Did you know that two students can have exactly the same study time and still produce different quiz scores? A scatter plot is supposed to show those differences. Yet a reader sometimes sees two or three dots directly above the same horizontal position and assumes the graph is broken, or that someone has plotted a pupil twice.
Repeated x-values in a scatter plot mean several observations have the same horizontal measurement while their vertical measurements may differ. Place each observation at its own ordered pair (x, y), including those in a vertical stack. It is not an error to have more than one y-value for one x-value; it is a clue about variation within that x-group. What requires separate checking is whether two identical ordered pairs are drawn on top of each other and therefore visually hidden.
Start with the meaning of each coordinate, not with the shape of the cloud
A scatter plot shows pairs of observations. Each dot represents one pair, not one x-value or one y-value taken separately. The x-axis might be minutes of practice; the y-axis might be correct answers out of ten. A vertical stack at x = 20 means more than one observation was made for 20 minutes of practice. It does not mean the score was measured on a different time scale.
Ask the learner to read one point aloud as a complete sentence: “For this recorded observation, 20 minutes of practice went with a score of 6 out of 10.” Then point to another dot at the same x-position and ask whether that dot must belong to the same pupil. Unless the source table identifies pupils and repeated measurements, the graph alone does not answer that question. The two points may represent two different pupils, repeated trials, or another legitimately specified unit of observation.
An original worked example: eight fictional pupils and four practice durations
Illustrative classroom data created for this lesson: A teacher records each of eight fictional pupils’ chosen practice duration and a later ten-question quiz result. For this exercise, practice durations are measured in minutes, and quiz scores are counts out of 10. No real pupil outcomes are reported and no causal claim is being made.
- Pupil A: (10 minutes, 4 correct); pupil B: (10 minutes, 7 correct).
- Pupil C: (20 minutes, 5 correct); pupil D: (20 minutes, 8 correct); pupil E: (20 minutes, 6 correct).
- Pupil F: (30 minutes, 7 correct); pupil G: (30 minutes, 9 correct).
- Pupil H: (40 minutes, 8 correct).
Check the table before drawing: there are 2 + 3 + 2 + 1 = 8 observations. The x-axis can be marked 0, 10, 20, 30 and 40 minutes; the y-axis can show whole-number scores from 0 through 10. Plot each ordered pair rather than averaging pupils who practised for the same duration.
At x = 10, draw dots at y = 4 and y = 7. At x = 20, draw dots at y = 5, 6 and 8. At x = 30, draw dots at y = 7 and 9. At x = 40, draw one dot at y = 8. The three dots at 20 minutes form a vertical stack because their x-values agree, not because they have identical y-values.
What can the vertical stack really tell us?
For 20 minutes of recorded practice, the observed scores are 5, 6 and 8 out of 10. Their arithmetic mean is (5 + 6 + 8) ÷ 3 = 19 ÷ 3, or about 6.33 correct answers. Their observed vertical range is 8 − 5 = 3 answers. Those are different summaries: the mean describes the centre of these three recorded scores, while the range describes the difference between the largest and smallest of those scores.
The two 10-minute scores have a range of 7 − 4 = 3 answers. The two 30-minute scores have a range of 9 − 7 = 2. Only one score was recorded for 40 minutes, so its observed range is 0 by a single-value calculation; that does not show that every pupil practising 40 minutes would score 8, or that the underlying variation is zero. An honest report says only one observation was available at 40 minutes.
This is the important transfer: a repeated x-value allows us to inspect the spread of y at a fixed x in the sample. It does not allow us to name the cause of the spread. Prior knowledge, task difficulty, concentration, measurement error and ordinary variation could all matter, and this small observational table cannot isolate them.
A different problem: identical x and y can conceal repeated records
Now imagine a ninth fictional pupil I who also has (20 minutes, 6 correct). If plotted with an ordinary opaque dot, I’s point may sit directly on top of pupil E’s point. The picture may appear to contain only eight visible dots although the table contains nine rows. This is called overplotting. It is not the same as the vertical separation of (20, 5), (20, 6) and (20, 8).
Before interpreting how many pupils share a result, compare the row count with the point count. Appropriate display choices include a clearly explained count annotation at (20, 6), a frequency label or a small amount of visual separation that is explicitly described as display-only jitter. Never shift a value silently and then present the shifted location as its actual measurement.
A one-minute learner diagnostic
Draw two small sketches. Sketch A contains dots at (2, 3), (2, 5) and (2, 7). Sketch B contains three records all at (2, 5), drawn with identical symbols on top of one another. Ask, “Which sketch reveals three separate y-values? Which could hide a frequency of three behind one visible mark?”
A learner who answers “A shows repeated measurements and B must have only one point” has confused visible symbol count with data-row count. Ask them to reconstruct the source pairs and to distinguish a vertical stack from an overlap. This diagnosis is better than assigning more generic graph questions.
Three rounds of practice from supported to independent
Round 1: teacher models coordinate language
Use the eight-pupil table above. Let the teacher place the first four dots and speak each pair: “twenty minutes, five correct.” Then the learner places the remaining four. Check whether x and y have been reversed and whether the count of plotted observations agrees with the source rows.
Round 2: learner explains variation without inventing causes
Offer the observations (5 hours, 2 books), (5 hours, 4 books), (6 hours, 3 books) for a fictional library-reading log. Ask why two marks share x = 5. A sound answer is that two records report five reading hours but different book counts. It is not sound to claim automatically that a five-hour reader “should” finish three books or that one student worked harder.
Round 3: independent transfer to a new scatter plot
Give the points (1, 2), (1, 2), (1, 4), (3, 5) and (3, 6), with axes labelled number of weekly practice sessions and completed tasks. There are five records, two distinct x-values, and four unique plotted coordinate locations because (1, 2) occurs twice. The student must explain how to show both the repeated coordinate and the distinct y-values. They should be able to do so without a demonstration.
Misconceptions worth repairing before they become habits
- “One x-value can have only one y-value.” That is a property of a mathematical function, not a rule for observational scatter plots. A collection of observed pairs need not define a function.
- “Three stacked dots mean a measurement mistake.” Repeated practice times can be legitimate. Check the original table and data definition before deleting or changing a row.
- “The highest dot proves practice caused success.” Association in an observational plot is not evidence of causation by itself.
- “A single point shows no variation in the group.” One observation provides no evidence about how widely unobserved outcomes would vary.
- “The picture always reveals the sample size.” Identical plotted coordinates can overlap. Check counts or a frequency-aware display.
What parents and teachers can actually check
A parent can use pencil and squared paper with five simple (x, y) pairs. Ask the learner to point to the x-axis unit, the y-axis unit, each represented observation and any repeated coordinate. The useful question is “How many records do you have, and how many separate locations are visible?” rather than “Does this look like a nice straight line?”
Teachers can record three observable milestones: the learner plots every source row accurately, explains why vertical stacks are permitted, and notices when identical pairs overprint. In a later unfamiliar example, ask the learner to justify a claim with the table rather than the apparent dot count alone. Accuracy, explanation and transfer should be noted separately.
Limits: when does this lesson stop being enough?
Dot size, transparency, overlapping locations and scaling decisions affect what viewers see. A crowded data set may need a count-based chart, a jittered display with the transformation disclosed, or a more specialised statistical model. A scatter plot with only a few points may not establish any stable trend. Never use a classroom illustration to promise how much practice time produces a particular score.
For technical background, the NIST Engineering Statistics Handbook’s scatter-plot explanation defines the graph as paired y-versus-x observations. Its no-relationship example illustrates that many y-values may occur for a given x-value. Those sources explain graphical principles, not results for the fictional pupils.
Frequently asked questions
Can a scatter plot contain points directly above one another?
Yes. The records can share the same x-value and differ in y. They will occupy different vertical positions on the same vertical line.
If two points have the same x and y, is one necessarily a duplicate error?
No. They could represent two genuine observations with equal measurements. Only the data collection records can tell you whether an entry was accidentally duplicated.
Should we average all the y-values at each x before drawing the plot?
Not by default. Averaging changes the chart from individual observations to a summary. It may be useful for a separately labelled display, but it conceals within-group variation and should not silently replace the raw scatter plot.
What does success in this lesson look like?
The learner can go from source table to accurate dots, explain a repeated x-position without panic, inspect hidden overlaps, and state limits on what the picture proves. That is evidence-based graphical literacy.
Continue learning with careful sources
Read the related Core Aim guide on preserving repeated values in a split stem-and-leaf plot for a different representation of repetition. For the wider learning route, visit the verified eduKate Education Hub. The unchanged Clementi Mathematics teaching reference demonstrates why diagnosis, explanation, progressive practice and independent checks belong together.
The best habit is small and powerful: count the records, read the axes, and then read the dots. A scatter plot becomes much more trustworthy once the learner knows which apparent repetitions are informative and which need a closer look.