Research Methods
Four fictional research questions, read one at a time. What is each one actually asking for, and what would you have to know before naming an analysis?
Simulated — fictional studies, no real project or dataset
By the end you should be able to read a research question and say what kind of claim it makes, and which designs could support it.
About 20 minutes. Nothing you do here is saved or sent anywhere.
Several of these questions can be read in more than one defensible way, and one of them is deliberately ambiguous. The tool marks readings as well supported, defensible or hard to defend from this wording rather than right and wrong, and explains what each reading would commit you to.
Students are routinely taught a flowchart: look at the data, count the groups, pick the test. Before you use the mapper, commit to a view about where a method choice really comes from.
One question at a time. Your reading builds up in the pinned card as you set each control, so you can always see the question you are reading and the reading you are giving it.
Four decisions per question. None of them is "which test?" — that comes last, and only once the first four are settled.
A researcher has interview transcripts from 18 students about group work, and a spreadsheet of those students' attendance, prior marks and self-reported engagement. They want to publish one paper.
The design a question needs follows from what the question is asking for — a description, a difference, an association, a meaning — and not from whether the data will be numbers or words. Reading the aim first is what makes the rest of the chain fall out, and noticing that a question is underspecified is a finding rather than a failure.