AP Statistics Unit 1 Notes: Learning to Summarize One-Variable Data

Variation in Categorical and Quantitative Variables

Statistics starts with a simple idea: real data are messy. Even when you measure the “same thing” (heights of students, number of texts sent yesterday, favorite music genre), the values usually differ from person to person. That natural difference is called variation. Understanding variation is the whole point of making graphs and summary descriptions—without variation, there would be nothing interesting to analyze.

A crucial first step is to recognize what kind of variable you have, because that determines what graphs and descriptions make sense.

Categorical vs. quantitative variables

A variable is a characteristic recorded for each individual in a dataset (an “individual” could be a person, a day, a product, a country, etc.). In AP Statistics, you’ll constantly decide whether a variable is categorical or quantitative:

• Categorical variable: places individuals into groups or categories. The values are labels.
  • Examples: eye color, political party, type of phone, “yes/no,” brand.
• Quantitative variable: takes numerical values where arithmetic makes sense in context.
  • Examples: height (in cm), number of siblings, reaction time (in ms), test score.

Why this matters: if you treat categories like numbers (for example, averaging “freshman=1, sophomore=2, junior=3, senior=4”), you can create summaries that look mathematical but are meaningless. The type of variable determines what “variation” means and how you display it.

What “variation” means for each type

For categorical variables, variation means the distribution of counts or proportions across categories. You ask:

• Which categories are most common?
• How different are the category frequencies?
• Are any categories rare or surprising?

For quantitative variables, variation means how spread out the numerical values are and how they cluster. You ask:

• What values are typical?
• How much do values differ from a typical value?
• Is the distribution symmetric, skewed, or does it have outliers?

A helpful way to think about it: categorical data vary by kind, quantitative data vary by amount.

Distribution: the unifying idea

A distribution tells you what values a variable takes and how often it takes them.

• For categorical variables, the distribution is the set of categories with their counts or proportions.
• For quantitative variables, the distribution is the set of numerical values with their frequencies (often grouped into intervals for convenience).

Almost everything in this unit—graphs, center, spread, shape, outliers—is a way of describing a distribution.

Example: identifying variable type and variation

Suppose a gym records data on 50 members:

• Variable A: “Primary workout type” (cardio, weights, classes, mixed)
• Variable B: “Minutes exercised last week”

Variable A is categorical. Variation shows up as differences in how many people choose each workout type.

Variable B is quantitative. Variation shows up as differences in minutes; you could meaningfully compute summaries like a mean or median, and you could look for clusters (many people around 120–180 minutes, for instance) or outliers (someone at 900 minutes).

What goes wrong (common classification traps)

• Numbers used as labels: ZIP codes, jersey numbers, phone numbers are categorical even though they’re digits. Averaging ZIP codes is nonsense.
• Quantitative but discrete: counts (like number of siblings) are quantitative even though you only see whole numbers.
• Binary categories (yes/no) are categorical, even though you might code them as 0/1 for convenience.

Exam Focus

• Typical question patterns:
  • Identify whether a variable is categorical or quantitative and justify.
  • Interpret what “variation” means for a given context (counts/proportions vs. spread/shape).
  • Choose an appropriate type of display based on variable type.
• Common mistakes:
  • Treating coded categories (like 1=male, 2=female) as quantitative and computing a mean.
  • Calling any variable with numbers “quantitative” without checking whether arithmetic is meaningful.
  • Describing categorical variation with quantitative language (like “skewed”).

Representing Data with Graphs (Dotplots, Histograms, Stemplots, Bar Charts)

Graphs are not decoration—they are tools for seeing features of a distribution that are hard to notice in a raw list of values. A good graph makes patterns visible (clusters, gaps, outliers, skew) and helps you communicate those patterns clearly.

The most important decision is matching the graph to the variable type:

• Categorical data are typically shown with bar charts.
• Quantitative data are commonly shown with dotplots, stemplots, and histograms (each has strengths and tradeoffs).

Bar charts (categorical)

A bar chart displays categories on one axis and counts (or proportions/percents) on the other. The bars are separated because the categories are distinct labels, not a number line.

Why it matters: Bar charts make it easy to compare category frequencies and to spot the most/least common categories.

How to make/interpret one well:

• Use a clear, descriptive title.
• Label axes (category names and either counts or percents).
• Keep a consistent scale starting at 0.
• Focus your description on comparisons: “Category A is about twice as common as Category B.”

Common confusion: A bar chart is not the same as a histogram. In a bar chart, the x-axis is categories and the bars are separated. In a histogram, the x-axis is a number line and bars touch.

Dotplots (quantitative, small to moderate datasets)

A dotplot places each data value as a dot above a number line. If values repeat, dots stack.

Why it matters: Dotplots show individual data points, so they are excellent for small datasets where you want to see exact values, clusters, gaps, and potential outliers.

How it works:

1. Draw a number line covering the data range.
2. For each observation, place a dot above its value.
3. Stack dots for repeated values.

When dotplots shine:

• Small samples (you can still see individuals).
• When you care about repeated exact values (like test scores out of 10).

What goes wrong:

• With large datasets, dotplots become cluttered; patterns become hard to see.

Stemplots (quantitative, preserves actual values)

A stemplot (stem-and-leaf plot) splits each data value into a stem (leading digits) and a leaf (trailing digit). It’s like a histogram that still lets you reconstruct the original data.

Why it matters: Like dotplots, stemplots preserve individual values, but they can be more compact when values have a consistent number of digits.

How it works (step-by-step):

1. Choose what digits will be the stem (often all but the last digit).
2. List stems in a vertical column in increasing order.
3. For each data value, write its leaf next to the corresponding stem.
4. Sort leaves within each stem from smallest to largest.
5. Include a key, such as “4|7 means 47.”

Split stems: If many values land in the same stem (too crowded), you can split stems (for example, one row for leaves 0–4 and another for 5–9).

What goes wrong:

• Missing or unclear keys—without a key, the scale is ambiguous.
• Inconsistent stem choices that hide structure (too few stems lumps everything together; too many stems creates a sparse, unhelpful display).

Histograms (quantitative, moderate to large datasets)

A histogram groups quantitative data into intervals called bins and displays the frequency (or relative frequency) in each bin. Bars touch because the x-axis is a continuous number line.

Why it matters: Histograms are the workhorse for showing the overall shape of a quantitative distribution—especially when there are many observations.

How it works conceptually:

• You trade away exact individual values to gain a clearer view of the overall pattern.

Key design choice: bin width (or number of bins)

• Wider bins smooth the graph (less detail).
• Narrower bins show more detail but can introduce “noisy” patterns.

A good histogram uses bins that reveal the important structure without overreacting to random bumps.

Frequency vs. relative frequency histograms:

• Frequency uses counts on the y-axis.
• Relative frequency uses proportions or percents on the y-axis.

Relative frequency is especially helpful when comparing groups with different sample sizes.

Choosing the right graph (and what you can say from it)

Different displays emphasize different information. A useful way to decide is to ask: “Do I need to see individual data values, or just the overall pattern?”

Worked example: same data, different displays

Suppose you have quiz scores (out of 20) for 15 students:

8, 9, 10, 10, 11, 12, 12, 12, 13, 14, 14, 15, 16, 18, 19

• A dotplot would show three dots stacked at 12 and two at 10 and 14—great for seeing repeats and the exact values.
• A stemplot might use stem 0 and 1 (since scores are two-digit at most) with leaves showing each score; it preserves exact values and is quick to read.
• A histogram might use bins 8–11, 12–15, 16–19, showing roughly where most scores fall; it highlights the overall shape but hides the fact that 12 occurs three times.

Different graphs answer slightly different questions; none is “the one correct graph” in all situations.

What goes wrong (graphing errors that cost points)

• Histogram vs. bar chart mix-up: touching bars for categorical data or separated bars for quantitative bins is a red flag.
• Bad bin choices: bins that are so wide they hide important features (like a gap) or so narrow they create misleading jaggedness.
• Unlabeled axes or missing units: you must communicate what the scale means.
• Using a pie chart to compare categories: pie charts make precise comparisons hard; AP Statistics emphasizes bar charts for categorical distributions.

Exam Focus

• Typical question patterns:
  • Choose an appropriate graph for a variable (or compare which display is better and why).
  • Interpret a graph’s features: identify clusters, gaps, peaks (modes), skew, potential outliers.
  • Explain how changing bin width in a histogram changes the appearance and what stays the same.
• Common mistakes:
  • Calling a histogram a bar chart (or describing histogram bars as “categories”).
  • Describing a graph without context (you need units and what the variable represents).
  • Over-interpreting small bumps in a histogram as meaningful “patterns” when they may be binning artifacts.

Describing the Distribution of a Quantitative Variable

Once you can display quantitative data, the next skill is to describe the distribution clearly and completely. In AP Statistics, a strong description is not a stream of adjectives—it’s a structured explanation that touches the key features and uses context.

A standard framework you’ll use all year is SOCS:

• Shape
• Outliers (and other unusual features like gaps)
• Center
• Spread

You can think of SOCS as the “four things a reader needs” to understand what the variable looks like in this group.

Shape: what the overall pattern looks like

Shape refers to the form of the distribution when you look at the whole graph.

Symmetric vs. skewed

• Symmetric: the left and right sides are roughly mirror images.
• Skewed right: a long tail extends to the right (toward larger values). Often happens with variables that have a lower bound at 0 and occasional large values (like income, wait times).
• Skewed left: a long tail extends to the left (toward smaller values). Often happens when there is a ceiling effect (like scores on an easy test where many are near 100).

A practical way to describe skew: ask which side has the “longer tail.”

Modality: peaks in the distribution

A mode is a peak in the distribution.

• Unimodal: one clear peak.
• Bimodal: two peaks (often suggests two subgroups mixed together).
• Multimodal: more than two peaks.

Why this matters: A bimodal distribution is a warning sign that one center/spread summary may be hiding two different behaviors.

Outliers and unusual features

An outlier is an observation that is unusually far from the rest of the data. Graphs can reveal outliers as isolated points (dotplot/stemplot) or isolated bars (histogram).

You should also look for:

• Gaps: intervals with no observations.
• Clusters: regions with many observations.

Why this matters: Outliers and gaps can strongly affect numerical summaries—especially the mean and standard deviation—and they can suggest issues like data entry errors or a special cause (for example, one student absent for half the week).

Center: a typical value

The center is a single value that represents a “typical” observation. Two common measures are:

• Mean: the arithmetic average.
• Median: the middle value when data are ordered.

Mean

If your data values are x_1, x_2, \dots, x_n, the sample mean is:

\bar{x} = \frac{1}{n}\sum_{i=1}^{n} x_i

• \bar{x} is the mean.
• n is the number of observations.
• x_i are the data values.

When mean is useful: roughly symmetric distributions with no extreme outliers.

Sensitivity: the mean is not resistant—outliers can pull it dramatically.

Median

The median is the midpoint of the ordered data (the 50th percentile). If n is odd, it’s the middle value. If n is even, it’s the average of the two middle values.

When median is useful: skewed distributions or data with outliers.

Resistance: the median is resistant—it does not move much when you add a very large or very small outlier.

Spread: how variable the data are

The spread tells you how far apart observations are.

Range (simple, but limited)

The range is:

\text{range} = \text{max} - \text{min}

It’s easy to compute, but it depends only on two values and is very sensitive to outliers.

Interquartile range (IQR): a resistant measure of spread

Quartiles split ordered data into four equal parts:

• Q_1: 25th percentile
• Q_3: 75th percentile

The interquartile range is:

\text{IQR} = Q_3 - Q_1

Why it matters: IQR measures the spread of the middle 50% of the data, so it is resistant to outliers and pairs naturally with the median.

Standard deviation: typical distance from the mean

The standard deviation measures (roughly) a typical distance of observations from the mean. The sample standard deviation is:

s = \sqrt{\frac{1}{n-1}\sum_{i=1}^{n} (x_i - \bar{x})^2}

• s is the sample standard deviation.
• The denominator n-1 is used for sample standard deviation (AP Statistics treats this as the standard formula for a sample).

How to interpret it: If s is large, observations tend to be far from the mean; if s is small, observations cluster near the mean.

Important: Like the mean, s is not resistant. Outliers inflate it.

Putting it together: matching summaries to shape

A common AP Statistics principle:

• For symmetric distributions with no outliers, describe center/spread with mean and standard deviation.
• For skewed distributions or with outliers, describe center/spread with median and IQR.

This isn’t a rule you mindlessly apply; it’s about choosing summaries that represent the distribution fairly.

Worked example: describing a distribution using SOCS

Suppose the variable is “minutes to finish a 5K” for a group of runners. A histogram shows:

• Most runners between 22 and 32 minutes
• A long right tail up to about 55 minutes
• One isolated bar around 55 minutes

A strong SOCS description might look like this (in context):

• Shape: The distribution is unimodal and skewed right (long tail toward slower times).
• Outliers/unusual: There appears to be a possible outlier around 55 minutes.
• Center: A typical finishing time is around 27–29 minutes (a median would be appropriate because of right skew).
• Spread: Most times fall roughly between 22 and 32 minutes, with an overall range up to about 55 minutes; the IQR would be a good numerical spread summary.

Notice what makes this good: it references the actual variable and units (minutes), uses appropriate shape language, and chooses center/spread measures that match the skew.

Example: how outliers affect mean vs. median

Consider the five values:

10, 11, 11, 12, 50

• The median is 11 (middle value).
• The mean is:

\bar{x} = \frac{10+11+11+12+50}{5} = 18.8

The mean is pulled upward by the outlier 50, while the median stays near where most values are. This is why you often prefer median/IQR for skewed data.

Writing good descriptions (what graders look for)

AP Statistics responses are graded for communication as much as computation. A good description of a quantitative distribution:

• Uses context (“test scores in points,” not just “values”).
• Uses shape language correctly (skewed left/right, symmetric, unimodal/bimodal).
• Identifies unusual features (outliers, gaps, clusters).
• Gives a reasonable center (with a number and units when possible).
• Gives a reasonable spread (with a number and units when possible).

A useful habit: when you state a number, attach units immediately (minutes, dollars, points). Many errors happen when students give correct statistics but unclear context.

Common misconceptions to avoid

• “Skewed means there’s an outlier.” Not necessarily. Skew describes an overall tail pattern; you can have skew without a single extreme point.
• “The mode is always the best center.” The mode can be useful, but AP Statistics typically emphasizes mean/median for center.
• “Standard deviation is the average distance from the mean.” It’s close in spirit, but it’s computed from squared deviations and a square root, so interpret it as a typical distance, not a literal average deviation.
• Mixing up which way skew goes: Skew is named for the direction of the tail, not where the bulk of the data is.

Exam Focus

• Typical question patterns:
  • “Describe the distribution” of a quantitative variable from a graph using SOCS.
  • Decide whether mean/SD or median/IQR is more appropriate and justify using shape/outliers.
  • Interpret a numerical summary (for example, explain what a standard deviation means in context).
• Common mistakes:
  • Listing SOCS words without tying them to the actual graph or context (no numbers/units).
  • Saying “skewed” without specifying left or right.
  • Using mean/SD for clearly skewed data with outliers, or claiming the median is affected heavily by an extreme value.