Learning Statistics with jamovi: A Tutorial for Beginners in Statistical Analysis

Danielle J. Navarro and David R. Foxcroft’s Learning Statistics with jamovi connects research design, statistical reasoning and practical analysis. The tutorial moves from describing and preparing data to inference, regression, analysis of variance, factor analysis and Bayesian methods, with an emphasis on understanding what an output can justify.

How to study this book

Work with the example data while reading, but write a research question and identify the variables before opening an analysis menu. Keep a log of data cleaning, assumptions, chosen procedures and interpretations. Reproduce an example, then change one feature to test your understanding. Treat software output as evidence requiring explanation, not as an automatic conclusion.

Chapter and section guide

Chapter 1 — Why Do We Learn Statistics

The opening chapter explains why informal judgement alone is not enough for evaluating evidence. Belief bias and Simpson’s paradox demonstrate how plausible impressions can conflict with a careful analysis of data. Statistics is placed within the larger process of research rather than offered as a replacement for good questions or design. Pay particular attention to the distinction between an observed pattern and a defensible explanation.

Study task: Invent a simple example where combining two groups could obscure or reverse the pattern within each group.

Chapter 2 — A Brief Introduction to Research Design

Measurement, operationalisation, reliability and validity establish the meaning of the variables that later enter an analysis. Experimental and non-experimental designs support different kinds of inference, and confounding can undermine an apparently persuasive association. Distinguish the scale of measurement from the role of a variable as predictor or outcome. A sophisticated analysis cannot repair every weakness in how a study was conceived.

Study task: Design a small study, define its main variable operationally and identify one threat each to internal and external validity.

Chapter 3 — Getting Started with jamovi

The practical introduction covers the spreadsheet, variable setup, importing data, computed variables, analyses and modules. The important habit is to verify how the software has interpreted a dataset before calculating anything. A column containing numbers might represent categories rather than a continuous measure, and an import can preserve unexpected codes or missing values. Save a clean project with a record of your decisions.

Study task: Import a small dataset and audit variable types, labels and missing-value conventions before running an analysis.

Chapter 4 — Descriptive Statistics

Measures of centre and spread summarise different features of a distribution. Mean, median, variance, standard deviation and related quantities should be chosen with the shape and measurement scale in mind. Group summaries and standard scores extend the comparison, while skew and kurtosis warn against relying on one headline average. Always link a numerical summary to the observations it is intended to describe.

Study task: Add an extreme value to a small dataset and compare its effect on the mean, median and standard deviation.

Chapter 5 — Drawing Graphs

Histograms, boxplots, violin plots and bar graphs answer different descriptive questions. The chapter makes visualisation part of analysis: distribution shape, group differences and unusual observations can become visible before an inferential procedure is chosen. Select the graph for the variable and question, and inspect labelling and scale before exporting. A visually striking difference still needs an appropriate account of uncertainty.

Study task: Display one continuous variable in two suitable ways and explain what each view makes easier to notice.

Chapter 6 — Pragmatic Matters

Tabulation, logical expressions, transformations, recoding and filtering turn a raw dataset into the form needed for analysis. These operations can also introduce hidden mistakes, especially when a filter changes the sample or a recode merges meaningful categories. Preserve the original variables, document every transformation and compare counts before and after a change so that the final results remain traceable.

Study task: Create a recoded variable and a filter, then report precisely which observations are included and why.

Chapter 7 — Introduction to Probability

Probability provides a language for uncertainty and a foundation for later sampling arguments. Frequentist and Bayesian interpretations are distinguished, while distributions such as the binomial and normal supply concrete models. Separate a probability attached to a discrete outcome from a density associated with a continuous variable. Check the assumptions of the proposed probability model before using its numerical results.

Study task: Contrast a binomial count with a continuous normal measurement and explain how probabilities are obtained in each case.

Chapter 8 — Estimating Unknown Quantities from a Sample

Samples, populations and sampling distributions connect observed statistics with unknown parameters. The law of large numbers and central limit theorem address different aspects of repeated sampling, while confidence intervals describe uncertainty under specified conditions. Bigger samples do not automatically remove selection bias. Learn to state what is estimated, how the sample was obtained and what the interval procedure means.

Study task: Distinguish a population distribution from the sampling distribution of the mean, then explain a confidence interval without assigning a frequentist probability to a fixed parameter.

Chapter 9 — Hypothesis Testing

Null and alternative hypotheses, test statistics, error rates and p-values form a conditional reasoning framework. Statistical significance is not a measure of effect importance, and a p-value is not the probability that the null hypothesis is true. Effect size, sample size and power influence what a study can detect. A good report states both the evidential result and its practical limitations.

Study task: Write a result statement that includes an effect estimate and uncertainty, and correct a mistaken interpretation of a p-value.

Chapter 10 — Categorical Data Analysis

Goodness-of-fit and association questions require methods suited to counts rather than continuous measurements. Chi-square procedures compare observed and expected frequencies, with assumptions that must be checked before interpretation. A contingency table can show an association without identifying its cause. Connect the statistical question to the arrangement of categories and report enough detail for the reader to understand the comparison.

Study task: Construct a two-way table, calculate the expected counts and explain whether the proposed chi-square analysis is appropriate.

Chapter 11 — Comparing Two Means

The chapter develops one-sample, independent-samples and paired comparisons, with alternatives and assumption checks relevant to the problem. The design determines whether observations are paired; this is not a choice to make after inspecting which test gives a smaller p-value. Interpret an estimated mean difference alongside its uncertainty and effect size, and consider the distributional conditions behind the chosen procedure.

Study task: Describe two studies with similar measurements but different pairing structures, and justify the analysis for each.

Chapter 12 — Correlation and Linear Regression

Correlation summarises a form of association, while regression models an outcome conditional on predictors. Scatterplots, coefficients, residuals and model checks help determine whether the fitted relationship describes the data adequately. Additional predictors change how a coefficient is interpreted, and association alone does not establish causation. Avoid extrapolating far beyond observed values or treating a fitted line as proof of an underlying mechanism.

Study task: Fit a simple relationship, inspect residuals and explain one reason the coefficient should not be read as a causal effect.

Chapter 13 — Comparing Several Means: One-Way ANOVA

ANOVA compares group means through a decomposition of variation. The overall test asks whether the means are all equal; it does not by itself identify every differing pair. Assumptions, planned comparisons or post-hoc procedures, and multiplicity therefore matter to the final interpretation. Keep the research question visible when deciding which follow-up comparisons are relevant.

Study task: Interpret a significant overall test without claiming that every group differs, then propose justified follow-up comparisons.

Chapter 14 — Factorial ANOVA

Multiple factors allow the analysis to examine main effects and interactions. An interaction means that the relationship involving one factor depends on another, so isolated averages can conceal the pattern of interest. Read cell means and interaction plots before summarising model terms. The chapter also develops the care needed when design, balance or repeated observations complicate a simple factorial picture.

Study task: Sketch a two-by-two interaction and explain why a main-effect summary alone would be misleading.

Chapter 15 — Factor Analysis

The chapter introduces methods for investigating patterns shared across measured variables, including principal components and factor-analytic approaches. Decisions about extraction, number of dimensions and rotation influence the interpretation, so a tidy loading table is not self-validating. Distinguish a mathematical summary of covariance from a justified claim about a psychological construct, and assess the quality of the measures supplying the data.

Study task: Compare a principal-component summary with a latent-factor interpretation and identify the extra claim made by the latter.

Chapter 16 — Bayesian Statistics

Bayesian inference combines prior information with a likelihood to obtain a posterior distribution, offering a different account of uncertainty from frequentist hypothesis testing. Model comparison and Bayes factors address relative evidence under stated specifications. Priors and modelling assumptions are part of the analysis and deserve scrutiny. The aim is not to replace one button with another but to understand the question each framework answers.

Study task: Explain how two defensible priors could lead to different posterior conclusions from a small dataset and propose a sensitivity check.

Applied assignment

Produce a short analysis report from a documented dataset. State the research question and design, show a suitable graph, justify one inferential procedure, check its assumptions and report the effect with uncertainty. Include the data-preparation log and a paragraph explaining which causal or population claims the design cannot support.

Suggested reading plan

Read Chapters 1–3 before practical analysis, then work through description and preparation in Chapters 4–6. Build the theory with Chapters 7–9 before the methods in Chapters 10–14. Finish with factor analysis and Bayesian reasoning, allowing separate sessions to compare their assumptions with earlier tools.

Questions for reflection

• Which limitations arise from the design rather than from the statistical test?
• What is the difference between statistical significance, effect size and uncertainty?
• How could cleaning, filtering or a modelling assumption change the conclusion?

Edition and source notes

Based on the 2025 Open Book Publishers edition by Navarro and Foxcroft, covering all sixteen numbered chapters. Software interfaces and optional modules can differ by installed version; use the source’s screenshots as guidance and verify the analysis settings in your own session. These original study tasks are not a reproduced answer key.

Consult the official source and full text.