Reading Beyond the Individual Dots
A dotplot lets us see both individual observations and where observations gather along a quantitative scale. Once the dots are plotted, the important question is not only “What values occurred?” but also “Where are observations concentrated, and where are they scarce or absent?” These patterns can help summarize quiz scores, measurement times, or other quantitative data without hiding the individual recorded values.
A visual pattern is a description of the plotted data, not an explanation of how the pattern arose. A cluster does not by itself show that a group of observations came from a separate population. A gap does not prove that values in that range are impossible, and a separated dot does not prove that a measurement is wrong. Careful interpretation keeps those distinctions clear.
What Counts as a Visible Pattern?
To describe a cluster, name the range of values where dots are relatively close together or numerous compared with nearby parts of the plot. Include the variable and its units when there are units. “Most recorded times are between 12 and 18 minutes” is more useful than “the dots are bunched up.” A cluster can contain many observations across a broad range, or a smaller concentration within a wider spread of data.
A gap has a more specific visual meaning: no observation is plotted in the stated interval. The size of a gap should be judged against the horizontal scale, not just the physical width of the blank space on a screen or page. Check the tick marks and labels before describing its endpoints. For integer-valued data, a gap from 10 to 11 means there are no observations at those recorded values; for continuously measured data, the plot may show no recorded observations throughout an interval.
An isolated value is noticeably separated from nearby observations. Say where it lies and how it is separated. For instance, “one time of 39 minutes lies well above the other recorded times, which range from 14 to 24 minutes” describes the plot. Calling that value an error, or assigning a technical label to it, would require additional criteria or information.
The terms are comparative rather than fixed rules. There is no universal number of dots that makes a cluster, or universal distance that makes a value isolated. A feature may look prominent when the display covers a narrow range and less prominent when a wider range is shown. The reader should use the scale and the full plotted distribution as context.
Worked Examples
Worked Example: Two Concentrations in Quiz Scores
A hypothetical class has 16 quiz scores, recorded as whole-number points. A dotplot has the following counts at each score:
| Score | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 |
|---|---|---|---|---|---|---|---|---|---|---|
| Number of dots | 1 | 2 | 3 | 2 | 0 | 0 | 1 | 2 | 3 | 2 |
The first visible concentration is from 6 to 9 points. Its count is \(1+2+3+2=8\) scores. The second is from 12 to 15 points, also with \(1+2+3+2=8\) scores. As a share of all 16 scores, each range contains \(8/16=0.50\), or 50%.
There are no dots at 10 or 11 points, so those recorded scores form a gap between the two concentrations. A careful description is: “The 16 quiz scores form concentrations from 6 to 9 points and from 12 to 15 points, with no scores of 10 or 11.” This describes the plotted values and gives counts to support the claim.
The display does not show why the scores gather in these ranges. It cannot establish that the class contains two kinds of students, or that the gap reflects a particular teaching or testing effect. Those would be explanations, not observations read directly from the plot.
Worked Example: A Separated Completion Time
In a hypothetical group of 16 hikers, trail completion times are recorded in minutes. The dotplot shows observations at 14, 15, 15, 16, 16, 17, 17, 18, 18, 19, 20, 21, 22, 23, 24, and 39 minutes.
Fifteen of the 16 times lie from 14 to 24 minutes, while one time is 39 minutes. The nearest recorded time to 39 is 24, so the separation is \(39-24=15\) minutes. The plot therefore shows one time well above the main concentration, with a blank interval between 24 and 39 minutes.
A suitable interpretation is: “Most recorded completion times are between 14 and 24 minutes; one time of 39 minutes is separated from them by a 15-minute difference from the next-highest time.” This does not show whether the hiker took a different route, stopped, or had a timing mistake. The graph alone provides no evidence to choose among those explanations.
Nor should “isolated” be treated as a verdict that the value is invalid. It identifies a visual relationship to the other observations. To decide whether the record is accurate, one would need to check the measurement and data-collection information.
Worked Example: Comparing Concentrations Across Score Ranges
A hypothetical placement test is scored from 0 to 100 points. In a dotplot of 20 scores, 12 observations fall from 40 to 49 points, 5 fall from 50 to 59, and 3 fall from 60 to 69. No observations lie outside 40 to 69.
The stated ranges account for all the observations because \(12+5+3=20\). The proportion in the 40–49 range is \(12/20=0.60\), or 60%; in 50–59 it is \(5/20=0.25\), or 25%; and in 60–69 it is \(3/20=0.15\), or 15%.
The plot has its strongest concentration in the 40–49 point range, with fewer observations in each of the next two ranges. This is a supported comparison of the displayed counts. It would be inaccurate to say that every score is concentrated between 40 and 49, because 8 of the 20 scores fall elsewhere.
This description also depends on the intervals being stated clearly. “There are more scores near the lower end” is vague; the count and the score range make the comparison interpretable. The counts describe these 20 hypothetical scores and do not, by themselves, establish the proportions for all possible test-takers.
Why a Gap Means No Plotted Observations
The logic behind reading a gap is simple. Each dot represents an observation at its horizontal-axis value. For a chosen interval, count the dots whose values fall in that interval. If the interval is blank on a complete dotplot, that count is zero. In symbols, if \(A\) is a range of values and \(I_i(A)\) equals 1 when observation \(i\) falls in \(A\) and 0 otherwise, the count in the range is:
A blank interval has \(I_i(A)=0\) for every observation, so the sum is zero. This justifies saying that the plotted sample has no observations in that interval. It does not justify saying that no individual in a larger population could have a value there. The graph displays recorded sample values, not all possible values.
Also check how the values were recorded. If a measurement is rounded to whole units, an apparent gap in the dotplot concerns the rounded values. It does not reveal whether unrounded measurements might have fallen between them. The graph should be interpreted at the precision and scale actually shown.
Common Mistakes and Careful Reading
- Calling a gap impossible. The wrong reasoning is, “There are no dots there, so that value cannot occur.” The correction is that the complete plot shows no recorded observations there; it does not show that the value is impossible in the population.
- Calling every separated dot an error. A dot far from the others is a visible isolated value, not proof of a recording mistake. Check the original record or measurement process before making that claim.
- Giving a cause for a cluster. A concentration may invite questions about how the data were generated, but the dotplot alone does not answer them. Report the values and pattern first, and use additional evidence to investigate possible explanations.
- Ignoring the axis scale. A blank span can look large or small depending on the labeled scale. Read the tick marks and use the units before describing the span.
- Using “most” without checking the count. If a range contains 12 of 20 scores, it contains a majority; if it contains 8 of 20, it does not. Count the observations before choosing words such as “most” or “a few.”
These descriptions matter because a concise visual summary can guide what questions to ask next. A cluster may prompt a closer look at observations within that range; a gap may prompt checking measurement precision or how data were collected; an isolated value may prompt verifying its record. Those are reasons to investigate, not conclusions established by the dotplot itself.
Check Your Understanding
Use the scale, count, and context to answer each question.
- A dotplot of 18 daily temperatures has 13 observations between 18 and 23 degrees and none between 24 and 27 degrees. Describe the concentration and the gap, including units.
- One dot lies well above the rest of a dotplot of delivery times. What can you say from the plot, and what cannot you conclude without more information?
- A student says, “There is a gap from 30 to 35, so values in that range cannot occur.” Explain what the plot supports and how the claim should be corrected.
- Why should you check the horizontal-axis labels before describing a gap as large?
- A range contains 9 of 15 observations. Calculate the fraction and percentage in that range, then state what this count does and does not tell you.