Quantitative Research Methods

Introduction

Quantitative Research Methods help you turn carefully defined questions into measurable evidence. In quantitative research, you collect numerical or countable data, examine patterns with statistical methods, and judge how strongly the evidence supports a conclusion. The method is used across statistics, psychology, sociology, economics, education, biology, business, public policy, and many other fields.

This aiMOOC is designed for Grades 11–13. You do not need advanced mathematics, but you should be comfortable with percentages, averages, graphs, and basic algebra. The central goal is not to calculate blindly. It is to understand how a research question, design, sample, measurement, analysis, and conclusion fit together.

A useful starting principle is this: good statistics cannot rescue weak research design. A large dataset may still be misleading if the sample is biased, the variables are badly measured, important confounders are ignored, or the analysis does not match the question.

Learning Goals

By the end of the course, you should be able to formulate a focused research question, distinguish populations from samples, operationalize variables, compare survey, observational, and experimental designs, recognize common sources of bias, summarize data, choose informative graphs, interpret correlation and simple regression, explain the logic of confidence intervals and hypothesis tests, and communicate limitations responsibly.

You should also be able to plan a small quantitative study in a school context while respecting privacy, consent, data minimization, and other relevant ethical rules.

From Questions to Testable Claims

Research Questions

A quantitative research question identifies measurable features of a phenomenon. Strong questions are specific enough to guide data collection and analysis. For example, instead of asking “Does sleep matter?”, you might ask: “Among students in a given school year, how is average sleep duration on school nights associated with a self-reported concentration score?”

That question identifies a group, two measurable variables, and a possible relationship. It does not yet claim that sleep causes concentration to change. The wording matters because the design determines which conclusions are justified.

Common quantitative question types include description such as “What is the typical commute time?”, comparison such as “Do two classes differ in average quiz scores?”, association such as “Are study time and quiz score related?”, and causal questions such as “Does a particular intervention change an outcome?”

Hypotheses

A hypothesis is a testable statement. In many studies, a substantive hypothesis predicts a direction or difference, while statistical analysis may use a null hypothesis that represents a reference position such as no difference or no association.

A useful hypothesis must connect to observable evidence. “Students learn better with good teaching” is too vague. “Students randomly assigned to Method A will have a higher mean score on the same post-test than students randomly assigned to Method B” is much more testable because the treatment, outcome, comparison, and analysis are clearer.

Variables and Measurement

Variables

A variable is a characteristic that can take different values. Examples include age, number of books read, reaction time, temperature, test score, transport mode, and agreement with a statement.

In an experiment, the independent variable is the factor deliberately changed or assigned by the researcher, and the dependent variable is the measured outcome. In an observational study, it is often clearer to speak of explanatory and response variables because the researcher is not necessarily manipulating anything.

A categorical variable places cases into groups, such as transport mode. A quantitative variable records numerical amounts, such as travel time in minutes. Quantitative variables may be discrete, such as a count of absences, or continuous, such as measured height.

Operationalization

Operationalization means defining exactly how an abstract idea will be measured. Concepts such as stress, political trust, motivation, well-being, or digital distraction cannot be analyzed until you specify indicators and procedures.

Suppose you want to study “academic engagement.” You might operationalize it using attendance, assignment completion, time on task, or responses to a validated questionnaire. These are not identical measures, so your choice affects what your results actually mean.

Good operational definitions are precise, repeatable, and connected to the concept. They also state units, timing, scoring rules, and how missing or unusual responses will be handled.

Levels of Measurement

A nominal variable consists of categories without an inherent order. An ordinal variable has ordered categories, but the distance between categories is not necessarily equal. A common single Likert-style item such as “strongly disagree” through “strongly agree” is ordinal. Interval measurements have meaningful equal differences but no true zero in the ratio sense, while ratio measurements have equal differences and a meaningful zero.

The measurement level influences which summaries and statistical procedures are sensible. You should never choose a statistical test only because software offers it; first consider what your values actually represent.

Research Designs

Surveys

A survey collects structured responses from participants. Surveys can efficiently describe attitudes, behaviors, experiences, or characteristics, but wording and sampling strongly affect the results.

A good questionnaire uses clear, neutral wording, avoids asking two things in one question, provides response options that fit the construct, and is tested with a small pilot group before full data collection. Leading questions, vague time frames, and unbalanced response scales can introduce measurement bias.

When asking sensitive questions, collect only what you need. Avoid unnecessary names, contact details, or combinations of variables that could identify individuals. Follow school rules and applicable legal and ethical requirements.

Observational Studies

In an observational study, researchers measure variables without assigning the main exposure or treatment. These studies are valuable when manipulation would be impossible, unethical, or unnecessary.

Observational data can reveal patterns and associations, but causal interpretation is difficult because groups may differ in other ways. A confounder is a variable related to both the explanatory variable and the outcome that can create, hide, or distort an association.

For example, if students who spend more time in a library also score higher, the library time itself might matter, but prior motivation, access to resources, or course difficulty could also influence both variables.

Experiments

In a controlled experiment, the researcher assigns conditions and measures outcomes. Random assignment helps create comparable groups by distributing known and unknown characteristics across conditions by chance. A control or comparison condition provides a reference.

Blinding, standardized procedures, and pre-specified outcomes can reduce bias. In some settings, a placebo may help isolate the effect of an active treatment from expectations, but school projects should only use procedures that are safe, ethical, and appropriate for students.

Random assignment supports stronger causal inference than simple observation when the study is well designed and implemented. It is different from random sampling: assignment addresses comparability between conditions, while sampling addresses how well a sample represents a population.

Sampling and Generalization

Population and Sample

The population is the full group about which you want to make a claim. The sample is the subset you actually observe. The target population must be stated precisely. “Teenagers” may be too broad if your data come only from one school or one voluntary online group.

In a simple random sample of a fixed size, every possible sample of that size has an equal chance of selection, which gives each eligible member the same inclusion chance. Stratified sampling first divides the population into relevant subgroups and then samples within those groups. Cluster sampling selects natural groups, such as classes, and studies all or some members within selected clusters. Systematic sampling uses a regular selection rule after a random start.

Convenience samples and voluntary response samples are often easier to obtain, but they can be strongly biased because inclusion is connected to availability or willingness to participate.

Sampling Bias and Sampling Variability

Sampling bias is a systematic problem in how cases enter the sample. It can occur through undercoverage, nonresponse, voluntary response, or other selection processes. Increasing the size of a biased sample does not automatically remove the bias.

Sampling variability is different. Even well-designed random samples differ from one another by chance. Larger random samples usually provide more precise estimates, all else equal, because random fluctuations tend to average out.

Representativeness depends more on how the sample is selected than on the percentage of the population sampled. A carefully selected sample of a few hundred people can sometimes represent a large population better than a much larger convenience sample.

Reliability, Validity, Bias, and Ethics

Reliability

Reliability concerns consistency. A reliable measurement procedure tends to produce similar results when the measured construct has not changed and conditions are comparable. Reliability may be examined through repeated measurements, agreement between observers, or consistency among items intended to measure the same construct.

High reliability is useful but not sufficient. A scale that is always five units too high may be consistent yet inaccurate.

Validity

Validity concerns whether the evidence and interpretation support the intended conclusion. Measurement validity asks whether an instrument captures the construct it is supposed to measure. Internal validity concerns whether a causal conclusion is credible within a study. External validity concerns how far findings can reasonably generalize to other people, places, times, or settings.

Validity is not a single switch that is either on or off. It is a question of how well the design, measurement, analysis, and interpretation support a particular claim.

Bias and Researcher Decisions

Bias can enter through sampling, wording, measurement, missing data, selective reporting, or analysis choices. Researchers reduce avoidable bias by planning in advance, documenting procedures, using appropriate controls, checking data quality, and reporting results even when they do not support the preferred hypothesis.

Transparency matters. Keep a codebook, record any data-cleaning decisions, and distinguish analyses planned before seeing the results from exploratory analyses developed afterward.

Ethics and Privacy

Quantitative research involving people requires respect for participants. Obtain appropriate informed consent or assent, minimize risks, allow voluntary participation, protect confidentiality, and collect only necessary data. For minors, schools and researchers may have additional responsibilities involving guardians, institutional approval, and local law.

Do not publish identifiable student-level data. Small subgroups can sometimes make people identifiable even when names are removed, so aggregate or suppress data when needed. Ethical quality is part of research quality, not an optional extra.

Descriptive Statistics

Center

Descriptive statistics summarize what the observed data look like. The mean is the arithmetic average. The median is the middle value after ordering the data. The mode is the most frequent value or category.

The mean uses every numerical value and is sensitive to extreme observations. The median is often more resistant to outliers and may better represent the center of a strongly skewed distribution. The mode can be useful for categorical data.

Spread

Two datasets can have the same center and very different variability. The range is the maximum minus the minimum. The interquartile range describes the spread of the middle half of ordered observations. The standard deviation describes how far values typically vary around the mean, using squared deviations in its calculation.

Spread matters because an average without information about variability can hide major differences. When comparing groups, consider both center and distribution.

Distributions and Outliers

A histogram displays the distribution of a quantitative variable by grouping values into intervals. Its shape can show symmetry, skewness, multiple peaks, gaps, and unusual values.

A box plot provides a compact summary using the median, quartiles, and a rule for flagging potential outliers. Different distributions can sometimes produce similar box plots, so use a histogram or raw data view when distribution shape matters.

An outlier is an observation unusually far from the main pattern. Do not remove an outlier just because it is inconvenient. First check for data-entry mistakes, measurement errors, unusual but genuine cases, and whether the analysis is sensitive to the observation.

Data Visualization

A graph is part of the analysis, not decoration. Match the graph to the variable type and question. Bar charts compare categorical counts or summaries. Histograms show distributions of quantitative variables. Scatterplots show relationships between two quantitative variables. Line graphs are often useful when an ordered variable such as time is central.

Label axes, include units, use a scale that does not distort the message, and avoid unnecessary three-dimensional effects. When comparing groups, consistent scales make visual differences easier to judge honestly.

A visualization should help a reader answer a research question. Ask yourself: What does the graph reveal that a table of numbers does not?

Correlation and Regression

Correlation

A correlation describes the direction and strength of association between two quantitative variables. Pearson's correlation coefficient, usually written as r, ranges from -1 to +1. Values near +1 indicate a strong positive linear relationship, values near -1 indicate a strong negative linear relationship, and values near 0 indicate little linear association.

Correlation can be affected by outliers and can miss strong nonlinear relationships. Always examine a scatterplot before relying on a single coefficient.

Most importantly, correlation does not by itself establish causation. An association may reflect confounding, reverse direction, common causes, selection effects, or chance.

Simple Linear Regression

Simple linear regression models a linear relationship between an explanatory variable and a quantitative response. The fitted line can be written conceptually as:

predicted outcome = intercept + slope × explanatory value

The slope estimates how much the predicted outcome changes for a one-unit increase in the explanatory variable. The intercept is the predicted outcome when the explanatory variable equals zero, but it may have little practical meaning if zero is outside the observed range.

A residual is the observed outcome minus the predicted outcome. Residual patterns help you check whether a straight-line model is reasonable. Regression does not automatically turn an observational association into a causal effect.

Statistical Inference

From Samples to Populations

Statistical inference uses sample data to learn about a population while accounting for uncertainty. An estimate based on a sample is called a statistic; the corresponding unknown population quantity is a parameter.

Examples include using a sample proportion to estimate a population proportion or a sample mean to estimate a population mean. Because different random samples produce different estimates, inference should communicate uncertainty rather than report a single number as if it were exact.

Confidence Intervals

A confidence interval combines an estimate with a margin that reflects sampling uncertainty under a statistical model. A 95% confidence procedure is designed so that, under its assumptions and repeated sampling, about 95% of intervals produced by the procedure would contain the true parameter.

That long-run interpretation is important. After one interval has been calculated, the frequentist statement is not that there is a 95% probability that a fixed unknown parameter lies inside that specific interval.

Wider intervals indicate less precision. Larger random samples usually narrow intervals, while higher confidence levels usually widen them, all else equal.

Hypothesis Testing and p-Values

A hypothesis test compares observed data with what a specified null model would make plausible. The p-value is the probability, assuming the null hypothesis and the model used for the test, of obtaining a result at least as incompatible with the null hypothesis as the observed result.

A small p-value can be evidence against the null model, but it is not the probability that the null hypothesis is true. It also does not measure the size or practical importance of an effect.

Decisions should not depend on a p-value alone. Consider the study design, effect estimate, uncertainty, measurement quality, assumptions, sample size, and the consequences of errors.

Statistical and Practical Significance

A very small effect can become statistically detectable in a large sample, while an important effect may remain uncertain in a small sample. Therefore, report an effect size and a confidence interval when possible rather than only whether a threshold was crossed.

Practical significance asks whether the magnitude matters in the real context. A change of one point may be important on one scale and trivial on another. Interpretation requires subject knowledge as well as statistical reasoning.

A Reproducible Quantitative Workflow

A strong school-level quantitative project can follow a clear chain from question to evidence:

Research question: State the population, variables, comparison or relationship, and the type of conclusion you hope to make.
Operationalization: Define exactly how each variable will be measured and coded.
Research design: Decide whether the study is a survey, observational study, experiment, or another suitable design.
Sampling: Define how cases will be selected and identify likely coverage or nonresponse problems.
Ethics: Obtain required approvals and consent, minimize personal data, and protect confidentiality.
Data management: Create a codebook, preserve raw data, document cleaning decisions, and record missing values consistently.
Exploratory data analysis: Check distributions, unusual values, missingness, and visual patterns before formal testing.
Statistical analysis: Choose summaries and inferential methods that match the variables, design, and assumptions.
Interpretation: Separate association from causation, report uncertainty, and state limitations.
Communication: Present methods and results clearly enough that another learner could understand and, where possible, reproduce the analysis.

A reproducible workflow helps you notice where a conclusion depends on a choice. It also makes your work easier to review, improve, and replicate.

Reading Quantitative Research Critically

When you read a news report, research paper, or infographic, do not focus only on the headline result. Ask who was studied, how the sample was obtained, what was measured, which comparison was made, whether the design supports causal claims, and how uncertainty was reported.

Check whether the graph matches the data, whether percentages have a clear denominator, whether missing data are discussed, and whether multiple outcomes or repeated analyses could have increased the chance of a striking result.

A strong critical reader can say both what the evidence supports and what it does not support. Scientific caution is not weakness; it is part of accurate reasoning.

Interactive Tasks

Quiz: Test Your Knowledge

What is a sample in quantitative research? (A subset of a population that is actually observed) (!The complete group a researcher wants to understand) (!A rule for proving a causal relationship) (!A graph that displays numerical data)

What does operationalization do? (Defines how a concept will be measured) (!Guarantees that a sample is random) (!Turns every association into causation) (!Removes all missing values from a dataset)

What is the main purpose of random assignment in an experiment? (To create comparable treatment groups by chance) (!To guarantee that the sample represents a country) (!To increase the number of questionnaire items) (!To eliminate the need for a control condition)

Which measure of center is usually more resistant to extreme outliers? (The median) (!The mean) (!The range) (!The standard deviation)

What does a strong correlation by itself establish? (A strong statistical association) (!A proven causal effect) (!A perfectly representative sample) (!A measurement without error)

What does a p-value describe? (The probability of data this extreme or more under the null model) (!The probability that the null hypothesis is true) (!The size of an effect in practical units) (!The percentage of the population in the sample)

How should a frequentist 95 percent confidence procedure be interpreted? (It produces intervals that contain the true parameter about 95 percent of the time in repeated sampling) (!It proves that the study result is correct 95 percent of the time) (!It means 95 percent of individual data values lie inside every interval) (!It guarantees that a specific interval contains the parameter)

What does reliability primarily concern? (Consistency of measurement) (!Causal interpretation) (!Population size) (!Graph selection)

In a controlled experiment what is the dependent variable? (The measured outcome) (!The factor assigned by the researcher) (!The complete target population) (!The random sampling frame)

Which situation is most likely to create sampling bias? (Only highly motivated volunteers choose whether to respond) (!A random sample differs slightly from another random sample) (!A histogram uses equal width bins) (!A dataset contains both a mean and a median)

Memory Game

Population	Full group about which a study aims to draw conclusions
Sample	Subset of cases that is actually observed
Variable	Characteristic that can take different values
Reliability	Consistency of a measurement procedure
Validity	Degree to which evidence supports the intended interpretation
Confounder	Third factor related to both an explanatory variable and an outcome
Outlier	Observation unusually far from the main data pattern

Drag and Drop

Match the correct terms.	Topic
Arithmetic average	Mean
Middle ordered value	Median
Relationship graph	Scatterplot
Randomly assigned comparison	Experiment
Long-run uncertainty range	Confidence interval

Match each description to the research term, then explain one match in your own words.

Crossword Puzzle

Sample	What do you call the subset of a population that is actually studied?
Median	Which measure is the middle value in ordered data?
Bias	What word describes a systematic distortion in a study?
Scatterplot	Which graph shows the relationship between two quantitative variables?
Reliability	What term describes consistency in measurement?
Confounder	What third variable can distort an observed association?

LearningApps

Cloze Text

A quantitative study begins with a focused

that can be connected to measurable evidence. The full group of interest is the

. The observed subset is the

. Turning an abstract concept into a precise measurement procedure is called

. A consistent measurement is described as

. In an experiment, random

helps create comparable conditions. A histogram helps you examine the shape of a

. The linear association between two quantitative variables can be summarized by a

. A confidence interval communicates statistical

. A p-value is calculated under a specified

model. Statistical significance should be interpreted together with the size of the

. A careful conclusion must distinguish association from

.

Open-Ended Tasks

Easy

Research Question Workshop: Write three quantitative research questions about school life, identify the population and variables in each, and revise the strongest question so that it can be answered with measurable evidence.
Variable Card: Create a one-page visual card for one abstract concept such as motivation or stress that shows at least two possible operational definitions and explains what each definition captures or misses.
Graph Critique: Find a public graph in a news article or report, make an annotated copy, and explain whether the graph type, axes, labels, and scale support an accurate interpretation.
Mini Questionnaire: Draft five neutral survey items about a harmless school topic, test them with two classmates, and revise any wording that causes confusion or ambiguity.

Standard

Random Sampling Simulation: Use slips of paper, a spreadsheet, or code to draw repeated random samples from a known class dataset, compare the sample means, and create a short explanation of sampling variability.
Observational Data Study: Collect a small anonymous dataset with two non-sensitive quantitative variables, produce a scatterplot, calculate or estimate the direction of association, and discuss at least two possible confounders.
Researcher Interview: Interview a teacher, laboratory worker, analyst, or researcher about how quantitative data are collected and checked, then summarize the quality-control steps in a written report or short audio recording.
Data Story Video: Produce a two-minute video that explains one dataset using a graph, a measure of center, a measure of spread, and a clear statement about what cannot be concluded.

Advanced

Controlled Experiment Proposal: Design a safe and ethical randomized school experiment using a harmless classroom procedure, specify treatment and comparison conditions, define the outcome, and explain how random assignment would support a causal claim.
Regression Investigation: Analyze a suitable open dataset with two quantitative variables, fit or estimate a simple linear regression, interpret the slope and residuals, and explain why the result does or does not justify prediction or causation.
Replication Audit: Choose a published quantitative result, reconstruct its research question, sample, variables, design, and main analysis from the available report, then identify which details another researcher would need to reproduce the study.
Research Report and Defense: Complete a small quantitative project and present a structured report with question, method, ethics, data display, analysis, uncertainty, limitations, and a five-minute oral defense responding to critical questions.

Learning Assessment

Design Diagnosis: Given a short study description, identify whether it is a survey, observational study, or experiment and justify which kinds of claims the design can support.
Sampling Evaluation: Compare a convenience sample and a probability-based sample for the same research question, then explain how selection could affect generalization.
Measurement Review: Evaluate two operational definitions of the same concept and argue which one has stronger validity for a specified purpose.
Graph and Summary Choice: Given several variable types and distribution shapes, choose suitable graphs and measures of center and spread, and justify each choice.
Association Interpretation: Interpret a scatterplot and correlation while naming at least two reasons why an observed association may not be causal.
Inference Reasoning: Explain a confidence interval and p-value in context without claiming that either gives the probability that a fixed hypothesis or parameter is true.
Evidence-Based Conclusion: Write a short conclusion that integrates effect size, uncertainty, study design, limitations, and the population to which the result can reasonably apply.

Evidence of Learning

Knowledge: You can explain populations, samples, variables, operationalization, common research designs, bias, reliability, validity, descriptive statistics, correlation, regression, confidence intervals, and hypothesis testing.

Skills: You can turn a question into measurable variables, choose a suitable design, evaluate a sampling strategy, create and interpret graphs, summarize distributions, reason about uncertainty, and distinguish association from causal evidence.

Products: Strong evidence may include a research proposal, questionnaire, codebook, anonymized dataset, graph portfolio, statistical analysis, written report, presentation, or short explanatory video.

Transfer: You can use these ideas to judge claims in news, science, business, public policy, and everyday decision-making, especially when a numerical result looks impressive but the design or measurement is weak.

Research integrity: You can document decisions, acknowledge limitations, protect participant privacy, and communicate results without exaggerating certainty.

OERs on the Topic

The English Wikipedia article below provides an open overview of quantitative research. Use it as a starting point, then compare definitions and methods with the examples and tasks in this course.