A graph can look convincing and still be the wrong model
Four counts are written on a noticeboard: 10, 20, 30, 40. A student calls the pattern exponential because the numbers are rising. Another student says that any rising sequence is a straight line. The first is wrong for these values at equally spaced times; the second is wrong as a general rule. The counts increase by a constant ten, so a linear model describes this particular table. Replace them with 10, 20, 40, 80 and repeated multiplication, rather than constant addition, becomes the important feature. The pattern changes even though both stories say attendance is increasing.
In G3 Mathematics, the goal is not to memorise three silhouettes and guess. It is to read a relationship between variables, identify evidence for a possible function, check the mathematical rule and explain where a real-world interpretation stops being justified. A clean sketch and a convincing argument grow from the same foundation: you know what the axes mean and what changes when the input advances.
This volume continues Vol 0226 on exponential growth graphs but gives it a new task. There, we learned the repeated-factor rule and its limits. Here, we must choose among linear, quadratic and exponential descriptions, reject a model when the evidence conflicts with it, and recognise when several rules fit a small number of observations. This is model selection, not another chapter of one graph type.
What the official G3 SEC Mathematics syllabus requires
The 2027 Singapore-Cambridge SEC G3 Mathematics subject code is K310, with the previous O-Level reference code 4052. The official syllabus covers linear functions of the form y = ax + b, quadratic functions of the form y = ax² + bx + c, the properties of quadratic graphs, and exponential graphs of the form y = k a^x where a is a positive integer. It also covers estimating the gradient of a curve by drawing a tangent. A student must learn what the syllabus actually defines rather than assuming every forecasting formula found online is examinable.
There are two written K310 papers. Paper 1 lasts 2 hours 15 minutes, has 90 marks, carries 50% and has about 26 short-answer questions. Paper 2 also lasts 2 hours 15 minutes, has 90 marks and carries 50%, with nine to ten questions; the final question focuses on mathematics in a real-world scenario. Essential working matters. Unless instructed otherwise, non-exact answers are generally given to three significant figures, or angles to one decimal place. See the official subject directory and the K310 2027 Mathematics syllabus.
The examples below are original study exercises, not SEAB past-paper questions. Their purpose is to make one decision habit visible. Mathematics does not gain credibility merely because a formula reproduces one or two data points. You must also be able to say why you selected it, what information you used, and what conclusions the context allows.
The three patterns students must be able to distinguish
Linear: when x moves by equal steps and y changes by equal differences, a linear rule may fit. For x = 0, 1, 2, 3 and y = 4, 7, 10, 13, the difference is consistently +3. One possible rule is y = 3x + 4. Its gradient, three units of y per one unit of x, is constant. The y-intercept is four, not three. Learn to separate rate of change from starting value.
Quadratic: for equally spaced x-values, constant non-zero second differences are a useful signal of a quadratic pattern. If y = 2, 5, 10, 17 at x = 0, 1, 2, 3, the first differences are +3, +5, +7; the second differences are +2, +2. The rule y = x² + 2x + 2 fits all four values. Its graph has curvature and, depending on the coefficient of x², a minimum or maximum point. Do not mistake a rising section of a parabola for a straight line merely because all its y-values increase.
Exponential: equal steps in x may produce equal ratios of consecutive non-zero y-values. For 3, 6, 12, 24, each term is twice the previous one. The rule y = 3 × 2^x fits x = 0, 1, 2, 3. Exponential behaviour is multiplicative, not merely fast growth. A sequence that grows by 10, 20, 30 and 40 is not automatically exponential simply because the increments themselves increase. Examine the ratios and the candidate equation.
The diagnostic before a graph sketch
- Confirm that input values are equally spaced before using simple first- and second-difference tests.
- Write the first differences, then second differences if needed, and compare non-zero consecutive ratios.
- Check the first given point, often x = 0, to identify a possible intercept or initial factor.
- Use one additional point to test a proposed formula. Two fitted points alone rarely distinguish graph families.
- Check whether the equation’s output has the correct real-world unit and whether negative or fractional values make sense in context.
- State the model’s domain and evidence limits; mathematical extension is not necessarily a physical prediction.
For a small finite data table, several formulas may agree with all recorded points. Diagnostics are useful recognition tools, not a proof of the process that generated the measurements. A table with only x = 0 and x = 1 cannot uniquely distinguish a straight line from a suitably chosen curve. In an examination, use the specified functional form when one is given and the available evidence when a choice is requested. In a real investigation, seek more observations rather than pretending the pattern is inevitable.
Thirty-four worked graph-family clinics
Worked clinic 01: Constant addition
Given: At x = 0,1,2,3, y is 8,13,18,23. Mathematical judgement: The differences are +5 each time; y = 5x + 8 fits.
Examination check: A higher y-value does not mean the rate is increasing. Here the gradient remains five. Verify x = 3 gives 23. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 02: Constant doubling
Given: The table at consecutive days is 4,8,16,32. Mathematical judgement: Each value is twice the previous one; y = 4 × 2^x is a candidate.
Examination check: Differences +4,+8,+16 are not constant. Do not force a linear gradient onto this table. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 03: Quadratic second differences
Given: For x = 0,1,2,3, y is 2,5,10,17. Mathematical judgement: Differences +3,+5,+7 and second differences +2,+2; y = x² + 2x + 2 fits.
Examination check: The increasing first differences matter. Curvature is a feature, not an error to be averaged away. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 04: A descending line
Given: Read y values 16,13,10,7 at x = 0,1,2,3. Mathematical judgement: The change is −3 each time; y = 16 − 3x.
Examination check: A negative gradient describes a decrease. It does not make the model invalid or ‘negative data’. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 05: A downward parabola
Given: Read y = 10,15,18,19 at x = 0,1,2,3. Mathematical judgement: First differences +5,+3,+1 give second difference −2; y = −x² + 6x + 10 fits.
Examination check: The values initially rise, but the rise slows. A downward-opening quadratic can grow over part of its domain. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 06: Multiplying by three
Given: For x = 0,1,2,3, y is 5,15,45,135. Mathematical judgement: The ratio is three, giving y = 5 × 3^x.
Examination check: The increase from one step to the next changes dramatically, but a constant ratio supplies a compact rule. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 07: Unequal x gaps
Given: Points are (0,2), (1,4), (3,16). Mathematical judgement: The x-step from 1 to 3 is two, so a ratio of four is consistent with 2^2; y = 2 × 2^x fits.
Examination check: Do not compare successive ratios without accounting for the size of the input step. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 08: Gaps of two time units
Given: Points are (0,3), (2,12), (4,48). Mathematical judgement: Each two-unit step multiplies y by four; the per-unit factor is two, so y = 3 × 2^x fits.
Examination check: A two-unit ratio of four is not automatically a one-unit factor of four. Track the input interval. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 09: A two-point ambiguity
Given: Only the points (0,2) and (1,5) are known. Mathematical judgement: Both y = 3x + 2 and y = 2 × (2.5)^x fit these two points as numerical models.
Examination check: Never call the family uniquely established from two values. More observations or an explicit form are needed. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 10: A quadratic turning point
Given: The function is y = (x − 2)² − 1. Mathematical judgement: Its minimum is −1 at x = 2; it opens upward and is symmetric about x = 2.
Examination check: The turning point is not necessarily the y-intercept. At x = 0, y is 3. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 11: A negative quadratic coefficient
Given: Consider y = −(x − 3)² + 9. Mathematical judgement: The maximum is nine at x = 3; the graph opens downward.
Examination check: The sign controls orientation. Do not say a negative coefficient makes every output negative. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 12: Intercept confusion
Given: For y = 4x + 7, one student calls four the starting value. Mathematical judgement: At x = 0, y = 7, so seven is the intercept and four is the gradient.
Examination check: Name the role of each coefficient rather than memorising its position without meaning. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 13: Initial exponential value
Given: For y = 6 × 3^x, find y when x = 0. Mathematical judgement: Because 3^0 = 1, y = 6. Six is the initial multiplier k.
Examination check: The base three is the repeated factor, not the starting value. Check these roles separately. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 14: The zero-pattern trap
Given: All displayed y-values are zero. Mathematical judgement: A zero line fits the recorded values, but it does not uniquely identify the generating process.
Examination check: Ratios of successive zero values are undefined, so the usual exponential-ratio test cannot be used. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 15: First differences that change
Given: A candidate sees 1,4,9,16. Mathematical judgement: Differences +3,+5,+7; second differences +2. The pattern is consistent with y = (x+1)² for x = 0,1,2,3.
Examination check: The sequence grows increasingly quickly but need not be exponential. Compare ratios as well as differences. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 16: A rapidly rising line
Given: A rate-of-pay table is 100,1100,2100,3100 at equal hourly steps. Mathematical judgement: The differences are each 1000, so a linear model fits despite the dramatic jump.
Examination check: A large fixed increment is still addition. Visual steepness alone is not a classification rule. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 17: A slowly growing exponential
Given: A model gives 100,110,121,133.1 for equal steps. Mathematical judgement: Each term is 1.1 times the previous one, a multiplicative pattern outside the syllabus’s simple positive-integer-base display form.
Examination check: Use the example to understand rates, but distinguish wider mathematical modelling from the specified K310 graph family y = k a^x with positive-integer a. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 18: A short run can deceive
Given: The table is 2,3,5 at x = 0,1,2. Mathematical judgement: Differences +1,+2 suggest a quadratic interpolation, but three points cannot prove the real process follows that rule forever.
Examination check: Finite data invite models; they do not establish unrestricted forecasts. A fourth observation may challenge the model. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 19: Check the proposed quadratic
Given: A student claims y = x² + x + 5 fits 5,7,11,17. Mathematical judgement: Substitution gives 5,7,11,17 for x = 0,1,2,3, so it fits those points.
Examination check: Checking all given points matters more than recognising a familiar curve shape without testing it. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 20: Wrong base from first pair
Given: A table begins 2,6,18,50. Mathematical judgement: The first two ratios suggest three, but 50 is not 54; y = 2 × 3^x fails at the fourth point.
Examination check: One successful ratio is not sufficient. Check later values, including data that may contain error or a change in process. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 21: Change of units
Given: A line’s slope is 3 litres per minute when time is measured in minutes. Mathematical judgement: If time is measured in seconds, the same physical rate is 0.05 litres per second.
Examination check: The graph’s numerical gradient depends on axis units. Give the unit as a ratio rather than a bare number. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 22: A decreasing quadratic section
Given: For y = (x − 4)², consider x = 0,1,2,3,4. Mathematical judgement: Outputs 16,9,4,1,0 decrease toward the minimum at x = 4.
Examination check: A quadratic can descend throughout the displayed window. Classification requires its rule or sufficient pattern, not merely the sign of change. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 23: Intercept outside the observed domain
Given: A school counts registrations from week 2 rather than week 0. Mathematical judgement: A line extrapolated to x = 0 may have a mathematical intercept, but that value was not observed.
Examination check: When explaining a real context, say whether the intercept is measured, calculated or extrapolated. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 24: A fixed ceiling
Given: A room holds at most 60 people, yet a formula predicts 80 registrations. Mathematical judgement: Eighty may be a mathematical forecast for demand but cannot be interpreted as eighty seated attendees.
Examination check: Models and capacities describe different quantities. State the constraint rather than declaring the arithmetic false. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 25: Comparing two fits
Given: A line and parabola both pass through two observed points. Mathematical judgement: Those points alone do not justify selecting one model as the true process.
Examination check: Use later data, a supplied functional form or sound contextual information to make the choice. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 26: The meaning of a tangent
Given: A curve shows cumulative visitors against time; a tangent rises 15 visitors in 3 minutes. Mathematical judgement: The tangent gradient estimates the local rate as 5 visitors per minute at that point.
Examination check: It is an instantaneous or local estimate, not automatically the average rate across the entire event. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 27: Equal ratio with negative values
Given: Values are −2,−4,−8 at equal x steps. Mathematical judgement: The ratio of two remains constant but the sign and context require care; y = −2 × 2^x fits.
Examination check: Do not automatically call the measured quantity impossible. Interpret whether negative y-values make sense for the specified variable. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 28: Time intervals are not equal
Given: Three readings are taken at 9 a.m., 10 a.m. and 1 p.m. Mathematical judgement: The successive time steps are one hour and three hours; ordinary term-to-term difference tests cannot be applied blindly.
Examination check: Convert the time information into x-values first. Equal row spacing on paper does not mean equal input spacing. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 29: The axis begins at 100
Given: A chart plots attendance only from 100 to 150 on the vertical scale. Mathematical judgement: A line can look visually dramatic when the axis is truncated; calculations still depend on actual coordinates.
Examination check: Check the axis origin and numerical scale rather than judging graph family or size of change by appearance. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 30: A minimum and two roots
Given: For y = (x − 1)(x − 3), find the x-intercepts. Mathematical judgement: The roots are x = 1 and x = 3; the axis of symmetry is x = 2 and the minimum y-value is −1.
Examination check: A quadratic’s intercepts and turning point provide related but different evidence. Sketch both. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 31: A growth model with three datapoints
Given: Observed participation is 6,12,24 over three equal periods. Mathematical judgement: The ratios are two and two, so y = 6 × 2^x fits the available points.
Examination check: Say the observations are consistent with doubling; do not say doubling is certain next year. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 32: A capacity-adjusted report
Given: A sequence predicts 8,16,32,64 applicants but the workshop has 40 seats. Mathematical judgement: Keep the 64 as the modelled number of applicants, not confirmed participants.
Examination check: The mathematical relation may describe expressed demand while practical acceptance is capped. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 33: Checking a claim after conversion
Given: A study compares two weeks: 4,8,12 minutes versus 240,480,720 seconds. Mathematical judgement: Both are linear sequences with the same physical changes expressed in different units.
Examination check: A shift in numerical scale changes the numbers but not the underlying relationship. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 34: A wrong second difference
Given: A student gets first differences 2,4,8,16 and calls second differences constant. Mathematical judgement: The second differences are 2,4,8, not constant; this table is more consistent with a multiplicative pattern.
Examination check: Write the actual difference rows. Do not describe them from intuition, because the second-difference test is numerical. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 35: Negative x in exponentials
Given: For y = 4 × 2^x, compute x = −1. Mathematical judgement: The result is 2, because 2^(−1) is one half.
Examination check: A negative x does not mean y becomes negative. This checks the rule beyond nonnegative integer inputs. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 36: Quadratic exact root check
Given: For y = x² − 5x + 6, determine when y = 0. Mathematical judgement: Factor as (x − 2)(x − 3). The zeros are x = 2 and x = 3.
Examination check: If the question is about graph intersections, roots give intercepts. They are not automatically solutions to a real-world decision without checking the context. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 37: Do not overfit the noise
Given: A real attendance record is 20,31,39,52. Mathematical judgement: The differences are 11,8,13; no exact simple fixed-difference or fixed-ratio pattern is established.
Examination check: Report that the four observations do not follow those exact models, rather than announcing an invented rule. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 38: Interpreting a negative prediction
Given: The linear model y = 12 − 3x yields y = −3 at x = 5. Mathematical judgement: Algebraically the value is −3, but a count of students cannot be negative.
Examination check: Distinguish the computed output from the meaningful domain and state that the model is unsuitable at this input. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 39: Wrong interval claim
Given: A chart marks x = 0,2,4,6 with y = 4,12,20,28. Mathematical judgement: The change is +8 per two x-units; the linear gradient is +4 per x-unit.
Examination check: A student who says the gradient is eight has ignored the horizontal change in the denominator. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Worked clinic 40: Evidence of symmetry
Given: A table around x = 2 gives y(0)=5,y(1)=2,y(2)=1,y(3)=2,y(4)=5. Mathematical judgement: The values are symmetric about x = 2 and consistent with a quadratic minimum there.
Examination check: Symmetry is useful graphical evidence, but always verify the stated rule if the problem provides one. Before leaving this problem, identify the input interval, test the rule with at least one coordinate and state the mathematical quantity in its correct unit. A fitting expression is not a licence to make a wider forecast than the evidence allows.
Four full problems with complete reasoning
Problem A: the school garden. A garden records weekly seedling heights, in centimetres, of 4, 7, 10 and 13 for weeks 0 to 3. A proposed rule is h = 3w + 4. Check it by substitution: h(0)=4; h(1)=7; h(2)=10; h(3)=13. The constant increase is three centimetres per week. The intercept is four centimetres at the reference week. Neither the equation nor these observations establishes that growth will remain linear indefinitely. Living plants have environmental and biological limits. A good answer names the period studied and the rate described by the model.
Problem B: the makerspace display. A fictional exhibition uses a quadratic height model h = −(t − 2)² + 9 for a moving model, with t measured in seconds and h in metres. Expand if necessary: h = −t² + 4t + 5. The turning point is (2,9), so the model’s maximum height is 9 metres at t = 2 seconds. At t = 0, h = 5 metres. This is a mathematical example only; the exact motion of a real model requires the assumptions in its context. Draw the axis of symmetry t = 2 and check the end-points of the time interval given by the question.
Problem C: repeated registration factors. A club records 5, 15, 45, 135 registrations in four equal observation periods. One formula that fits is R = 5 × 3^n for n = 0,1,2,3. Test n = 2: 5 × 9 = 45. A student writes, ‘135 means 405 participants are certain next period.’ That is an unjustified leap. The formula would produce 405 if the factor continued; the observations cannot prove that it will. Demand, capacity and time boundaries must be distinguished. Write ‘the model predicts’ rather than ‘the school will have’.
Problem D: select, then reject. A notice reports attendance 12, 18, 26, 38 over equal days. Differences are 6, 8 and 12; second differences are 2 and 4; ratios are 1.5, about 1.444 and about 1.462. These numbers do not exhibit an exact constant first difference, an exact constant second difference or an exact constant ratio. You might fit a function using a separate modelling method, but the simple pattern tests do not establish one of the three families uniquely. The defensible conclusion is limited: the data increase, while the exact form remains unsupported. This is often better reasoning than an elaborate wrong formula.
Where students lose marks in graph questions
A correct graph-family label can still accompany a poor solution. Typical weaknesses include forgetting axis units, reading the wrong scale, omitting substitutions, miscalculating second differences, mixing a rate of change with an intercept, ignoring the x-interval, confusing mathematical possibility with physical feasibility, and quoting a decimal answer before the problem specifies the appropriate accuracy. These errors are repairable because each one corresponds to a visible check. Students should learn to inspect their own steps instead of relying on the appearance of a final answer.
For a comparison problem, organise working in this order: input values; candidate pattern; explicit formula; verification against observed points; interpretation of the answer. For a quadratic sketch, add turning point, axis of symmetry and intercepts where applicable. For a linear graph, show gradient and intercept accurately. For an exponential graph, identify k, the repeated factor, and the relevant values. For a local rate on a curve, draw the tangent carefully and compute vertical change divided by horizontal change with units.
Independent practice in 35 minutes
- First five minutes: review definitions of constant difference, constant second difference, constant ratio and the three basic function forms.
- Next ten minutes: solve five short classification tables without drawing the graphs; write a reason for each choice.
- Next ten minutes: select two functions and sketch them, labelling axes, intercepts, symmetry or repeated-factor behaviour where applicable.
- Next five minutes: invent an example with too little information to identify a unique model; explain why two points are insufficient.
- Last five minutes: check calculations, units and whether any conclusion extends beyond the measured range.
Questions and careful distinctions
Does a table with constant second differences prove a real phenomenon is quadratic? Not universally. It is consistent with a quadratic rule over equal input steps; more information is needed about the underlying process. Can a quadratic rise and then fall? Yes, depending on its sign and domain. Does an exponential always mean ‘rapid’? No. The defining feature is the multiplicative structure, not the visual drama of one chosen scale. Can several rules fit a small dataset? Certainly. The specified function family or broader evidence is then important.
Does a modelling question always ask for a forecast? No. It may ask for a graph, a comparison, a coordinate, a gradient or a limitation. Should I use an advanced forecasting tool? Not when the task can be solved within the official G3 syllabus and the question expects clear working. The most impressive method is the one that answers the question correctly, visibly and within its given assumptions.
Your connected learning route
Return to the G3 SEC learner-guide hub. For repeated factors and the meaning of an exponential model, review Vol 0226. For estimating local slope, return to Vol 0222 on tangent gradients. If the earlier algebraic and numerical foundation is unreliable, bridge through the Primary 6 Mathematics Learning Hub and the PSLE Learning Guide before resuming. Each route has a separate purpose; this volume owns the comparison of function families.
When you finish a graph question, ask yourself two questions: ‘What mathematical change does this rule represent?’ and ‘Which part of my conclusion is an observation, which is a calculation and which is only a prediction?’ Answer those well and you have made a considerable step toward independent G3 performance.
