Online
Where the marks go depends on the department. An economics or business course tends to reward the modelling decision and the diagnostics behind it. A psychology course wants the assumption checks and the reporting conventions followed exactly. A nursing or public health course wants the number converted into something a ward could act on next quarter, with the uncertainty stated rather than buried. The same correct analysis can sit in three different bands depending on which of those the rubric is asking for, and the rubric will tell you if you read it before starting.
So the work is mostly the last step: taking output and stating it in the units of practice, with appropriate caution about what the design supports. Confidence intervals matter more here than in most versions of the course, because clinical decisions are made about ranges rather than point estimates.
The recurring failure in psychology writing is a conclusion the design cannot support: correlational work described causally, convenience samples generalised to populations, statistical significance reported as though it were importance.
Markers here are trained to catch precisely that, so the discipline of staying inside what your design permits is worth more than any amount of computational fluency. Effect size and interval reporting are usually explicit rubric requirements rather than good practice, and omitting them costs marks directly.
Here the statistics are instrumental. Nobody is grading your regression for its own sake; they are grading whether it supports a recommendation somebody could act on, what it would cost to act, and what you would monitor afterwards.
Which means an analytically perfect answer that stops at the output scores in the middle. The step everyone skips is the one from finding to decision, and it is the step the entire course exists to teach.
Choosing a test is a decision tree rather than a memory exercise. What kind of outcome, how many groups and are they independent, and are you comparing groups or examining a relationship. Written on one page that tree answers most of what an exam will ask.
There is also a practical test worth applying before anything is submitted: cover the output and describe what it shows. If that comes out fluently, the analysis is yours. If it does not, you are holding a result rather than a finding, and this is a subject where the next assignment builds directly on the one before it.
Biostatistics, psychology methods, business analytics or a general course. They are marked on different things.
What your data and design permit you to claim, established before any test is chosen.
Into the units your discipline uses, in language you could defend without notes.
No, and the link is weaker than you think. Graduate statistics is largely reasoning about evidence rather than calculation, which software does. Professionals who make careful decisions from imperfect information already have the underlying skill and are missing the vocabulary, which is quick to acquire.
Use the one your course assumes, since the assignments and the exam are written around its menus and its output. Beyond that the choice matters far less than it appears to: what a test does and when it is the wrong test are properties of the statistics, not of the software, and those carry across intact. Picking up a second package once you hold the first is a weekend, not a term.
State it, say what you did, and justify the choice — that is a stronger answer than pretending it held, and it is explicitly rewarded. Depending on the violation you might transform the data, use a non-parametric alternative, or proceed with a clearly stated limitation.
Because a figure you cannot narrate is a liability in a course that ends with a closed-book exam and, later, in a room where somebody asks why you chose that test. Everything comes back with the reasoning attached and the assumption checks shown, and the bar is that you can talk through the output before it goes anywhere.
The same ideas at higher stakes, applied to your own data rather than a supplied set. Students who learn the coursework properly find the dissertation analysis far less frightening, which is the strongest argument for doing it now rather than getting through and meeting it again in two years.
Tell us the program, the term ahead and where the grade is leaking. A specialist in your field replies with a plan and a quote within two hours. The reading costs nothing, and sometimes the honest answer is that we are the wrong fit.
Online