From Noticing a Feature to Writing a Full-Credit Description
In Common Mistakes in Describing Distributions, you learned to avoid vague language, name skew from the tail, and keep claims tied to the displayed data. This tutorial takes the next step: it shows how to revise a weak description into one that is specific, supported, and responsive to the question.
A strong description is not necessarily long. It gives the reader enough information to understand which distribution is being described and why the statements fit the display. The goal is not to insert every possible statistic. Instead, select the relevant features and state them accurately, in context.
AP questions can ask for an overall description, a particular feature, or a comparison. The exact scoring rubric depends on the question, so no single sentence guarantees credit in every situation. Still, the same revision habits help: check the prompt, use specific evidence, and make each number meaningful in context.
A Practical Revision Routine
Before revising a sentence, identify what the question actually asks. If it asks for shape, center, and spread, a response that reports only the median is incomplete. If it asks whether a distribution is skewed, a long list of unrelated summaries does not improve the answer.
Name the group and the quantitative variable, including units when they are available. This gives every later claim a clear subject.
Trade words like “weird,” “high,” or “spread out” for visible features, such as a tail, a cluster over a stated range, a gap, or a value separated from the rest.
As in Choosing Mean or Median to Describe Center and Choosing IQR or Standard Deviation to Describe Spread, match the summaries to the distribution’s shape and unusual values.
Give units and say what each summary describes. A value by itself is not a complete statement about center or spread.
Describe the observations shown. Do not assume a cause, call a value an error without evidence, or generalize to a larger population unless the question supports that.
This routine builds on the SOCS framework from The SOCS Framework for Describing Distributions. Think of it as a revision tool rather than a new order for describing every graph. A good final response includes the parts the question needs, not a checklist pasted in regardless of the task.
Worked Examples: Weak Wording and Stronger Revisions
Worked Example: Replace Vague Shape and Spread Language
A dotplot represents the waiting times, in minutes, for 11 fictional customers at a service counter. The observations are 12, 14, 15, 16, 17, 18, 19, 20, 22, 24, and 41. A student writes, “The times are kind of spread out, and there is a weird one.” Revise the description to report shape, an unusual feature, center, and spread.
Solution. First, sort and inspect the observations. They are already in increasing order. Most lie from 12 to 24 minutes, followed by a larger value of 41 minutes. That separated high observation creates a tail toward larger times, so the distribution is right-skewed. The median is the sixth observation, 18 minutes.
For the IQR, use the median-of-halves method. The lower five observations are 12, 14, 15, 16, and 17, so \(Q_1=15\) minutes. The upper five are 19, 20, 22, 24, and 41, so \(Q_3=22\) minutes. Therefore,
The 1.5 IQR fences are \(15-1.5(7)=4.5\) minutes and \(22+1.5(7)=32.5\) minutes. Since 41 is above 32.5, it is flagged by the 1.5 IQR rule. That rule flags an observation as a potential outlier; it does not prove that the observation is an error.
Revised description. “For these 11 fictional customers, waiting times are concentrated from 12 to 24 minutes, with a right tail and one higher value of 41 minutes that is flagged by the 1.5 IQR rule. The median wait is 18 minutes, and the IQR is 7 minutes, so the middle half of the waiting times spans 7 minutes.”
This revision replaces “kind of spread out” with a named measure and its interpretation, and replaces “weird one” with a specific value and a cautious description. It also names the group, variable, and units.
Worked Example: Add Context to a Correct but Incomplete Observation
A histogram summarizes the daily number of reusable containers returned to a fictional school dining area over 30 days. The histogram has one main peak, and its counts taper toward larger values. Most days fall between 40 and 70 returns; a few days have between 80 and 100. A student writes, “It’s skewed.” What should be improved?
Solution. The statement does not say which distribution is skewed or give evidence for the label. The histogram shows a concentration from 40 to 70 returns and a thinner tail toward larger counts. That supports describing the distribution as right-skewed. The response should name the variable, group, and units. Because no center or spread summaries are provided, do not invent them.
Revised description. “For the 30 days shown, the distribution of reusable-container returns at the fictional school dining area is unimodal and right-skewed. Most days had 40 to 70 returns, while a few days had 80 to 100 returns, extending the tail toward larger counts.”
The revision uses the histogram’s intervals as evidence, rather than treating “skewed” as a complete description. It does not claim that returns are typically a particular exact number or explain why some days had more returns. A histogram groups values into bins, so its intervals support approximate descriptions, not claims about exact values within each bin.
Worked Example: Describe What a Boxplot Supports
A modified boxplot displays delivery times, in days, for a fictional set of local orders. The lower whisker extends from 1 to 2 days, the box runs from \(Q_1=2\) to \(Q_3=5\) days, the median is 3 days, and the upper whisker extends from 5 to 8 days. A separate plotted point is at 12 days. A student writes, “The data are normal, and 12 is a mistake.” Rewrite the statement using only what the boxplot supports.
Solution. A boxplot does not show enough detail to establish that a distribution is normal. The upper half of the box, from 3 to 5 days, is longer than the lower half, from 2 to 3 days. The upper whisker, from 5 to 8 days, is also longer than the lower whisker, from 1 to 2 days. These features suggest right skew, though a boxplot gives only a summary of the distribution’s shape. The separate point at 12 days is shown as a potential outlier; the display does not establish that it is a mistake.
The IQR is \(Q_3-Q_1=5-2=3\) days. It describes the width of the middle 50% of the delivery times. The median is 3 days.
Revised description. “For these fictional local orders, delivery times have a median of 3 days and an IQR of 3 days, so the middle half spans 3 days. The longer upper sections of the boxplot suggest right skew, and the plotted value at 12 days is a potential outlier, not necessarily a recording error.”
This version distinguishes what the plot suggests from what it proves. In particular, it does not diagnose a normal distribution from a boxplot or assign a cause to the separate point.
Worked Example: Make a Comparison Direct and Fair
Two dotplots show fictional times, in minutes, for students to complete the same short puzzle. Group A has a median of 8 minutes, an IQR of 3 minutes, and one value of 19 minutes separated above the main cluster. Group B has a median of 10 minutes and an IQR of 2 minutes, with no similarly separated value. A student writes, “A is better, but B is more consistent.” Revise the comparison without assuming what “better” means.
Solution. The word “better” is unclear: it might mean faster, more accurate, or something else, and the information gives only completion times. A smaller median time means the typical completion time, as represented by the median, is lower for Group A. A smaller IQR means the middle half of Group B’s times is less spread out. Neither comparison establishes that one group is better overall.
Revised comparison. “For the fictional students shown, Group A has a lower median puzzle-completion time than Group B (8 minutes compared with 10 minutes). Group B has a smaller IQR (2 minutes compared with 3 minutes), so its middle half of completion times is less spread out. Group A also has one value of 19 minutes separated above its main cluster.”
This revision states which group has the lower center and which has the smaller spread, with units and a clear interpretation. It avoids turning a description of times into a judgment about performance that the question has not defined.
Common Mistakes and What a Strong Response Does
- Keeping a vague word and adding no evidence. “The scores are unusual” leaves the reader guessing. A stronger response identifies the feature: for example, “one score of 4 points lies below the main cluster from 12 to 18 points.”
- Listing a statistic without interpreting it. “The IQR is 7” omits both units and meaning. A complete statement identifies the variable and says that the middle half spans 7 of its units.
- Reporting only the center when the prompt asks for a distribution. A median alone does not describe shape or spread. Check the prompt and include the requested features, using suitable measures.
- Calling a graph normal or uniform without enough support. A boxplot cannot reveal the detailed pattern needed to describe peaks or gaps. Use a histogram or another suitable display when those features matter.
- Giving an explanation that the graph cannot establish. A high value may be a potential outlier, but the display alone does not show whether it is a mistake or what caused it. State what is visible and separate it from possible explanations.
- Making an undefined comparison. Words like “better” or “more typical” can hide what is being compared. State the actual difference in center or spread, with values and units.
- Writing every observation instead of describing the pattern. A list of values may be accurate but does not explain the distribution. Summarize the pattern and use a few relevant values or intervals as evidence.
A quick final check is to ask: Could a reader identify the group and variable? Is each claim supported by the display or a stated summary? Are the units clear? Did I answer the question without adding an unsupported cause or conclusion? If any answer is no, revise that sentence.
Check Your Understanding
For each item, focus on how the wording can be made more specific and better supported.
- A histogram shows a concentration of fictional plant heights from 12 to 18 centimeters and a thin tail toward 30 centimeters. Rewrite: “The plants are weirdly shaped.”
- A response says, “The median is 14.” Name two pieces of context that would help make this a complete statement.
- A boxplot has a separate point at 25 minutes. What can you say about that point, and what should you avoid claiming without further evidence?
- Two groups have median wait times of 6 and 9 minutes. Why is “the first group is better” not a sufficiently precise comparison?
- A question asks for shape, center, and spread. A student reports only the mean. What additional information should the student consider including?