Each dot records a pair of values from one observation. In the invented example, the point (3, 8) means that this observation has x-value 3 and y-value 8. It does not mean that an object has travelled from the previous point to this one. Before drawing a line, the student needs to know what relationship the two coordinates describe.

Think of each pair as a card with two measurements written on it. Plotting the cards displays how the measurements vary together across observations. Unless the question supplies an order with a relevant meaning, the order in which those cards were recorded does not create a path. A scatter cloud shows association, not automatically a connected route.

Joining (1, 4) to (2, 4) and then to (3, 8) creates a horizontal segment followed by a rising segment. Those segments introduce values between the observations according to that particular connection rule. The raw data alone do not say that the true relationship stays exactly flat and then suddenly rises in those places.

A line of best fit, when a linear trend is appropriate, serves a different purpose. It summarises the overall relationship rather than reproducing every observed value. The dots remain visible as observations, and their departures from the line remain part of the information. A good explanation distinguishes those two roles instead of pretending that the fitted line is the original data.

Ask your child to finish the sentence, ‘This point represents …’ using the two variables. If they describe a journey, revisit the context. If they describe paired measurements correctly but still join all dots, focus on the line’s purpose. Understanding the point comes first, because an attractive graph can still communicate a relationship that the observations do not establish.

The same six paired observations can be written in different orders without changing the scatter diagram. Plotting (6, 12) first and (1, 4) last places them at exactly the same coordinates as before. The overall cloud is unchanged. But joining points in recording order would create a different path, showing that the path depends on an arbitrary choice.

Sorting by x-value avoids a tangled path, but it still does not turn dot-to-dot segments into a straight-line fit. In the main example, sorted joins form alternating flat and rising pieces. That follows the six observations closely, while the guide y = 2x + 1 summarises their common upward tendency. These are different constructions with different interpretations.

A time-series graph may deliberately join consecutive values to display change over time. That is a separate graph job, supported by the time ordering and the task. A scatter diagram can also use time as one variable, but the instruction still matters: plotting paired observations and fitting a trend is not automatically the same as connecting a recorded sequence.

Do not teach ‘never join points’ as a universal rule for every graph. Function graphs, cumulative-frequency curves and specified time plots can require connections or curves. The repair is to identify the graph type and requested representation. A scatter question about a line of best fit should not inherit a joining habit merely because another exercise used it.

One useful check is to shuffle a short list of paired cards and ask whether the plotted cloud should change. It should not, because every pair is preserved. Then ask what would happen to a line joining cards in the new order. This contrast makes the difference between data positions and an invented connection visible without adding difficult arithmetic.

Keep each x-value attached to its corresponding y-value. The six invented pairs are (1, 4), (2, 4), (3, 8), (4, 8), (5, 12) and (6, 12). Sorting one column independently would create new pairs and therefore different observations. A scatter diagram is built from pairings, not from two separate lists that can be rearranged freely.

The table below preserves the pairs and records the value of our illustrative guide at each x. The guide column is not an observed measurement; it is a model value. Showing both helps the student distinguish plotting the original dots from reading a fitted relationship. Only the observed y-values supply the six scatter points.

Label both axes with the variables and any supplied units. Choose sensible scales and read their increments before plotting. A position three small squares across is not necessarily x = 3 if the scale uses two units per square. The graph should communicate actual values, not only a pattern of marks with unlabelled distances.

Plot one pair at a time, locating x horizontally and y vertically unless the question specifies otherwise. Mark the point clearly without making dots so large that small differences disappear. Check repeated y-values: points at x = 1 and x = 2 both have y = 4, but they are still two distinct paired observations.

Finally, compare the plotted positions with the original table. A correct-looking positive trend does not prove that every point is accurate. One swapped coordinate can still leave a cloud that appears plausible. The plotting check should preserve each observation before a fitted line or estimate is added, because later reasoning depends on the actual locations.

A fitted line describes a model for the overall relationship. Observations can vary around that model because of measurement differences, other influences or the fact that a simple line only approximates a more complex pattern. Passing through every point is therefore not the defining job of a straight line of best fit.

In the main example, y = 2x + 1 gives fitted values 3, 5, 7, 9, 11 and 13. The observed values are 4, 4, 8, 8, 12 and 12. None lies exactly on this guide, yet the departures are small and alternate above and below. The line captures the overall positive relationship without claiming exact agreement.

A student may worry that missing a dot makes the line inaccurate. Ask whether the exercise seeks an exact curve through known mathematical coordinates or a summary of scattered measurements. Those tasks are different. In a scatter diagram, the distance between a dot and the line is information about how that observation compares with the trend.

Joining all dots can conceal this model-versus-observation distinction. It makes each recorded value part of the constructed path, which may suggest certainty between neighbouring observations that the data do not justify. A straight guide keeps the difference visible: observations remain dots, while the fitted relationship remains a separate line.

Use the word estimate when reading from the fitted line. At x = 3.5, the guide gives y = 8, but there is no recorded observation at x = 3.5 in our table. The model supplies the estimate. This wording helps the student avoid presenting a prediction as if it were another measured data point.

For a roughly linear scatter cloud, a reasonable hand-drawn line should follow the overall direction and pass through the central region of the observations. It should not chase individual dots. Look at the whole cloud before placing a ruler. A line that hugs one extreme edge may systematically miss the central relationship even if it passes through several points.

Having points on both sides is a useful visual check, but equal counts alone do not guarantee a good line. Three points just above and three very far below can indicate poor placement. Consider both the number and the size of departures across the range. The informal aim is a sensible central summary, not a rigid rule that decides the line by counting only.

Our illustrative guide y = 2x + 1 leaves vertical departures +1, −1, +1, −1, +1 and −1. It follows the cloud and balances those departures simply. It is useful for teaching the difference from dot-to-dot joins. We are not claiming that every acceptable hand-drawn answer must match its exact equation.

Formal statistical fitting methods choose lines using specified criteria, and different methods can produce different models. An ordinary hand-drawn exercise does not automatically require calculating such a fit. Follow the method requested and explain what was done. Do not label a visually selected guide as a computed least-squares result.

Ask your child to explain the placement before calculating an estimate. A response such as ‘the line follows the upward cloud through its middle, with observations nearby on both sides’ is more informative than ‘I joined the first and last dots’. The explanation makes the choice assessable and shows whether the student considered all observations.

The first and last observations may lie above or below the central trend. A line through them is influenced by those particular departures, not just by the overall cloud. In the main data, a line joining (1, 4) to (6, 12) has gradient 8/5 = 1.6, whereas our illustrative central guide has gradient 2.

The endpoint line has equation y = 1.6x + 2.4. At x = 3.5 it also gives 8, so this one estimate does not expose the difference. At x = 2, however, it gives 5.6 while the illustrative guide gives 5. Agreement at one central input does not make the lines identical or prove that the endpoint method is generally sound.

Do not respond by teaching another shortcut such as ‘always ignore the endpoints’. They are genuine observations and should normally remain in the cloud. The problem is giving two chosen observations automatic authority over all the others. A fitted line should reflect the pattern and the fitting method, not a rule that privileges points simply because their x-values are extreme.

Sometimes a suitable line happens to pass through an endpoint, and sometimes it does not. The position alone is not decisive. Check how the line relates to the rest of the observations. If it creates large systematic misses across the middle, revisit its placement rather than insisting that the endpoint must remain on the line.

A useful parent prompt is, ‘What did the other four points contribute to this line?’ If the child selected the first and last automatically, the answer reveals the shortcut. Invite them to inspect the whole cloud, explain a central line and compare the resulting estimates. The repair concerns model selection, not merely calculating a different gradient.

A positive association means that larger x-values tend to accompany larger y-values in the observed cloud. A negative association means larger x-values tend to accompany smaller y-values. These describe tendencies across observations, not guarantees about every pair. The main data show a positive pattern even though some consecutive y-values are equal.

For a contrasting negative example, use x-values 1 through 6 with y-values 14, 14, 10, 10, 6 and 6. The guide y = −2x + 17 runs centrally through that downward cloud. Its estimates decrease as x increases. The negative gradient reflects the direction, while departures from the line remain visible as scatter.

An unclear linear pattern should not be forced into a confident upward or downward line. For example, the six y-values 4, 10, 7, 7, 10 and 4 at x = 1 through 6 are symmetric around their middle and do not show the same straight-line trend. A straight fit may provide little useful summary of such a pattern.

No clear linear association does not establish that the variables are independent or unrelated in every possible way. A curved relationship can be present without a strong straight-line trend. Inspect the shape and the context. Avoid a conclusion stronger than the graph supports merely because the plotted dots do not form a simple rising band.

Ask for direction and form separately: ‘Is there a clear tendency, and does it look roughly linear?’ This prevents a student from treating any visible pattern as a reason to draw a straight best-fit line. The question’s instructions still guide the task, but understanding the cloud helps the child interpret the result responsibly.

Two clouds can have similar upward direction but different amounts of scatter. Points tightly clustered around a line support more precise estimates within that simple model than widely dispersed points do, assuming the data and context are otherwise comparable. Direction and strength are separate features. A steep line is not automatically a strong association.

In the main example, every observation lies one vertical y-unit from the illustrative guide. That is a deliberately tidy teaching cloud. Real or supplied data may have uneven departures. Look across the full range rather than judging fit from one convenient cluster. A line can appear satisfactory near one x-value and perform poorly elsewhere.

Compare a hypothetical point at x = 3 with y = 8 to one at x = 3 with y = 18. Both may lie above an upward guide, but their departures have different sizes. The distance matters when assessing how closely the line summarises the observations. Counting them both as simply ‘above’ loses that difference.

Graph scales can affect how steep or spread out a cloud looks on the page. Changing the vertical scale stretches or compresses visual distances without changing the underlying measured values. Read the axes before using a visual impression. A compressed y-axis can make substantial variation look small, and a stretched one can make small differences look dramatic.

For an informal school-level estimate, state the model result without pretending that the actual outcome must equal it. A tutor can ask the student to describe whether the surrounding points are close or dispersed. That observation gives meaning to the word estimate and connects the plotted cloud to the confidence one should place in a prediction.

To estimate y for a given x, locate the x-value on the horizontal axis, move to the fitted line and then read horizontally to the y-axis. The estimate comes from the line, not from whichever dot happens to be nearest. Make the guide lines light enough that the original observations and fitted line remain readable.

With the illustrative equation y = 2x + 1, x = 3.5 gives y = 2(3.5) + 1 = 8. This input lies between the observed x-values 1 and 6. The calculation explains the graphical reading, but a question that asks for a graph estimate may expect the line to be drawn and used visibly.

If the fitted line differs slightly from another reasonable hand-drawn line, estimates can also differ slightly. Do not promise an exact universal tolerance without a supplied marking instruction. Explain the method and read the scale carefully. A parent can check whether the estimate is consistent with the child’s own line before comparing it with a printed solution.

An observation at x = 3 has y = 8 in our data, while the guide estimates y = 7 there. Neither value should silently replace the other. One is a recorded measurement; the other is the model’s prediction at the same input. A good answer identifies which quantity the question requests rather than choosing the value that looks tidiest.

Before giving the final number, state ‘estimated y when x is …’ and include any units supplied for y. This preserves the direction of the request. It also creates a useful check against reading the axes backwards or reporting an observed coordinate when the task asked for a value on the fitted trend.

A graph exercise may ask for the x-value corresponding to a given y on the drawn guide. Start from the y-axis, move horizontally to that line and then vertically to the x-axis. With y = 2x + 1 and y = 9, the line gives x = 4. This is a graphical inverse reading of the chosen relationship.

Algebraically, rearrange y = 2x + 1 to x = (y − 1)/2. Substituting y = 9 gives (9 − 1)/2 = 4. A common slip is to use the original expression 2y + 1, which treats the supplied y as though it were x. Naming the known and requested variable before calculating helps prevent that reversal.

An inverse reading from a hand-drawn line is not automatically the same as a formally fitted statistical model for predicting x from y. Different statistical fitting criteria can produce different regression relationships. In this tutorial, we are reading the stated illustrative guide, not claiming that reversing a fitted equation is always the optimal prediction method.

Keep the input within a sensible part of the drawn range and context. If the line is nearly horizontal, a small change in y can produce a large or poorly determined x reading. If the relationship is not one-to-one in a curved model, more than one x may correspond to a y. The graph’s form matters.

Ask the child to mark the starting axis and destination axis with words. This small step can reveal an interpretation error before arithmetic begins. If the direction is correct but the result is wrong, repair the rearrangement or scale reading. Do not treat every failed inverse estimate as evidence that the whole scatter-diagram concept is insecure.

Interpolation uses the fitted model within the observed input range. Our x-values run from 1 to 6, so estimating at x = 3.5 is interpolation. Extrapolation extends the model beyond that range, such as predicting at x = 10. The same formula can produce a number, but the observations do not directly show how the relationship behaves that far away.

The guide y = 2x + 1 gives 21 at x = 10. That arithmetic is correct for the guide. It does not establish that a real measured quantity would equal 21, or even continue increasing linearly there. Other constraints or changing relationships may intervene. Separate calculation under a model from evidence that the model remains suitable.

Do not say that every extrapolation is automatically wrong. A question may deliberately request one and ask about its reliability. Give the model estimate and explain the limitation using the observed range. The useful judgement is how far the input lies beyond the data and whether the context supports continuing the trend.

Similarly, interpolation is not a guarantee of accuracy. A poor fit, sparse observations or substantial scatter can weaken estimates inside the range. It is usually less speculative than distant extrapolation for the same model, but it still relies on assumptions. State the estimate as a model-based reading rather than a certain outcome.

Ask your child to identify the smallest and largest observed x-values before estimating. Then classify the requested x as inside or outside that interval. This simple range check gives the reliability discussion a concrete basis. It is more useful than attaching the word reliable to every answer that happens to fall neatly on the ruler-drawn line.

An unusual point lies away from the main pattern, but that alone does not prove it is a mistake. It might be a recording error, a different situation or a genuine observation that the simple model does not explain well. Identify it and consider the context before deciding what to do. Do not erase inconvenient data merely to make a tidy line.

Suppose an additional invented observation appears at (3.5, 25) beside the main cloud. It sits far above the illustrative guide’s value 8 at that x. The departure is noticeable and should be described. But the graph alone does not tell us whether 25 was copied incorrectly or represents a different underlying process.

If the question explicitly instructs the student to ignore a stated erroneous observation, follow that instruction and explain the exclusion. If no such information is given, preserve the point and discuss its possible influence. A teacher or tutor can clarify the intended fitting approach when the exercise leaves that choice open.

An unusual point can affect the position of a fitted line, especially under a formal criterion that weighs large departures strongly. A hand-drawn line should not simply bend to pass through it while abandoning the rest of the cloud. Inspect both the overall pattern and the point’s location, and explain how the selected guide reflects them.

Ask, ‘What evidence tells us this point is wrong?’ If the answer is only ‘it does not fit’, the conclusion is too strong. This question builds a useful distinction between noticing an observation and rejecting it. The scatter diagram should help the child reason about variation, not teach them to delete every measurement that resists a preferred pattern.

A scatter diagram can show that two measured quantities vary together across observations. It does not by itself establish that changing one causes the other. A shared influence, reverse direction or the way observations were selected may help explain the pattern. Keep the conclusion tied to association unless additional evidence supports a causal claim.

For an invented example, a positive relation between two scores could reflect both being influenced by preparation, prior knowledge or another factor. It would be too strong to claim that increasing the first score directly makes the second score rise by the fitted gradient. The line describes observed association within the supplied data, not an intervention.

Likewise, a fitted estimate does not promise an individual outcome. A child reading y = 2x + 1 should understand that it is a simple model for the chosen cloud. The observed departures already show that exact prediction is not the model’s role. Distinguishing correlation from causation also helps prevent overconfident interpretation of a tidy-looking line.

Use language such as ‘larger x-values tend to accompany larger y-values in these observations’. Avoid replacing it with ‘x makes y increase’ unless the task supplies evidence for that claim. The first statement is precise and useful. It says what the graph shows without inventing a mechanism that was never measured.

This chapter is a reading check rather than an invitation to analyse a full research study during every Mathematics exercise. Ask the child to state the conclusion at the level supported by the graph. A tutor can then explain any additional context the question provides, keeping the model, observations and causal claims distinct.

The gradient of a straight guide is change in y divided by change in x. If y is measured in units Y and x in units X, the gradient has units Y per X. It is not merely a number of graph squares. Read values from the axes and use points on the drawn line to calculate its rate of change.

For y = 2x + 1, points (1, 3) and (6, 13) lie on the guide. Their gradient is (13 − 3)/(6 − 1) = 10/5 = 2. These coordinates are model points; neither needs to be an original observation. Using two well-separated points on the line can reduce reading error compared with choosing adjacent tiny sections.

The rise and run must be measured in coordinate values. If one horizontal square represents two x-units and one vertical square represents five y-units, counting squares alone gives a different ratio from the numerical gradient. This is the same scale-reading habit needed when plotting the original paired data.

The intercept 1 belongs to y when x = 0 on the model. If zero lies outside the observed range, that intercept may lack a direct practical interpretation. Do not invent a real starting value from a line extended beyond the data. It is a parameter of the guide unless the context makes its meaning clear.

Ask your child to identify whether they are calculating the gradient of the fitted line or of a segment joining two observed points. Both calculations can be performed, but they describe different objects. A correct rise-over-run procedure does not fix a wrongly selected pair of points, so the chosen line must be named first.

A roughly curved cloud should not automatically be summarised by a straight line as though its direction were constant. Inspect the shape before fitting. If the question asks for a suitable curve, follow that instruction. If it specifically asks for a straight line, carry out the requested construction while understanding that a straight model may miss systematic curvature.

A curve drawn through the central pattern is still different from joining every point in a jagged path. The model should reflect the overall form, while individual departures remain visible. Do not assume that the only alternatives are a straight line or exact dot-to-dot connections. The task and observed relationship determine the appropriate representation.

Repeatedly bending a guide to hit every observation can overfit the particular data rather than summarise a useful relationship. Conversely, forcing a ruler line through a strongly curved pattern can obscure how the relationship changes. At this level, the practical check is whether the chosen representation serves the graph question and reflects the cloud’s form.

An unclear straight-line pattern may therefore still contain structure. For a simple conceptual illustration, observations arranged around a U-shape can have low linear association while showing a clear non-linear tendency. Avoid concluding that ‘no strong straight-line trend’ means ‘nothing is happening’. Describe the form that the diagram actually supports.

Keep extensions proportionate to your child’s current exercise. This article’s main repair concerns an appropriate straight-line best-fit task. Curvature is a boundary check that prevents turning the repair into another universal instruction. A tutor should explain why a straight line is being used here and how a changed pattern could require a changed model.

A numerical comparison can clarify what two candidate guides predict, even when the question asks for a hand-drawn line. For our illustrative central guide y = 2x + 1, the six departures observed minus predicted are +1, −1, +1, −1, +1 and −1. This gives an organised description of how the line sits relative to the data.

The endpoint guide y = 1.6x + 2.4 predicts 4, 5.6, 7.2, 8.8, 10.4 and 12. Its departures are 0, −1.6, +0.8, −0.8, +1.6 and 0. It passes through both endpoints but differs elsewhere. The comparison shows why endpoint passage alone is not a sufficient fitting principle.

The squared departures for the illustrative guide sum to 6. For the endpoint guide they sum to 6.4. Under that particular comparison, the illustrative guide has the smaller total. This is a teaching check of two proposed lines, not a proof that either is the unique least-squares optimum among every possible line.

If a formal fitting method is requested, use its actual criterion and procedure rather than choosing between two convenient candidates. Our simple guide was selected for transparent illustration and central placement. Avoid presenting a small comparison as a complete statistical optimisation or adding advanced calculations that the child’s current task does not require.

A tutor can use this chapter to connect visual reasoning to numerical evidence when helpful. The student should still begin by identifying the purpose of the line and preserving paired observations. Extra arithmetic is useful only when it resolves a real uncertainty. It should not conceal the basic distinction between fitting a relationship and connecting a sequence.

A small-group 3-pax Secondary 4 Mathematics tutorial should ask each student what a point represents before checking the drawn line. Give a short paired-data table and ask for a scatter diagram without connections. Then ask what the requested best-fit line is meant to show. The explanation distinguishes a graph-purpose misunderstanding from an inaccurate ruler placement.

Next check pair preservation, axes and scales. A child may understand the idea of a fitted line but plot coordinates wrongly, swap x and y or sort one column independently. Those errors need a plotting repair. A child who plots all points correctly and then joins them automatically needs a representation repair instead.

Use a controlled contrast between a scatter task and a specified time-sequence task. Ask why a connection can be meaningful in one and inappropriate as a best-fit construction in the other. This avoids teaching an absolute ‘never join dots’ rule that will fail on the next graph type. The student’s reason should refer to the task and data.

In a group of three, each learner should draw and explain a guide independently before comparing choices. One child can describe central placement, another can read an estimate, and the third can check its observed range. The tutor must still inspect everyone’s work. A confident group’s agreement does not show that every member can select the representation alone.

Finish with a new cloud or changed request, such as estimating y at an input inside the range. Improvement means the child distinguishes observations from fitted values and explains the line without being reminded not to join dots. Record that specific evidence and the next gap. A calm lesson can make progress visible without promising guaranteed results.

Start with ‘What does each dot represent?’ rather than ‘Why did you join them?’ Let your child explain the paired values and whether the task supplies a meaningful sequence. This makes the graph’s purpose visible before discussing the mistake. A neutral question is more useful than describing the whole effort as careless.

Then ask, ‘Are we trying to visit every observation, or summarise the overall relationship?’ For a straight-line best-fit question, invite the child to inspect the cloud’s middle and direction. Let them propose a line and explain its placement. Drawing it yourself can correct the page while leaving the selection decision hidden.

Use the six invented pairs for a short comparison if the current worksheet is crowded. A dot-to-dot path has flat and rising pieces, while the illustrative central guide is straight. Ask what each construction communicates. Keep the numbers simple so the child can focus on the distinction rather than struggle with calculation at the same time.

If tiredness makes the discussion unproductive, keep the original graph and note the exact issue for the teacher or tutor: ‘paired points plotted correctly; connected them in x order for a best-fit task’. That description preserves a useful diagnosis. Parent support can be brief and still help the professional choose the next repair.

Look for independent transfer later with another scatter cloud. Ask for the line’s purpose, an estimate and a range check. Acknowledge the specific reasoning now shown rather than saying the whole statistics topic is mastered. The aim is a dependable graph decision that your child can repeat when the context and numbers change.

Try these before reading the answers. Question 1: for the six main pairs, explain why joining them in recording order does not automatically produce a best-fit line. Question 2: use the illustrative guide y = 2x + 1 to estimate y when x = 4.5. State whether the input is inside the observed x-range.

Question 3: read x from the same guide when y = 9. Question 4: use points (1, 3) and (6, 13) on the guide to find its gradient. Explain why these need not be observed dots. Question 5: the guide gives y = 21 at x = 10. Identify the type of estimate and state one reason for caution.

Question 6: a student says that a positive scatter trend proves x causes y to rise and that the fitted line must pass through every observation. Correct both claims in plain language. In each response, identify whether you are discussing raw observations, the fitted model or a conclusion about the underlying situation.

Answers to Questions 1–3: recording order can change while the paired observations stay the same, so a joining path is not the same as an overall fit. At x = 4.5, y = 10, and the input lies within 1 to 6. When y = 9, x = (9 − 1)/2 = 4. Both readings refer to the illustrative guide.

Answers to Questions 4–6: gradient = (13 − 3)/(6 − 1) = 2; choose points on the drawn line to measure that line. The estimate at x = 10 is extrapolation beyond the observed range. A positive association alone does not prove causation, and a fitted line summarises the cloud rather than being required to pass through every dot.

Read the graph request and identify the variables. Decide whether the task is plotting paired observations, connecting a supplied sequence, drawing a fitted line or curve, or interpreting an existing model. The presence of dots alone does not decide the construction. Keep the graph’s purpose attached to the words of the question.

Preserve the pairs, label axes and units, and read the scales before plotting. Check the individual points against the supplied table. Then inspect the cloud’s direction, form and spread. If an appropriate straight-line fit is requested, draw a reasonable central guide rather than connecting observations automatically or choosing endpoints without considering the rest.

Keep observations and model values distinct. When estimating, start from the requested input axis and read the fitted relationship, stating the result as an estimate with its units. If finding a gradient, use well-separated points on the drawn line and calculate changes in coordinate values, not only in graph squares.

Check whether the requested input lies inside the observed range. Explain extrapolation limits and avoid treating a fitted number as a certain outcome. Notice unusual points without automatically deleting them, and describe correlation without inventing causation. These checks keep the interpretation proportionate to what the supplied observations and model actually show.

For parents considering Secondary 4 Mathematics tuition in Punggol, a useful checkpoint is, ‘Can my child explain why this line summarises the cloud instead of joining every dot?’ If yes, move to a changed estimate or graph type. If not, keep the repair focused on paired observations and representation. A clear graph decision supports more confident and independent Mathematics work.