aiMOOC

Statistical distributions and scatter graphs - Mathematics Year 8

QR-Code

Introduction

Can two groups have the same mean and still tell very different stories? Compare complete distributions, then read positive, negative and non-linear-looking point patterns without turning correlation into a causal claim.

You will compare two five-value distributions and interpret three actual schematic scatter graphs. All teaching scores and point lists are invented; no real learner data or research result is being reported.

Compare two distributions using centre and spread

A single-variable distribution records how one quantity varies across observations. The mean is the sum divided by the number of observations. The median is the central ordered value, or the average of the two central values for an even number of observations. The range is maximum minus minimum. Each summary answers a different question; no one number describes the whole distribution.

Compare the two complete data sets. Values are invented counts of completed practice items in the same stated session. They are sorted, measured in the same unit and both have five observations. This permits a comparison of these supplied groups, but does not establish that a school-wide or national sample is representative.

Group	Sorted counts	Sum	Number	Mean	Median	Range
A	2, 3, 4, 5, 26	40	5	8	4	24
B	6, 7, 8, 9, 10	40	5	8	8	4

For A, 2+3+4+5+26=40, so mean 40÷5=8. The third ordered count is 4, giving median 4. The range is 26−2=24. For B, 6+7+8+9+10=40, so mean 8, median 8 and range 4. Both have the same mean, but A has four low counts and one much larger count; B is clustered more tightly around 8. Neither has a unique most frequent value because each supplied value appears once.

If the question asks for a typical count near the main cluster in A, the median 4 is informative and less affected by 26 than the mean 8. If the question asks for an equal-share total per observation, the mean 8 has a direct interpretation. Do not announce that the median is always better or remove 26 just to make a mean look nicer. The context may justify it as a genuine value; it may also require checking a transcription error. State the purpose and inspect the raw data.

Independent checks and limits. Reconstruct the sum using mean×number: 8×5=40 in both groups. Confirm the middle value after sorting, not before. The ranges show different total spreads but not the distribution shape. For example 0, 5, 10 and 0, 0, 10 both have range 10, yet their middle values are 5 and 0. Equal range does not mean equal clustering or centre.

If 26 in A were a confirmed recording error and corrected to 6, the new sorted set 2, 3, 4, 5, 6 would have sum 20, mean 4, median 4 and range 4. That conditional calculation is not permission to alter real data without evidence. Keep original and corrected versions labelled and explain the reason for a justified change.

Read positive and negative paired-point patterns

A scatter graph records two numerical variables together. Each point is one observation's x value and y value. Two observations may share x while having different y. Read the labelled units and scales, then look at the whole cloud: direction, how closely points follow a pattern, clusters and unusual points. A scatter graph is not a categorical bar chart, and points are not automatically joined in order.

Here x and y are invented dimensionless scores. All three original graphs use x 1–5 and y 0–10 at one-unit intervals; the exact lists accompany the diagrams for access and checking. A small diagram cannot prove the pattern would continue outside its displayed data.

Graph A: a main rising group and one unusual point.

Original Graph A: paired invented x,y scores
o = one observation; . = empty grid point
 y
10 | .  .  .  .  . 
 9 | .  .  .  .  o 
 8 | .  .  .  .  . 
 7 | .  .  .  o  . 
 6 | .  .  .  .  . 
 5 | .  .  o  .  . 
 4 | .  .  .  .  . 
 3 | .  o  .  .  . 
 2 | .  .  .  .  . 
 1 | o  .  .  .  o 
 0 | .  .  .  .  . 
   +---------------
     1  2  3  4  5  x
Both axes use one-unit intervals.

Exact point list: (1, 1), (2, 3), (3, 5), (4, 7), (5, 9), (5, 1).

Five points rise from(1, 1) through(3, 5) to(5, 9). The extra point(5, 1) sits well below that main rising pattern. Across all six observations there is still an overall positive linear tendency, but it is not a perfect straight-line relationship. Call the low point unusual relative to the main pattern, check its provenance and retain it unless there is a justified reason for exclusion. Two different y values at x 5 are allowed because these are observations, not a rule giving one output for every input.

Graph B: a falling pattern.

Original Graph B: paired invented x,y scores
o = one observation; . = empty grid point
 y
10 | .  .  .  .  . 
 9 | o  .  .  .  . 
 8 | .  .  .  .  . 
 7 | .  o  .  .  . 
 6 | .  .  .  .  . 
 5 | .  .  o  .  . 
 4 | .  .  .  .  . 
 3 | .  .  .  o  . 
 2 | .  .  .  .  . 
 1 | .  .  .  .  o 
 0 | .  .  .  .  . 
   +---------------
     1  2  3  4  5  x
Both axes use one-unit intervals.

Exact point list: (1, 9), (2, 7), (3, 5), (4, 3), (5, 1).

As x increases, the displayed y values decrease in a straight pattern: this sample shows negative correlation. It does not mean that the y scores themselves are negative; all five are positive. In these invented values each rise of one x unit corresponds to a fall of two y units. That describes the plotted relationship, not a demonstrated cause.

Compare a trend description with a claim about an individual. A trend can support a cautious estimate within the observed range, but an unusual point means that the same x need not give the predicted y. A line should summarise a cloud rather than join points in their data-entry order. Never infer a causal effect just from the direction of a plotted line.

Interpret a third graph and resist causal shortcuts

Graph C: no overall rising or falling linear tendency.

Original Graph C: paired invented x,y scores
o = one observation; . = empty grid point
 y
10 | .  .  .  .  . 
 9 | .  .  .  .  . 
 8 | .  o  .  o  . 
 7 | .  .  .  .  . 
 6 | .  .  .  .  . 
 5 | .  .  o  .  . 
 4 | .  .  .  .  . 
 3 | .  .  .  .  . 
 2 | o  .  .  .  o 
 1 | .  .  .  .  . 
 0 | .  .  .  .  . 
   +---------------
     1  2  3  4  5  x
Both axes use one-unit intervals.

Exact point list: (1, 2), (2, 8), (3, 5), (4, 8), (5, 2).

The two end points have y 2, the next two inward points have y 8 and the centre has y 5. The left-to-right rise is balanced by the fall. There is no overall positive or negative linear trend in these five points. However, their arrangement is not evidence of no relationship of any kind: it suggests a possible curved or grouped pattern worth investigating with more data. “No linear correlation” and “no relationship” are different claims.

Why correlation is not a cause. Even a strong positive or negative pattern can have other explanations: a third variable, selection effects, measurement problems or chance in a small sample. A hypothetical connection between practice time and a test score would not, by itself, show that adding practice caused the whole score difference. Initial skill or access to support could matter. These explanations are possibilities to investigate, not findings from our invented scores.

Compare the original Corbettmaths scatter-graph tutorial with our exact A, B, C point lists. Its visible plotted line belongs to its own example. Read both axes, inspect unusual points and distinguish an estimate from a guaranteed individual value. Original: corbettmaths, Scatter Graphs - Corbettmaths. The written explanations and native tasks work without sound or the external player.

Three-graph assessment. For A name the positive main pattern and low point(5, 1), and explain why the full sample is not perfect. For B state negative correlation and distinguish decreasing values from negative numbers. For C state no overall linear direction, while explicitly leaving open non-linear structure. Support each conclusion with at least two listed coordinates rather than an impression of the page layout.

New distribution transfer. The ordered set 3, 4, 5, 6, 22 has sum 40, mean 8, median 5 and range 19. Compare it with B: the means agree, but median 5 versus 8 and range 19 versus 4 show different centre and spread. The value 22 strongly affects the mean; its presence alone is not proof of an invalid observation. State whether your summary is intended to show equal share, a central ordered observation or overall spread.

Compare raw-value calculation with visual reasoning. The former gives exact summaries for the supplied lists; the latter shows patterns and unusual values that a single summary can conceal. Retain both. Real sampling quality cannot be checked from five invented numbers alone, and a truncated vertical axis can make a graph look more dramatic without changing the data. Keep scales visible, avoid extrapolating beyond the observed x range and record only actual later retrieval or research work.

Interactive Tasks

Quiz: Test Your Knowledge

Group A 2, 3, 4, 5, 26: mean? (8) (!4) (!26) (!24)

That group: median? (4) (!8) (!5) (!26)

That group: range? (24) (!26) (!4) (!8)

Group B 6, 7, 8, 9, 10: median? (8) (!4) (!10) (!40)

Which comparison is correct? (Both means are 8 but their medians and ranges differ) (!Both distributions are identical) (!Both ranges are 4) (!Both medians are 4)

Which point is unusual relative to Graph A main rising group? (5, 1) (!1, 1) (!3, 5) (!4, 7)

Graph B shows which direction? (Negative correlation) (!Positive correlation) (!No variation in y) (!A causal proof)

Graph C supports which conclusion? (No overall linear direction; other structure remains possible) (!No relationship of any kind is proven) (!Perfect positive correlation) (!All observations have the same y)

What does a scatter point represent? (One paired x, y observation) (!A whole category bar) (!A guaranteed future outcome) (!A causal mechanism)

What should happen before deleting an unusual value? (Investigate and justify any exclusion) (!Delete it to improve the graph) (!Replace it with the mean automatically) (!Ignore its recorded provenance)

Memory Game

Mean	Total divided by number of observations
Median	Central value after ordering the data
Range	Maximum minus minimum
Correlation	Association between two numerical variables
Unusual point	Observation away from the main plotted pattern
Causation	A cause producing an effect, not established by correlation alone

Drag and Drop

Mean 8	Equal-share summary for the supplied Group A
Median 4	Middle ordered value of Group A
Range 24	Maximum-minus-minimum spread of Group A
Negative trend	Direction in Graph B
Investigate 5, 1	Responsible response to Graph A unusual low observation

Crossword Puzzle

MEAN	Sum divided by count
MEDIAN	Ordered central value
RANGE	Maximum minus minimum
SCATTER	Graph type plotting paired observations
CORRELATION	Association without an automatic causal conclusion
SAMPLE	Observed subset whose selection affects generalisation

LearningApps

Optional extra practice. Choose an appropriate activity with your teacher. If the external collection is unavailable, use the native activities and independent tasks in this course.

Cloze Text

Open Tasks

Easy

Calculate the source summaries: Show sum 40, mean 8, median 4 and range 24 for Group A.
Compare with Group B: Calculate all three summaries and identify the common mean.
Read one paired point: Explain x and y for 5, 1 in Graph A.
Distinguish graph types: Contrast paired scatter points with the separate bar-chart illustration.

Standard

Interpret all three graphs: Use point evidence for direction, unusual values and limits.
Choose a summary for a purpose: Compare equal-share mean with typical-cluster median in Group A.
Check a correction conditionally: Recalculate only if 26 is confirmed as a recording error and corrected to 6.
Avoid causal inference: Give two alternative explanations for a hypothetical observed association.

Advanced

Transfer to a new distribution: Calculate summaries for 3, 4, 5, 6, 22 and compare with B.
Explain the curved-pattern limit: Use Graph C to distinguish no linear trend from no relationship.
Audit sample and scales: Explain why five supplied observations cannot prove a population conclusion.
Retain a justified investigation: Keep raw data, both summary methods, all three graph interpretations and actual later corrections.

Learning Check

Compare both full distributions: Calculate centre and spread, then interpret the difference.
Interpret Graph A: Use the full six-point list without hiding the unusual observation.
Interpret Graphs B and C: Distinguish negative trend and no overall linear direction.
Purpose, range and causal errors: Correct three unjustified inferences.
Unfamiliar transfer and conditional correction: Check 3, 4, 5, 6, 22 and a confirmed 26-to 6 correction.
Methods, portfolio and evidence limits: Retain raw values and actual diagram-based conclusions.

Worked checks and assessment guidance

A: sum 40, n 5, mean 8, median 4, range 24. B: sum 40, n 5, mean 8, median 8, range 4. A has four lower values and one much larger value; B is more tightly clustered. No unique most frequent value occurs in either set.
Main pattern rises from 1, 1 to 5, 9; extra 5, 1 is far below it. Overall tendency remains positive but is not perfect. Equal x values with different y are allowed. Investigate provenance before any exclusion.
B falls 9, 7, 5, 3, 1 as x rises 1–5: negative correlation with positive y values. C values 2, 8, 5, 8, 2 have no overall linear direction; this does not rule out a curved or grouped relationship.
The median is not always best: mean has an equal-share meaning. Equal range does not prove equal distribution shape, as 0, 5, 10 versus 0, 0, 10 shows. Correlation does not prove cause; confounding, selection or measurement could contribute.
New set sum 40, mean 8, median 5, range 19; compare B median 8, range 4. Only if justified, corrected A 2, 3, 4, 5, 6 has mean 4, median 4, range 4. Label original and changed data distinctly.
Keep exact summary calculations, independent total checks, the three labelled plots and point lists, first-error correction and new comparison. No real students, research finding, population representation, causal result or delayed mastery is invented.

Learning evidence

Retain two complete distribution comparisons, three plotted point-pattern interpretations with exact lists, an unfamiliar transfer, justified method choices and explicit sampling/causation limits.

OER and Wikipedia

Department for Education: England mathematics programme of study, Key Stage 3, geometry, measures, statistics, fluency, reasoning and problem solving. Year 8 allocation is editorial within KS3. All numerical cases, original schematic graphs and feedback are newly authored teaching material. Actual Commons creators/licences follow; original-publisher videos supplement the complete written teaching. Reviewed 12 November 2026.

Simple bar chart.svg, Masur, Public domain. Unchanged original; caption and learning activities are original.

2D Cartesian Coordinates.svg, H Padleckas, Public domain. Unchanged original; caption and learning activities are original.