eduKateSengkang · Family learning
Find your next learning step
Choose the route closest to your question. Read one worked example, try a fresh case and use the explanation to check the decision behind your answer.
Full chapter index · Start the diagnostic · Continue at the learning hub
A scatter plot is useful when each observation has two numerical measurements. This guide helps families preserve those pairs, plot them accurately and describe patterns without claiming more than the evidence supports. All datasets below are original fictional teaching examples, not measurements of Sengkang children, schools or services.
Choose one route and one practice set. This is a flexible learning resource, not a statement that every section belongs to a particular school syllabus. Formal correlation calculations are outside its scope; the emphasis is clear reading, justified estimates and careful explanations.
Coordinates for children · Linear graphs for families
Choose a chapter
Keep measurements paired · Chapters 1–4
Plot and describe · Chapters 5–9
Interpret carefully · Chapters 10–15
Estimate and practise · Chapters 16–20
Explain and continue · Chapters 21–24
CHAPTER 1 OF 24 · Keep measurements paired
1. Start with one observation and two measurements
Imagine six cardboard ramps. For each ramp, a child records its height in centimetres and the distance a toy car travels in centimetres. One row might say height 8, distance 43. That row belongs together because both values describe the same trial. A scatter plot places a dot at the corresponding coordinate. The dot is a compact record of a relationship within one observation, not two unrelated numbers chosen from separate lists.
The observation could instead be a seedling, a container or a day. Before plotting, complete the sentence “Each dot represents one ___.” If that sentence is unclear, pause. A graph cannot repair uncertainty about what was measured. Two columns labelled height and distance need enough context to tell a reader which height and which distance belong to each other.
For a first family task, use three pairs: (4, 24), (8, 43) and (12, 57). Say what the first coordinate means and then what the second means. Ask whether (8, 57) is an observation in this dataset. It is not. Both numbers appear somewhere in the list, but they were not recorded together. This distinction is the foundation of honest plotting.
Keep a small identifier such as trial A beside each row during preparation. The identifier does not normally become an axis value; it allows someone to trace a dot back to the original record. That trace is useful if a measurement seems unusual or a copied coordinate needs checking.
CHAPTER 2 OF 24 · Keep measurements paired
2. Decide whether a scatter plot fits the question
A scatter plot helps investigate how two numerical variables vary together. “Does greater ramp height tend to accompany greater travel distance in these trials?” is suitable. “How many children chose each colour?” is a category-count question and usually needs a different display. A list of favourite colours does not become numerical simply because colours receive arbitrary codes such as red equals one and blue equals two.
Compare a line graph of one plant’s height over ten days with a scatter plot of ten different plants’ heights and leaf counts. The first has a meaningful time sequence within one plant. The second compares paired measurements across plants. Both may use numerical axes, but the intended relationships differ. Choosing a graph begins with the question and observation, not with the available coloured pencils.
Some datasets can support more than one useful display. A time series can be plotted as points, and a scatter plot can use time as a variable. What matters is what the chosen display communicates. Avoid teaching that every plot with dots is a scatter plot for investigating association or that every numerical horizontal axis requires joining observations.
Ask the learner to propose a question before opening graph paper. Then identify the observational unit, the two measurements and the reason they belong together. If the task has only one numerical measurement and a category label, explain why a bar chart or another summary may answer it more directly. That decision is part of statistical reasoning.
CHAPTER 3 OF 24 · Keep measurements paired
3. Preserve the pairs when preparing a table
Consider original records for paper strips: A has length 10 cm and mass 2 g; B has length 15 cm and mass 3 g; C has length 20 cm and mass 5 g; D has length 25 cm and mass 4 g. The slightly unexpected final mass is not a reason to pair it with a different length. The data must retain the observations actually supplied.
Sorting the rows by length is acceptable if each entire row moves together. Sorting the length column and mass column separately destroys the pairs and may manufacture a neat pattern. A family check is to read one identifier before and after sorting. Strip C must still have length 20 and mass 5. If it does not, the preparation changed the evidence.
Missing measurements need explicit treatment. If strip E has a recorded length but no mass, it cannot supply a complete point on this two-variable scatter plot. Mark the mass as missing and explain whether E was omitted from the display. Do not replace the missing value with zero unless zero was actually measured. Missing and zero describe different information.
Use a simple audit before plotting: count the complete pairs, check units, inspect obvious copying mistakes and retain uncertain entries for clarification. An apparent error may be a real observation. Correct it only when the source record supports the correction. The graph should be a faithful representation of the table, not an attractive improvement on it.
Suppose horizontal x represents ramp height in centimetres and vertical y represents travel distance in centimetres. The pair (8, 43) means move to 8 on the horizontal scale and then to 43 on the vertical scale. Reversing the axes can still produce a valid display if every label and point is consistently reversed, but it answers and presents the relationship differently.
Write variable names and units, not just x and y. “Height” alone could refer to a ramp, child or shelf. “Ramp height, cm” supplies a usable description. If another person reads the graph without the original worksheet, the axes should still tell them what each coordinate measures. A title can then identify the dataset or question without carrying every technical detail.
There is no general rule that the variable on the horizontal axis must be a proven cause. In a planned experiment, the deliberately varied input often appears there. In an observational comparison, the choice may reflect the question or convention. Placing a variable on x does not establish that it causes changes in y. Keep the graphical choice separate from the strength of the evidence.
For checking, select two dots and translate them into complete sentences with units. If the learner says “eight and forty-three,” ask eight what and forty-three what. This small language step exposes swapped axes and missing context before the learner begins making claims about the overall pattern.
CHAPTER 5 OF 24 · Plot and describe
5. Choose a scale that can be read consistently
A useful scale covers the relevant values and uses equal numerical increments at equal physical intervals. If one vertical square represents 5 cm, that relationship must remain consistent along a standard linear axis. A scale that uses 5, 10, 20 and 50 at equally spaced marks changes the mapping and can seriously distort the display unless a different scale is explicitly intended and understood.
For ramp heights from 4 to 20 cm, horizontal marks every 2 cm may work. For distances from 24 to 81 cm, vertical marks every 10 cm may be convenient, with smaller divisions for plotting intermediate values. The horizontal and vertical scales need not use the same numerical increment. Different units or ranges often make different scales sensible.
Scatter-plot axes do not always need to begin at zero. A display comparing heights from 140 to 180 cm may use an axis covering that range, provided the labels make the range clear. Truncating an axis changes visual emphasis, so read the numbers rather than treating the plotted angle as a universal measure of strength. The same data can look steeper under a different aspect ratio.
Ask the child to locate a value halfway between labelled marks before plotting all the data. If the child cannot place 35 between 30 and 40 consistently, work on scale reading first. A beautiful pattern built on an inconsistent scale is less useful than a rough but accurately labelled graph.
Use pairs (2, 15), (4, 22), (6, 31), (8, 35) and (10, 48), with x as the number of identical paper layers and y as the load supported in grams in a fictional test. Plot one point at a time from a complete row. Do not read the entire x column first and then try to remember which y values belong to it.
After placing a dot, read its coordinates back from the axes and compare them with the row. This reversal is an effective check because it tests the graph as a reader will encounter it. If (6, 31) has been placed at (6, 13), the read-back reveals the transposition. Merely looking at the expected upward pattern may not.
Use small, clear dots or crosses. Huge symbols can cover neighbouring observations, especially when values are close. If the exercise specifies a symbol, follow it. If a digital plot allows hovering, a tooltip can help identify rows, but the underlying axes and units still need to work without interaction. Technology should support the reference, not replace basic labelling.
Count completed rows and plotted observations at the end. Different coordinates should normally produce corresponding visible marks, but identical coordinates may overlap, a complication discussed next. A count discrepancy is a reason to investigate rather than immediately add a dot in an empty area. The correct location must come from a particular recorded pair.
Suppose five observations are (3, 12), (3, 12), (4, 15), (5, 18) and (5, 18). A simple plot may show only three visible positions because two positions each contain overlapping observations. Three visible marks do not prove there were only three measurements. The table contains five observations, and a display should not silently erase that fact when the count matters.
For a small teaching dataset, write a note such as “two observations at this coordinate” or use a clearly explained frequency symbol. Digital methods may use transparency, proportional symbol sizes or slight jitter. Jitter deliberately shifts the visual position and must be explained; a shifted point should not be mistaken for an exact measured coordinate. Beginners can retain the table and use explicit count notes.
Do not add a small random amount to an actual recorded value merely to make the dots separate. Presentation adjustments and data changes are different operations. If a plot uses jitter for visibility, retain the original numbers and state what was adjusted visually. Otherwise a reader might infer variation that the measurements did not contain.
Ask two questions: how many observations are listed, and how many distinct coordinate positions occur? For the example, the answers are five and three. Both counts are correct because they count different things. This distinction also explains why a screenshot of dots may be insufficient evidence for estimating a dataset’s total size without a legend or accompanying record.
In pairs (1, 12), (2, 18), (3, 17), (4, 26), (5, 29) and (6, 33), larger x values generally accompany larger y values. That is an upward or positive association in this small fictional dataset. The drop from 18 to 17 does not automatically destroy the overall tendency. Real and teaching data can vary around a pattern rather than follow a perfect staircase.
Describe the variables, not just the slope direction: “Among these observations, greater practice time tends to accompany a higher task score,” if those are the stated measurements. Do not write “more x means more y” without identifying the quantities. A useful explanation tells the reader what changed together and limits the claim to the observations or population justified by the collection method.
Positive does not mean beneficial. If x is waiting time and y is the number of complaints, an upward association may describe an undesirable situation. The mathematical word names direction, not approval. Likewise, negative association is not automatically bad. Keeping everyday value judgements separate from statistical terminology prevents confusing explanations.
For practice, remove one middle point and describe whether the direction appears substantially different. Then restore it. The aim is not a formal significance test; it is to make the learner inspect the full collection rather than base a conclusion on one pair. A tendency describes how the observations sit together across the plotted range.
CHAPTER 9 OF 24 · Plot and describe
9. Read negative association without negative values
Suppose original pairs for remaining water after different elapsed times are (1, 90), (2, 81), (3, 73), (4, 60), (5, 55) and (6, 42). Larger x values generally accompany smaller y values. This is a downward or negative association. Every coordinate can be positive even though the association is negative. The word describes direction of co-variation, not whether a number lies below zero.
State the measurements and units: “Across these trials, longer elapsed time tends to accompany less remaining water.” If the process is a single container measured repeatedly, mention that design rather than pretending each row is an independent container. The observations may still be plotted, but the collection structure affects the questions that can reasonably be asked about the pattern.
A common error is to say “y decreases as you move left.” For a standard horizontal axis increasing to the right, describing a negative association usually means y tends to decrease as x increases to the right. Trace two widely separated points and compare their actual numbers. Gestures alone can hide a reversed reading.
Ask the learner to create a sentence about a hypothetical situation where a negative association would be welcome, such as fewer defects accompanying more inspection time. Then explain why that invented context is only a teaching illustration. A plausible story does not supply real evidence about a service, school or activity unless measurements were actually collected.
CHAPTER 10 OF 24 · Interpret carefully
10. No clear linear pattern is still information
Consider pairs (1, 12), (2, 27), (3, 10), (4, 25), (5, 14) and (6, 23). These do not form a convincing narrow upward or downward band. A careful elementary description is that there is no clear linear association visible in this small set. It is less defensible to declare that the variables can never be related in any population or circumstance.
The qualifier linear matters. A curved relationship can be strong while a straight-line description fits poorly. A narrow x range, imprecise measurements or a small sample can also make a pattern hard to see. “I cannot see a clear straight trend here” describes the display; “there is no relationship whatsoever” makes a much broader claim.
For family practice, ask what additional information would help: more observations across a wider relevant range, better measurement definitions or knowledge of how the cases were chosen. Do not respond to uncertainty by drawing a line anyway. A line should represent a justified summary, not an expectation that every exercise must end with one.
Compare a shuffled collection with a U-shaped collection such as (1, 9), (2, 4), (3, 1), (4, 4), (5, 9). Both resist a simple one-direction straight trend, but the second has an obvious curved pattern. The learner should describe what is visible before choosing a technical label that may conceal that difference.
CHAPTER 11 OF 24 · Interpret carefully
11. Strength concerns scatter around a pattern
An apparent strong linear association has points relatively close to a straight trend; an apparent weaker one has more spread around that trend. Steepness is a different feature. A shallow but tightly aligned band can show a clearer linear association than a steep, widely scattered collection. Changing axis scales can alter the visual angle without changing the underlying paired measurements.
Compare A: (1, 11), (2, 12), (3, 13), (4, 14), with B: (1, 5), (2, 30), (3, 16), (4, 49). A rises gently and exactly in this constructed example. B covers a larger vertical change but is less tightly aligned. Calling B stronger solely because its y values rise further confuses amount of change with closeness to a relationship.
Qualitative strength judgements should acknowledge a small dataset and the intended pattern. A formal correlation coefficient, if later taught, provides a particular numerical summary with its own limitations. It does not replace reading for curves, groups, unusual points and collection design. This guide uses visual language and does not claim numerical thresholds for strong or weak association.
Ask a learner to justify a comparison with the same axes and comparable ranges where possible. Useful language includes “the points lie closer to a straight band” or “the vertical values vary more for similar horizontal values.” “It looks better” is not yet an explanation. A good description identifies the graphical feature used to support the judgement.
Original pairs (1, 2), (2, 4), (3, 8), (4, 16) and (5, 32) rise at an increasing rate. A straight line can miss their shape badly. Describe an upward curved pattern rather than pretending the increase is constant. The change in y between successive x values grows from two to four to eight to sixteen, which supports the visual observation.
Another collection might rise and then level off, such as (1, 10), (2, 18), (3, 23), (4, 26), (5, 27). A straight line over a short part of the range might provide a rough local approximation, but extending it indefinitely could imply increases inconsistent with the observed levelling. Model choice should respond to the shape and practical context.
Do not join every dot to create a dramatic zigzag unless a meaningful sequence is part of the question. In a scatter plot across different objects, the row order may be arbitrary. A smooth trend summary is also an interpretation, not a record of a measured path between objects. Retain the distinction between individual observations and a fitted relationship.
For a family activity, calculate successive differences for equally spaced x values in a tiny constructed dataset. Then compare them with the graph. This connects numerical and visual representations. It does not mean equal differences are required in measured data; the exercise simply helps the learner recognise what a constant-change model would look like and why a clear curve calls for caution.
CHAPTER 13 OF 24 · Interpret carefully
13. Unusual points are questions to investigate
Suppose most pairs rise from (2, 20) to (10, 61), but one observation is (8, 4). That point sits far below the main collection. It may reflect a copying error, a different testing condition or a genuine unusual case. Calling it an outlier in a descriptive discussion does not tell you which explanation is correct. The original record and collection notes are needed.
Start by checking units and identifiers. Was 4 entered instead of 40? Was the measurement in metres while others were centimetres? Was the object a different material? These are questions, not permission to rewrite the value. If the recorded value is valid, keep it and consider whether the population or model needs a more careful description.
Compare conclusions with and without the point, stating exactly what was done. A trend may appear much tighter after removal, but that does not justify removal for appearance alone. If a verified entry error is corrected, document the correction. If an observation is excluded because it lies outside a defined population, explain the inclusion rule and apply it consistently.
In a family lesson, reward “I need to check the record” as a legitimate response. It shows an understanding that data are evidence from a process. A graph is not a puzzle in which every awkward dot must disappear so that the expected answer becomes tidy. Explaining uncertainty honestly is part of the skill.
Imagine two fictional materials tested separately. Within each material, longer strips support slightly less load. Yet one material is both generally longer and much stronger than the other. If all points are combined without labels, the overall cloud may slope upward. The combined pattern and the within-material patterns answer different questions. Group membership is relevant information, not a decorative colour choice.
A simple constructed pair of groups illustrates this. Group A contains (1, 12), (2, 11), (3, 10). Group B contains (7, 32), (8, 31), (9, 30). Within each group y falls as x rises. Across the combined groups, the high-x group has much higher y. Describing only one overall upward association hides the within-group direction.
Use a legend and distinguish symbols as well as colours when group identity matters. A plot that relies entirely on red and green can be difficult for some readers to interpret. Explain what each group represents and whether comparison is appropriate. Do not invent group labels after seeing the pattern merely to force a desired explanation.
This chapter is an extension, not a requirement to analyse every beginner’s picture for complex effects. Its practical lesson is modest: ask whether the observations combine different conditions or populations. A parent can prompt “Were all these objects tested in the same way?” before accepting a broad claim about the whole set.
CHAPTER 15 OF 24 · Interpret carefully
15. Association alone does not prove a cause
Suppose a fictional reading club records book length and time spent reading. Longer books may accompany longer reading times. The scatter plot describes that association, but a reader still needs context: were reading speeds similar, were books equally difficult, and did everyone finish? A relationship between two columns does not by itself explain the mechanism or exclude other relevant factors.
A particularly misleading example would compare shoe size and vocabulary score across children of very different ages. Both measurements may increase with age. The graph would not show that buying larger shoes improves vocabulary. Age could help explain the observed association. This is a teaching illustration, not a claim based on measurements of actual children.
Useful causal evidence depends on study design and other information, not on how straight the dots look. A controlled experiment can sometimes help isolate an effect; an observational comparison may leave alternative explanations. Even a well-designed study needs careful interpretation. For this family guide, the safe habit is to distinguish “tends to occur with” from “makes happen.”
Ask the learner to rewrite an overclaim: “More stickers cause higher scores.” If the only supplied evidence is a scatter plot, a better statement is “In this dataset, children with more stickers tended to have higher scores.” Then ask what collection details would matter. Perhaps stickers were awarded after the scores. The direction of explanation cannot be read from axis placement alone.
CHAPTER 16 OF 24 · Estimate and practise
16. A trend line summarises rather than connects
When a roughly straight pattern is suitable, a line of best fit can summarise the tendency. It should run through the central region of the cloud, with observations reasonably distributed around it for the intended method. It is not normally a dot-to-dot path, and it need not pass through the origin. A measured relationship may have a nonzero intercept or be relevant only over a limited range.
For a hand-drawn elementary estimate, use a ruler and judge the full collection. A precise statistical fitted line uses a specified procedure and can differ from a visual estimate. Do not claim that a hand-drawn line is uniquely exact. If a task supplies an official line, use that line rather than replacing it with a preferred one while answering an estimation question.
Two points selected on the fitted line can help read its gradient, even if they are not original observations. Keep that distinction clear. The slope describes the model’s estimated change in y per unit x. It does not mean every measured object changes by exactly that amount, nor does it turn the model into proof of causation.
Before fitting, inspect for a clear curve, separate groups or a highly influential unusual observation. A line can conceal these features while appearing mathematically convenient. If a straight-line summary is not appropriate, state the limitation. A useful model simplifies the observations while preserving the feature relevant to the question; it should not erase the reason the observations need careful reading.
CHAPTER 17 OF 24 · Estimate and practise
17. Interpolation and extrapolation have different risks
If recorded ramp heights span 4 to 20 cm, an estimate at 12 cm lies inside the observed range. That is interpolation. An estimate at 35 cm lies outside it and is extrapolation. Both may use the same fitted line, but the second relies more heavily on assuming the relationship continues beyond the evidence. The practical process may change at greater heights.
Suppose a supplied model is y = 3x + 10 for the observed range. At x = 12, the model gives 46. At x = 35, it gives 115. These are model outputs, not two newly measured results. The second should be labelled as an extrapolated estimate and considered against the physical situation. A toy car may leave the track or encounter a barrier before that assumed distance.
Uncertainty also remains inside the range because observations scatter. If values near x = 12 vary between 40 and 53, presenting 46 as a guaranteed outcome is misleading. A school exercise may request a single approximate answer from a line; include the approximate nature and units. Do not add many decimal places merely because a calculator can display them.
Ask the learner to mark the observed x range and locate the requested estimate before calculating. This order makes the evidence boundary visible. A numerically correct substitution can still support a weak practical claim if the request lies far outside the measured conditions. Checking the range is part of answering, not an optional afterthought.
CHAPTER 18 OF 24 · Estimate and practise
18. Diagnose the first error in an explanation
Read four fictional responses. A says “The data are positively associated because all y values are above zero.” B says “There are ten people because I can see ten dots.” C says “The line rises sharply, so the relationship must be strong.” D says “Practice time causes high scores because it is on the horizontal axis.” Each confuses a different feature with a conclusion.
A confuses the sign of values with the direction of association. Compare y as x increases. B overlooks overlapping coordinates and needs the observation count or legend. C confuses steepness with closeness to a trend and ignores scale choice. D treats graphical placement as causal evidence. The corrections require different teaching examples, so labelling all four answers careless would not help much.
Ask the child to repair one answer using the actual data or a clearly stated limitation. A good repair might be “The points generally rise from left to right, so these measurements show a positive association.” For B, “At least ten distinct positions are visible; the number of observations is not established by the picture alone” is more careful when no overlap information is supplied.
Choose a follow-up that targets the unstable decision. Use a downward plot entirely in the positive quadrant for A, overlapping pairs for B, rescaled axes for C and a plausible common-cause story for D. One carefully chosen contrast can reveal whether the child has changed the reasoning, whereas another page of identical plots may only rehearse the same mistake.
| Claim | Missing distinction | Better check |
|---|---|---|
| Positive because y is above zero | Direction versus sign | Compare y as x increases |
| Ten dots means ten people | Positions versus observations | Inspect overlap and records |
| Steep means strong | Slope versus scatter | Inspect closeness to a trend |
| x causes y | Display versus causal evidence | Inspect collection design |
CHAPTER 19 OF 24 · Estimate and practise
19. Independent practice: preserve and plot pairs
Use this original fictional dataset for five paper gliders. Each pair gives folded wing width in centimetres and measured travel distance in metres: A (6, 3), B (8, 5), C (10, 4), D (12, 7), E (14, 8). Create a labelled scatter plot. State the observational unit, the two variables and their units before drawing. Choose scales that let someone read every coordinate clearly.
Answer these questions independently. Which glider travelled 4 metres? Is (8, 7) a recorded observation? What does the point (12, 7) mean in a complete sentence? If the rows are sorted by travel distance, which width must remain beside 4? Does a general upward tendency require every successive travel distance to increase? Explain with a specific pair of neighbouring widths.
Now add F (10, 4), a second glider with the same pair as C. How many gliders are recorded, and how many distinct coordinate positions appear? Explain a display method that preserves the existence of F without pretending its measurements were different. If F’s width were missing rather than 10, could it be placed as a complete point? Explain why zero would not be an honest replacement for a missing width.
Finally, swap the axis assignments consistently and sketch the new display. Translate one point back into words and compare its meaning with the original graph. The pairing should remain the same even though the coordinates reverse. Keep the original table beside both displays so that you can check whether a visual change has accidentally become a data change.
CHAPTER 20 OF 24 · Estimate and practise
20. Explained answers to the plotting practice
Each observation is one glider. The original horizontal axis can show folded wing width in centimetres and the vertical axis travel distance in metres. C travelled 4 metres. The pair (8, 7) is not recorded: B has width 8 and distance 5, while D has width 12 and distance 7. Combining their values would create an observation that was never supplied.
The point (12, 7) says that the glider with a folded wing width of 12 centimetres travelled 7 metres in this fictional test. When rows are sorted, width 10 remains paired with distance 4 for C. There is a general upward tendency despite the fall from distance 5 at width 8 to distance 4 at width 10. A tendency can include variation around the overall pattern.
After F is added, there are six gliders and five distinct coordinate positions. A note indicating two observations at (10, 4), or another clearly explained frequency display, preserves the count. A missing width prevents F from providing a complete pair for this plot. Zero would describe a measured zero width, which the supplied record does not say. Do not silently substitute it for unavailable information.
With distance on the horizontal axis and width on the vertical axis, D appears at (7, 12). The full sentence still describes the same glider and measurements. The axes have changed, not the observation. If the table and graph disagree, locate the exact row before correcting anything. An overall upward shape alone cannot verify that every individual pair has been preserved accurately.
CHAPTER 21 OF 24 · Explain and continue
21. Independent practice: describe and limit claims
Set A contains (1, 9), (2, 15), (3, 14), (4, 22), (5, 27). Set B contains (1, 28), (2, 24), (3, 19), (4, 17), (5, 11). Describe the overall direction of each without claiming that direction is a moral judgement or a proof of cause. For A, explain why one small downward step does not automatically remove the general tendency.
Set C contains (1, 1), (2, 4), (3, 9), (4, 16), (5, 25). Would a constant-change straight-line description capture its structure well? Explain using the differences in y. Set D combines group P: (1, 12), (2, 11), (3, 10), and group Q: (7, 32), (8, 31), (9, 30). Compare the within-group directions with the combined visual pattern. State why a legend would matter.
For a fictional dataset spanning x from 2 to 10, a supplied line is y = 4x + 3. Estimate y at x = 6 and at x = 15. Classify each estimate by its position relative to the observed range. Explain why even the inside-range estimate is not necessarily an exact measured outcome. Include a sentence warning against treating the second estimate as guaranteed.
Finally, critique this claim: “The plot of number of library visits and reading score slopes upward, so every extra visit makes a child’s score rise by exactly five marks.” Identify at least three unsupported steps. Then write a more careful description that preserves what an upward pattern could show while leaving the size, cause and universality of an effect unresolved.
CHAPTER 22 OF 24 · Explain and continue
22. Explained answers to interpretation practice
A shows an overall positive association and B an overall negative association. In A, the value falls from 15 to 14 between x = 2 and x = 3, but the collection still generally rises across the range. Positive and negative name direction of association. They do not classify outcomes as good or bad, and neither description alone proves a causal explanation.
C curves upward with successive y increases of 3, 5, 7 and 9. A constant-change model misses that increasing rate. In D, each group slopes downward internally, while the high-x group sits much higher in y than the low-x group, making the combined collection look upward. A legend preserves the group distinction and prevents an overall summary from concealing different within-group relationships.
The supplied line gives y = 27 at x = 6 and y = 63 at x = 15. The first is interpolation because 6 lies between 2 and 10. The second is extrapolation because 15 lies outside that range. Both are model estimates. Scatter around a line means an individual observed result need not equal the fitted value, and continuation beyond the measured range requires an additional assumption that may fail.
The library claim assumes causation, an exact effect of five marks and a universal response for every child. An upward plot alone establishes none of those. It may also ignore age, previous reading skill, how visits were counted and whether scores preceded the visits. A careful revision is: “In the supplied observations, more library visits tended to accompany higher reading scores. The plot alone does not establish the cause or a guaranteed gain for an individual child.”
Begin with the record, not the trend label. Ask the learner to name one observation, read its two measurements and find its dot. Then ask for an overall description and one limitation. This sequence connects table, coordinate and explanation. A child who can name positive association but cannot interpret a single point may need help with representation before practising more terminology.
For a manageable session, choose one five-point fictional dataset and change just one feature. Add an overlapping observation, change an axis scale or mark a separate testing condition. Ask what changed in the evidence and what changed only in presentation. These controlled contrasts make the reasoning visible without requiring a long new collection of numbers every time.
Keep an error note with the decision involved: paired values, axis units, scale, overall direction, strength, evidence range or causal claim. “Mistakes with graphs” is too broad to guide the next lesson. A fresh example should test the same decision with different values. Accurate copying of a corrected plot does not yet show that the learner can explain a new one independently.
If asking a teacher or tutor for support, bring a graph, its underlying table and the child’s explanation. That combination reveals whether the difficulty concerns reading coordinates, noticing the full pattern or expressing uncertainty. The goal is a reader who can say what the dots show, trace the evidence and recognise the boundary of a claim, rather than merely produce a graph that looks like a model answer.
CHAPTER 24 OF 24 · Explain and continue
24. Family questions and the next learning step
Must every plot have a trend line? No. A line should serve a suitable question and relationship. A curve, separate groups or no clear pattern may make a single straight line misleading. If a task supplies a line, read it as a model and distinguish its estimates from measured dots. If drawing one yourself, state when it is a rough visual summary.
Can one unusual dot be ignored? Only with a defensible reason and a transparent explanation. A verified transcription error can be corrected; a genuine unusual observation may contain valuable information. Removing points because they weaken an expected conclusion changes the evidence unfairly. Keep the original record and document any correction or exclusion rule.
Does a perfect straight pattern prove causation? No. Even exact co-variation can arise from a shared factor or a constructed rule. To discuss cause, ask how the data were collected and what alternative explanations remain. This guide develops cautious interpretation rather than a complete course in study design or formal statistical inference.
What should a learner study next? If plotting is difficult, revisit coordinates and scale reading. If line estimates are confusing, revisit gradient, intercepts and equations. If explanations overclaim, practise distinguishing observation, model output and causal inference. The linked mathematics hub and reference below offer continuation. Choose the next step from the learner’s actual response, and keep original fictional examples clearly separate from claims about real families or institutions.
References and further reading
The worked scenes and practice sets are original teaching examples. These references offer formal definitions or documented usage; the connected eduKate guides provide foundations and further study.
NIST: scatter plots and association · Coordinates for children · Linear graphs for families
