Module 4 · Statistics Foundation

Statistical inference.

This module explains how statisticians move from sample data to population conclusions. Students learn sampling distributions, standard error, confidence intervals, hypothesis tests, p-values, power, sample size and method selection. The emphasis is on careful reasoning, not mechanical testing.

Start Lesson 4.1 →View all lessons

Lessons

14–15 hrs

Study time

Coding

Foundation → Advanced

Level

What this module builds

The reasoning system behind statistical evidence.

Descriptive statistics summarise what was observed. Probability describes uncertainty. Statistical inference combines both: it uses probability models to decide what sample evidence says about a wider population.

Sampling variability

Inference begins with the idea that different samples produce different estimates, even from the same population.

Uncertainty around estimates

Standard error and confidence intervals show how much uncertainty surrounds a sample statistic.

Evidence against a model

Hypothesis tests compare observed data with what would be expected if a null model were true.

Decision errors

Statistical decisions can be wrong. Students learn Type I error, Type II error, power and their design implications.

Study design

Sample size, variability, allocation, dropout and bias control shape the strength of statistical evidence.

Method choice

The final lesson brings the module together by matching research questions, data types and assumptions to suitable inference methods.

By the end

Students should be able to interpret evidence responsibly.

Explain why sample statistics vary from sample to sample.

Define and interpret standard error.

Construct and interpret confidence intervals correctly.

Explain confidence level using long-run coverage.

State null and alternative hypotheses clearly.

Calculate and interpret test statistics and p-values.

Distinguish Type I error, Type II error and power.

Explain why statistical significance is not practical importance.

Plan sample size using precision and power logic.

Choose appropriate inference methods for common study questions.

Inference reasoning workflow

Ask six questions before reporting evidence.

Good inference is not just formula selection. Students must define the parameter, understand the estimator, check assumptions, quantify uncertainty, make careful decisions and judge practical meaning.

What is the target parameter?

Inference must begin with the population quantity being estimated or tested.

What is the estimator?

Identify the statistic calculated from the sample, such as a mean, proportion or difference.

What assumptions are being made?

Independence, sample size, distribution shape and measurement quality affect validity.

How much uncertainty remains?

Use standard error, confidence intervals and design context to judge precision.

What decision is justified?

Hypothesis tests support reject or fail-to-reject decisions, but not absolute proof.

Does the result matter?

Statistical significance should be interpreted with effect size and practical importance.

Formula map

Inference formulae all come from sampling variability.

Confidence intervals, hypothesis tests, power and sample size calculations are connected by one central idea: how much a sample statistic varies from sample to sample.

Standard error

SE(x̄) = σ / √n

Measures how much sample means vary across repeated samples.

Confidence interval

estimate ± critical value × SE

Turns a point estimate into a range of plausible parameter values.

Test statistic

z = (estimate − null value) / SE

Measures distance from the null in standard-error units.

Power

Power = 1 − β

Probability of detecting a specified real effect.

Mean sample size

n = (z*σ / ME)²

Used when planning precision for a confidence interval around a mean.

Two-group SE

SE = σ√(1/n₁ + 1/n₂)

Shows why balanced group allocation is often efficient.

Common inference traps

This module teaches careful interpretation.

Many statistical mistakes come from overinterpreting p-values, ignoring uncertainty or assuming large datasets automatically produce valid conclusions.

Large n does not remove bias

A large biased sample can produce a very precise but misleading estimate.

Small p-values are not effect sizes

A tiny p-value can occur for a tiny effect if the sample size is very large.

Non-significant is not no effect

CapstoneMethod selection

Choosing the right inference method

A capstone lesson on choosing between t-tests, proportion methods, chi-square tests, rank-based methods and correct reporting.

Method choiceAssumptionsReporting

Open lesson

Start →

How to study this module

Always connect the formula to the research question.

Inference is not about memorising procedures. Before applying a method, identify the parameter, estimator, sampling assumption, standard error, uncertainty statement and practical meaning. This habit prevents mechanical and misleading interpretation.

Module completion

Ready for modelling and applied statistical decisions.

After this module, students should understand how sample evidence becomes statistical evidence. They will be ready to study relationships, regression, model assumptions and applied reporting with a stronger inference foundation.

Back to course →