Every module we teach, and exactly what is in it
Research statistics taught the way they are actually used — the assumption that chooses the test, the SPSS path in full, and the sentence that reports it.
44
modules planned
25
available now
23h 20m
of learning time
A · Research foundations
What research actually asks of you
Two studies can run the identical test on the identical data and only one of them survives review. The difference is never the arithmetic.
From research question to statistical analysis
Two checks. The first asks whether the question is worth doing; the second asks whether it is answerable at all.
Research designs and what each one can prove
That is the entire distinction, and it is the only thing that changes what your study can prove.
Variables and levels of measurement
Fifteen minutes, and it is the highest-return quarter of an hour in this course.
Hypotheses, p-values, errors and power
This is the part that feels backwards, and it stays feeling backwards until you notice what it buys you.
Sampling and sample size
That is the whole distinction, and everything you are allowed to say about a population follows from which side of it you are on.
Validity, reliability and the instrument
Twenty minutes, and it will tell you whether you owe anybody anything.
B · Data preparation in SPSS
The SPSS workspace
SPSS looks like one program. It is really four windows that pass work to each other, and each one saves a different kind of file.
Codebook and data entry
The order is the content of this module. Every step is cheap while the ones after it have not happened yet, and expensive once they have.
Getting data in: Excel, Forms, Kobo, REDCap
Every import produces a working file, not a finished one. Keeping them separate is what lets you answer “where did this number come from?” a year later.
Cleaning data you did not collect
Duplicates come first because a duplicated case multiplies every other problem it carries. Outliers come last because you cannot judge an extreme value until the impossible ones are gone.
Missing data
You cannot choose a method before you have a view about the mechanism. Almost every bad decision about missing data is question 3 answered without question 2.
Transform: recode, compute, score
Every transformation in SPSS is one of these four, and every one of them writes to a new variable and leaves the original alone.
Restructure: select, split, merge, long/wide
Every operation in this module answers one of those two questions. The first three are reversible and persistent; the last three change the file itself.
Syntax, and why you should never click twice
This is the whole of Track B, in the order the modules taught it, as five sections of one text file you can re-run from nothing.
C · Describing data
Descriptive statistics and Table 1
Choosing the right summary for each measurement level.
Charts that do not look like SPSS charts
Choosing the chart the data justifies.
Normality and distribution checks
What "assumes normality" actually refers to, and the fact that it is usually the residuals and not the raw variable.
D · Comparing groups
Chi-square tests
Chi-square goodness of fit and chi-square test of independence, and which question each answers.
One-sample and independent-samples t-test
What question each t-test answers.
Paired-samples t-test
What makes data paired - before/after, matched pairs, two measures on the same person.
One-way ANOVA and post hoc
Why three t-tests is the wrong answer, with the inflated-error arithmetic on the slide.
Two-way and factorial ANOVA
Main effects and interaction, explained before any output appears.
Repeated measures ANOVA
When measurements are repeated on the same participants and why that demands a different test.
Non-parametric alternatives
When a non-parametric test is the right answer and when it is an excuse.
ANCOVA
What a covariate does and what "adjusted means" means.
E · Relationships and prediction
Correlation
What r measures and what it does not.
Simple linear regression
What regression adds that correlation does not - a prediction and a unit.
Multiple linear regression
What "holding the others constant" means and why it is the whole point.
Binary logistic regression
Why a yes/no outcome breaks linear regression.
Ordinal and multinomial logistic regression
When the outcome has three or more categories, and whether their order matters.
Moderation and mediation with PROCESS
The difference between a moderator and a mediator, with a diagram for each and a study where each is the right question.
F · Scales and multivariate
Reliability analysis
Internal consistency and what Cronbach's alpha actually measures.
Exploratory factor analysis and PCA
What factor analysis is for and how PCA differs from common factor analysis.
MANOVA
When several dependent variables belong in one analysis, and when they do not.
Cluster analysis
Segmentation as a research question.
G · Specialist methods
Survival analysis
Time-to-event data and why it cannot be analysed as a simple outcome.
Agreement and reliability between raters
Why percentage agreement is not enough.
Diagnostic accuracy and ROC curves
The 2x2 diagnostic table.
Complex samples and survey weights
Why multistage survey data cannot be analysed as a simple random sample.
H · Reporting and defence
Reporting statistics in APA style
The anatomy of a reported result - statistic, degrees of freedom, value, p, effect size, confidence interval.
Tables and figures for journals
What belongs in a table and what belongs in the text.
Writing the results and discussion
The order results should be presented in, and why it mirrors the objectives.
Defending your analysis
## Suggested bundles
How it is sold
Take one module, a track, or the whole thing
Every module can be bought on its own if you only need one test. Most people take a programme — a named bundle that covers a whole job of work — and the modules you have already finished carry across if you upgrade later.
See the programmes