Difficulty: Introductory | Prerequisites: None. This is Week 1 material.
This is the foundation of everything else in MATH 399N. Before you can run a hypothesis test or build a regression model, you need to know what kind of data you are working with, how it was collected, and what you are allowed to do with it mathematically. The concepts here (population vs sample, measurement levels, variable types) reappear in every later topic, from confidence intervals to ANOVA. If you are coming in cold, start here.
Statistics is about collecting, organising and interpreting data to make decisions. Data splits into qualitative (labels) and quantitative (numbers), and the four measurement levels (nominal, ordinal, interval, ratio) determine which mathematical operations make sense. Every study involves variables that play different roles, and recognising those roles is half the battle in designing or critiquing research.
Data
Information coming from observations, counts, measurements, or responses. In simple terms, data is the raw material you feed into any statistical analysis.
Statistics
The science of collecting, organising, analysing and interpreting data in order to make decisions. Think of it as the toolkit for turning raw numbers (or labels) into useful conclusions.
Population
The collection of all outcomes, responses, measurements or counts that are of interest. In simple terms, it is the entire group you want to draw conclusions about. Look for keywords: "all", "every", "each".
Sample
A subset of the population. Think of it as the smaller slice you can feasibly measure when measuring everyone is impractical.
Parameter
A number that describes a population characteristic. (Mnemonic: Parameter and Population both start with P.)
Statistic
A number that describes a sample characteristic. (Mnemonic: Statistic and Sample both start with S.)
Qualitative (categorical) data
Attributes, labels or non-numerical entries such as marital status, telephone number, SSN, political preference. Think of it as data that labels or categorises rather than measures.
Quantitative (numerical) data
Numerical measurements or counts. Can be discrete (counting) or continuous (measuring). Examples: salary, weight, age.
Explanatory variable
The variable that attempts to explain or influence changes in another variable. Also called the independent variable, input variable, or predictor variable.
Response variable
The variable that is affected by the explanatory variable. Also called the dependent variable, outcome variable, or output variable.
Lurking variable
An additional variable that can cloud a study's outcome. Also called a confounding variable, confounding factor, or confounder.
The population/sample distinction runs through the entire course. Nearly every exam question that asks you to identify one or the other hinges on a single word:
"All", "every", "each" in the problem statement points to a population.
"A few", "some", "200 of the 10,000" points to a sample.
Once you know which you are looking at, the numerical summary follows:
A number computed from the population is a parameter.
A number computed from the sample is a statistic.
The mnemonic is straightforward: Parameter goes with Population, Statistic goes with Sample.
Every data value you encounter is either qualitative or quantitative. The distinction matters because it determines which statistical tools apply.
Qualitative (categorical) data
Consists of labels, names, or attributes.
You cannot do arithmetic on it in a meaningful way. Averaging phone numbers, for instance, produces nonsense.
Examples: marital status, telephone number, SSN, political preference, favourite sport.
Quick test: if summing or averaging the values would be meaningless, the data is qualitative.
Quantitative (numerical) data
Consists of numerical measurements or counts.
Arithmetic operations (mean, standard deviation) are meaningful.
Splits further into two sub-types:
Discrete: data you get by counting (number of students in a class, number of defective items).
Continuous: data you get by measuring (weight, height, temperature, time).
Examples: salary, weight, age.
A useful mnemonic from the source material: quaLitative = Labels, quaNtitative = Numerical.
The four levels form a hierarchy from weakest to strongest. Each higher level inherits the properties of the levels below it and adds something new.
1. Nominal (weakest)
Qualitative data only.
Names, labels, or categories with no inherent order.
No arithmetic is possible.
Examples: married/divorced/widowed, eye colour, jersey numbers (the number on a shirt is a label, not a quantity).
2. Ordinal
Can be qualitative or quantitative.
Data can be arranged in order or rank (1st to last), but the differences between ranks are not meaningful or consistent.
You know that 1st place is better than 2nd, but you cannot say how much better.
Examples: top-5 sports teams, satisfaction ratings (poor/fair/good/excellent), class rank.
3. Interval
Quantitative data only.
Data can be ordered, and the differences between values are meaningful and consistent.
There is no true (inherent) zero. Zero is just a position on the scale, not the absence of the thing being measured.
Examples: temperature in Celsius or Fahrenheit (0°C does not mean "no temperature"), calendar years (2000 AD, 500 BC).
4. Ratio (strongest)
Quantitative data only.
Data can be ordered, differences are meaningful, and there is a true zero representing the complete absence of the quantity.
Ratios between values are meaningful: 20 kg is twice as heavy as 10 kg.
Examples: weight, height, age, salary, distance.
A common exam question gives you a data example and asks you to name its measurement level. The key test: does zero mean "none"? If yes, it is ratio. If zero is just a point on the scale, it is interval.
When a study examines how one thing affects another, the variables take on specific roles.
Explanatory variable (independent / input / predictor)
The variable the researcher manipulates or uses to explain changes in the outcome.
Different values of the explanatory variable are called treatments.
Example: in a study on fertiliser and crop yield, the type of fertiliser is the explanatory variable.
Response variable (dependent / outcome / output)
The variable that changes as a result of the explanatory variable.
Example: crop yield is the response variable in the fertiliser study.
Lurking variable (confounding variable / confounder)
An additional variable, not part of the study design, that can distort or cloud the results.
A lurking variable can make it look as though the explanatory variable caused a change when it did not, or hide a real effect.
Example: if plots that received fertiliser also happened to get more sunlight, sunlight is a lurking variable.
Being able to identify which variable is which is a recurring exam skill. When reading a study description, ask: what is being changed on purpose (explanatory), what is being measured as a result (response), and what else might be influencing the outcome (lurking)?
Students often think that any number makes data quantitative. Telephone numbers and SSNs are numerical in form but qualitative in nature, because averaging them is meaningless.
Ordinal data tempts people into computing means. A mean of satisfaction ratings (1 = poor, 5 = excellent) treats the gaps between categories as equal, which ordinal data does not guarantee.
The difference between interval and ratio trips people up regularly. The test is simple: does zero mean the complete absence of the thing? Temperature in Celsius: no (0°C is a real temperature). Weight in kilograms: yes (0 kg means nothing is there). Celsius is interval; weight is ratio.
Students sometimes confuse a lurking variable with a response variable. The lurking variable is not part of the study design at all. If the researchers are measuring it as an outcome, it is a response variable, not a lurking one.
⚠️ Expect questions that give a scenario and ask: "Is this a population or a sample?" Look for the keywords (all/every/each vs a subset).
⚠️ Identifying the level of measurement (nominal, ordinal, interval, ratio) from a given example is a near-certain exam item. Practise the zero test.
⚠️ Variable identification questions (explanatory vs response vs lurking) appear regularly, often embedded in short study descriptions.
⚠️ The distinction between discrete and continuous quantitative data is frequently tested. Counting = discrete, measuring = continuous.
True or false: A telephone number is quantitative data because it contains digits. Answer: False. It is qualitative because arithmetic on it is meaningless.
Fill in the blank: A number that describes a population characteristic is called a __________. Answer: Parameter.
True or false: Ordinal data allows you to calculate meaningful differences between ranks. Answer: False. You can rank ordinal data, but the differences between ranks are not guaranteed to be equal or meaningful.
Fill in the blank: The measurement level that has a true zero is called __________. Answer: Ratio.
True or false: The number of students in a classroom is an example of continuous data. Answer: False. It is discrete, because you count whole students.
Q: A researcher surveys every student at a university about their favourite study method. Is "every student at the university" a population or a sample?
A: It is the population, because it includes all individuals of interest (signalled by "every").
Q: A report states that the average income of 500 surveyed households is £42,000. Is £42,000 a parameter or a statistic?
A: It is a statistic, because it describes a sample (500 households), not the entire population.
Q: Classify the following as qualitative or quantitative: (a) hair colour, (b) number of pets owned, (c) postal code.
A: (a) Qualitative, (b) quantitative (discrete), (c) qualitative (even though postal codes contain digits, averaging them is meaningless).
Q: Temperature in Fahrenheit is which level of measurement, and why?
A: Interval. Differences between temperatures are meaningful and consistent, but 0°F does not represent the complete absence of temperature.
Q: In a study testing whether a new drug lowers blood pressure, identify the explanatory variable, the response variable, and a possible lurking variable.
A: Explanatory variable: whether the patient receives the drug or a placebo. Response variable: blood pressure. Possible lurking variable: the patient's diet or exercise habits, which could also affect blood pressure independently.
The population vs sample distinction feeds directly into inferential statistics: the entire point of confidence intervals and hypothesis tests is to say something about a population based on a sample.
Measurement levels come back when choosing which statistical test to run. Nominal and ordinal data call for non-parametric tests (chi-square, for instance), while interval and ratio data open the door to means-based tests (t-tests, ANOVA).
Understanding variable types is essential for regression analysis later in the course, where you will formally model the relationship between explanatory and response variables.
data types, qualitative data, quantitative data, categorical data, numerical data, nominal ordinal interval ratio, levels of measurement, measurement scales, population vs sample, parameter vs statistic, discrete vs continuous, explanatory variable, independent variable, predictor variable, response variable, dependent variable, outcome variable, lurking variable, confounding variable, confounder, MATH 399N, intro statistics, ASU statistics