, reproducibility, trueness, precision, and method-performance concepts; current parts require controlled access and version review.
- JCGM 106 — The role of measurement uncertainty in conformity assessmentV21-SRC-019
Decision rules, guard bands, acceptance limits, and uncertainty in pass/fail decisions.
- EPA — Quality System and Quality Assurance Project PlansV21-SRC-026
Data-quality objectives, sampling design, QA/QC, documentation, validation, and corrective action.
- Montgomery — Design and Analysis of ExperimentsV21-SRC-030
General experimental-design reference for randomization, replication, blocking, factorials, interactions, and model checking.
Downloads
No lesson-specific download is approved for this release. Use browser print/save-to-PDF when you need an offline reading copy.