KNOWLEDGE WAREHOUSE · OBJECT 12
Statistics and Data
Statistics is the Mathematics of learning from data under uncertainty. It combines representation, summary, variability, sampling, modelling and inference so evidence can support—but not exceed—the conclusions justified by the data.
A calculation can be correct while the statistical conclusion is wrong.
Core ideas
- Data types and data collection.
- Tables, charts, histograms, box plots and scatter plots.
- Mean, median, mode, range, quartiles and standard deviation.
- Distribution shape, centre, spread and outliers.
- Correlation and regression.
- Sampling, probability models, confidence and hypothesis testing at advanced stages.
Prerequisites and representations
Prerequisites: number sense, percentage, ratio, algebra, graph reading and probability. Representations include raw data, frequency tables, plots, summary statistics, probability distributions and model output.
Failure signatures
- Chooses mean automatically even when skew/outliers make median more informative.
- Reads a graph without checking axis scale or sample size.
- Confuses correlation with causation.
- Computes standard deviation but cannot interpret what spread means.
- Performs a hypothesis test mechanically without stating the model, hypotheses or conclusion in context.
Diagnostic probes
- Construct two datasets with the same mean but very different spread.
- When would median be a better summary than mean?
- What information is lost when raw data are reduced to one average?
- Does a strong correlation prove one variable causes the other? Give a counterexample.
- Explain a p-value or hypothesis-test conclusion in the language of the original context rather than formula notation.
Repair and transfer
Repair by reconnecting calculation to the question being asked. Require the learner to describe distribution shape, compare representations and justify the chosen summary/model before calculating. Transfer is verified when the learner can critique unfamiliar data displays, identify limitations and communicate a bounded conclusion.
Stage progression
Primary introduces tables, graphs and simple averages. Secondary develops statistical diagrams and summary measures. H1 Mathematics gives substantial weight to Probability and Statistics; H2 extends probability distributions, sampling, hypothesis testing, correlation and regression. Current SEAB H1 specifically frames Mathematics and Statistics as preparation for informed decision-making and business/social-science study.
Downstream dependencies
Data science, research, economics, social science, medicine, engineering, AI evaluation and evidence-based decision-making all depend on statistical reasoning.
TECHNOLOGY: spreadsheets, graphing calculators, statistical software and AI can accelerate analysis; model choice, assumptions, interpretation and causal restraint must remain humanly accountable.
PHASE 4 · STATISTICS & DATA READER GUIDE
Quick Read: what is statistics really doing?
Statistics turns data into evidence. The learner must decide what the data represent, how to summarise variation and what conclusions the evidence does—and does not—justify.
A calculator can produce means, standard deviations, regressions and test statistics quickly. The mathematical work is not finished there. The student still has to know which summary is appropriate, what assumptions are being made and how to translate the result back into careful language.
One-sentence answer: statistics becomes secure when the learner can move from data to representation to summary to interpretation without claiming more than the data support.
Data first: what was measured?
- Variable: what quantity or category was observed?
- Population: which larger group is of interest?
- Sample: which observations were actually collected?
- Measurement process: how were the values obtained?
- Context: what conditions might affect interpretation?
A numerical summary is only as meaningful as the data-generating process beneath it. Poor sampling or ambiguous measurement cannot be repaired by more sophisticated calculation.
Centre and spread answer different questions
| Summary | Question it answers | Typical risk |
|---|---|---|
| Mean | Where is the numerical balance point? | Can be pulled by extreme values. |
| Median | What is the middle ordered value? | Can hide how widely values vary. |
| Range / IQR | How spread out are the observations? | A single spread measure cannot describe the whole distribution. |
| Standard deviation | How dispersed are values around the mean? | Meaning is lost if treated as calculator output only. |
Two datasets can have the same mean and very different variability. Strong interpretation therefore considers location and spread together.
A graph is an argument about the data
Bar charts, histograms, box plots, scatterplots and cumulative displays reveal different structures. Choosing a graph is therefore part of statistical reasoning.
- Use a representation that matches the variable type.
- Check axes, scales and bin choices.
- Ask what patterns become visible and what information is hidden.
- Do not infer causation from a visual association alone.
A misleading scale can exaggerate a small difference; a wide histogram bin can hide structure. Representation choices affect what the reader sees.
Correlation is not causation
A statistical association tells us that variables move together in the observed data. It does not by itself establish that changing one causes the other. Confounding variables, selection effects or reverse direction can create or distort the pattern.
A regression line can describe association. Causal explanation requires stronger design and evidence.
Three statistics students who need different repair
- Student A calculates the mean correctly but cannot decide whether it is representative. Repair interpretation and distribution reading.
- Student B produces a regression output but states that one variable causes the other. Repair evidence language and the distinction between association and causation.
- Student C understands the concept but misreads calculator output. The issue may be tool control, labels or notation rather than statistical meaning.
Again, the same low mark can hide different mechanisms. Written interpretation and the choice of representation are as diagnostic as arithmetic.
From description to inference
Descriptive statistics summarise observed data. Inferential statistics use sample evidence to reason about a wider population or model under stated assumptions.
- Define the question.
- Understand the sample and population.
- Choose an appropriate model or procedure.
- Compute the statistic or interval.
- Interpret the result in context.
- State uncertainty and limitations.
The important development is epistemic: the learner must separate what was directly observed from what is inferred.
Statistics across the school journey
| Stage | Statistical demand | Key transition |
|---|---|---|
| Primary | Read and construct simple tables and graphs; compare data. | Move from individual values to patterns in a set. |
| Secondary | Measures of centre/spread, data representations and interpretation. | Compare distributions, not only single summaries. |
| JC Mathematics | Probability distributions, sampling, correlation/regression and inference. | Use sample evidence to make cautious model-based conclusions. |
Calculator output must be translated
- Know which variable each output refers to.
- Check whether the selected model matches the question.
- Distinguish sample statistics from population parameters.
- Read signs and magnitude in context.
- State conclusions using the language justified by the procedure.
- Question implausible output instead of treating the calculator as authority.
The calculator should reduce computational burden while leaving model choice and interpretation with the learner.
What parents can notice
- Can the learner explain what the data measure?
- Can they say why one summary is more useful than another?
- Can they describe spread as well as average?
- Do they distinguish correlation from causation?
- Can they interpret calculator output in ordinary language?
- Can they identify limitations in sampling or measurement?
A useful question is: “What does the data actually allow you to claim?” That keeps statistical reasoning disciplined.
Frequently asked questions
Why can two datasets with the same mean feel different?
Because centre does not describe spread or shape. Variability and distribution structure matter.
Is a strong correlation proof of cause?
No. Correlation measures association. Causal claims need stronger evidence about design, confounding and direction.
Why do statistics questions require so much writing?
Because the conclusion is part of the Mathematics. A numerical result without interpretation may not answer the statistical question.
How do we know statistical understanding has transferred?
The learner can inspect an unfamiliar dataset, choose a justified representation or summary, use appropriate technology and state a conclusion with correct uncertainty.
The larger idea: statistics is disciplined compression of evidence
Data are detailed and noisy. Statistics compress them into summaries, models and conclusions. Every compression loses some information, so the learner must know what was preserved and what was hidden.
The mature student therefore treats a statistical result as an evidence statement with conditions, not as a number detached from how the data were produced.
Statistics is strongest when the calculation becomes a careful statement about evidence rather than a machine-produced answer.