statistical reports vocab

Lurking/confounding variable

 A confounding variable is an outside influence that affects both the independent variable and the dependent variable, making it harder to determine the actual relationship between the two variables.

In simple terms, it’s a “hidden” factor that can mess up your experiment or study by giving you misleading results.

Randomisation

In experiments, participants are randomly assigned to groups. This helps ensure that confounding variables are equally distributed across groups, so their influence is minimised.

Restriction

Limiting the study to a specific subgroup to eliminate a confounding variable.

Matching

Researchers pair participants with similar values of a potential confounding variable across groups.

Stratification

Analysing results separately within different levels (strata) of the confounding variable.

Bias

A systematic error that causes the results of a study or survey to be inaccurate or misleading.

In statistical samples, this happens when the sample doesn’t accurately represent the population you’re trying to study.

This leads to results that are systematically off, and not just by chance, but in a particular direction.

Selection bias

Occurs when the sample is not randomly selected, and doesn’t reflect the target population.

Non-response bias

Happens when people who don’t respond to a survey differ significantly from those who do.

Measurement/response bias

Arises from how the data is collected — such as leading questions, or faulty instruments.

Voluntary response bias

When participation is optional, and only people with strong opinions respond.

Undercoverage bias

When some groups in the population are left out or under-represented.

Survey

A method of collecting information from a group of people — usually by asking them a set of questions. They are in the form of long questionnaires with a series of open and close ended questions. Each query gathers in-depth insights about audience preferences, needs, and behaviours from a specific market segment.

Multiple questions, often more detailed; used to collect broader information, attitudes, or behaviours; can be qualitative, quantitative, or both.

They are more like a short documentary.

Poll

A method of collecting information that involves asking a single query or a very limited set of questions. Questions are quick to the point, and allow you to draw conclusions about the opinions and attitudes of a larger population.

Usually one question, maybe two; designed to gauge quick opinions on a single topic; often used in real-time settings; fast and simple questions.

They are more like snapshots.

Numerical/quantitative data

Data in the form of numbers. These numbers can be a count of objects, number, or any data which uses numbers, and it is always measurable.D

Discrete data

Data that represents a set of countable items that are countably finite. It always takes up whole number values.

Continuous data

Data that can be described as everything in-between an interval. Often measurements.

Qualitative/categorical data

Non-numerical data that uses description or vocabulary of a variable (e.g. colours, flavours, ethnicity, etc). It represents types of data which may be divided into groups.

Variables of interest

The specific characteristic or property you’re trying to learn about or measure (it is the response variable).

Population of interest

The entire group that you want to learn about or draw conclusions from.

Causation

When one variable directly causes a change in the other.

Causal claim

A conclusion as to what caused a change in the response variable. A statement that says one thing directly causes another.

Controlled experiments, elimination of other possible explanations (lurking variables), and strong evidence of a direct effect are required.

Correlation

When two variables are related, and change together in a pattern. The variables move together, but not necessarily because of each other.

Explanatory variable (independent variable)

The variable you change or think is the cause. It helps explain or predict the response.

Plotted on the x-axis.

Response variable (dependent variable)

The variable you measure. It responds to changes in the other variable.

Plotted on the y-axis.

Random sample

A group selected from a population where every individual has an equal chance of being chosen. Decreases bias in results, and is more likely to represent the whole population.

Sampling frame

The list or source used to choose the sample from the population; the part of the population where the sample is extracted from.