Types of Data

Why This Matters

In Probability and Statistics, everything starts with understanding what kind of data you're working with. The type of data determines which statistical methods you can use, which graphs are appropriate, and which summary statistics make sense. A common exam mistake is applying the wrong statistical test because a student misidentified their data type.

Data classification works like a decision tree: first, is it qualitative or quantitative? If qualitative, is it nominal or ordinal? If quantitative, is it discrete or continuous, and does it have a true zero? These distinctions matter because they unlock or restrict what you can do mathematically. Don't just memorize definitions; know what operations each data type permits and why.

---

Qualitative Data: Describing Categories

Qualitative data describes attributes or characteristics rather than quantities. The key principle: you can count how many fall into each category, but you can't perform meaningful arithmetic on the categories themselves.

Nominal Data

• No inherent order or ranking. Categories are simply different from one another with no "greater than" or "less than" relationship.
• Only mode is meaningful as a measure of central tendency. Mean and median cannot be calculated because there's no way to order or average labels.
• Examples: blood type (A, B, AB, O), eye color, zip codes. Even though zip codes are numbers, arithmetic on them is meaningless (averaging two zip codes tells you nothing), so they're nominal.

Ordinal Data

• Categories have a meaningful order, but the intervals between ranks are not necessarily equal. The gap between "strongly disagree" and "disagree" might not be the same size as the gap between "agree" and "strongly agree."
• Median is appropriate, but mean is problematic because you can't assume equal spacing between levels.
• Examples: survey responses (strongly disagree to strongly agree), letter grades (A, B, C, D, F), and socioeconomic status categories (low, middle, high).

---

Quantitative Data: Measuring Amounts

Quantitative data represents numerical values where arithmetic operations can be meaningful. The distinguishing feature: you can add, subtract, and calculate means with these values.

Discrete Data

• Takes only specific, countable values, typically integers representing counts of items or occurrences.
• Cannot be subdivided into smaller meaningful units. You can't have 2.7 students in a classroom.
• Examples: number of siblings, dice rolls, defective items in a batch. The test is simple: if you're counting things, it's discrete.

Continuous Data

• Can take any value within a range. Theoretically infinite precision is possible.
• Measured rather than counted, limited only by the precision of your measuring instrument.
• Examples: height (you could be 170.284 cm tall), time, blood pressure. The value 5.2847 seconds is just as valid as 5 seconds.

---

Measurement Scales: What Math Is Allowed?

The distinction between interval and ratio data determines which mathematical operations produce meaningful results. This matters because ratios and percentages only make sense when zero means "none."

Interval Data

• Equal intervals between values, but no true zero point. Zero doesn't mean the absence of the quantity.
• Ratios are not meaningful. $40°F$ is not "twice as hot" as $20°F$. You can see why if you convert: $40°F \approx 4.4°C$ and $20°F \approx -6.7°C$. The "twice as much" relationship disappears.
• Examples: temperature in Celsius or Fahrenheit, calendar years, IQ scores. You can say how much more, but not how many times more.

Ratio Data

• A true zero point exists, meaning zero represents complete absence of the quantity.
• All mathematical operations are valid. You can meaningfully say someone earning $\$80{,}000$ makes twice as much as someone earning $\$40{,}000$.
• Examples: weight, height, age, income, distance. Most physical measurements fall here.

---

Data Collection Structure: When Was It Gathered?

How data is collected over time affects which analyses are appropriate. Time structure in your data determines whether you're comparing groups or tracking change.

Cross-Sectional Data

• A snapshot at a single point in time across multiple subjects or groups.
• Useful for comparing different populations, but cannot establish causation or trends over time.
• Examples: a survey of voter preferences before an election, or comparing test scores across schools in one semester.

Time Series Data

• Collected repeatedly over time, often at regular intervals from the same source.
• Reveals trends, cycles, and seasonal patterns, making it essential for forecasting.
• Examples: daily stock prices, monthly unemployment rates, annual GDP. Any data tracked chronologically fits here.

---

Quick Reference Table

---

Self-Check Questions

1. A researcher records the number of text messages each student sends per day. Is this discrete or continuous data? What if they recorded the time spent texting instead?

2. Which two data types both involve categories but differ in whether order matters? Give an example of each from a medical context.

3. Temperature measured in Kelvin has a true zero (absolute zero). Is Kelvin temperature interval or ratio data? How does this differ from Celsius?

4. A study compares income levels across five countries in 2024. Another study tracks one country's income from 2000 to 2024. Classify each data structure and explain what conclusions each can support.

5. A survey asks respondents to rate their satisfaction as very dissatisfied, dissatisfied, neutral, satisfied, or very satisfied. What type of data is this? Can you calculate a meaningful average? Justify your answer.