Read the tables in the order they are used
Not top to bottom as printed. Statistical software emits tables in the order it computed them, which is not the order in which they answer your question, and that mismatch is most of why output feels impenetrable.
For a comparison of three or more groups the useful order is: how many people were in each group and what their averages were, then whether the test's preconditions were satisfied, then the comparison itself, then which specific pairs differ. Anything else printed is diagnostic material for the analyst. Once you know that sequence you can ignore roughly two thirds of what appeared on screen.
- Descriptives: group sizes and averages, so the reader knows what is being compared
- The preconditions check, reported whether it passed or failed
- The main comparison, carrying the statistic, its degrees of freedom and a probability
- A measure of how large the difference is, which the comparison itself does not tell you
- Pairwise comparisons, but only where the main test justified looking
- Everything else, which belongs to the analyst rather than the paper
The preconditions row is not optional
Students skip it because it looks like housekeeping and because a failure feels like bad news. It is neither. The check tells you whether the comparison underneath it can be trusted, so reporting it is what makes everything after it meaningful.
When it passes, one sentence covers it and you move on. When it fails, you have three respectable options and a marker who is watching for which you take: switch to a test that does not need the assumption, adjust the data and declare it, or continue while naming the violation as a limitation. Graduate criteria frequently award marks specifically for handling this in the open, which makes a failed check an opportunity rather than a setback.
Rebuild the table, do not screenshot it
Software tables carry borders, footnotes and column headings that belong to the program rather than to any style manual, and pasting one in as an image guarantees it matches nothing your course requires.
So the values get typed into a table you build yourself, following whichever manual your program names, with a caption above it and the abbreviations explained beneath. It takes ten minutes and it is visibly different from what most of the class submits. It also forces you to decide which numbers matter, which is the part of the exercise being assessed and the part a screenshot lets you avoid entirely.
Three numbers reach your paper
From the main comparison you need the test statistic, its degrees of freedom, and the probability. From a separate calculation you need something describing how big the difference actually is. That is four figures, formatted to whichever manual your course uses, and everything else stays out.
The reason the fourth matters is that the first three cannot tell you it. A probability answers whether a difference this large would be surprising if there were genuinely nothing going on. It says nothing about whether the difference is big enough for anyone to act on, and with a large sample a trivial gap will clear the threshold routinely. Reporting both, then saying what the combination means, is what separates a results section from a screenshot.
Turn it into the sentence
The whole exercise ends in one construction, and it is worth memorising because almost every comparison you write for the rest of your degree fits it. Name the test and what it compared. Give the figures. State the direction of the difference in ordinary words. Say what it means for the question you asked.
Four clauses, no jargon beyond the unavoidable, and no hedging. Students lose marks by stopping after the figures, which leaves a marker holding numbers and no interpretation. The interpretation is the assessed part; the numbers are merely the evidence for it. Where the analysis belongs to a dissertation rather than a course, the plan should exist before you collect anything.
Questions people actually ask.
Should I paste the output straight in?
Avoid it, because marks go missing almost every time. What the program prints obeys no style guide and carries far more than any write-up wants. Retype only the handful of figures your analysis genuinely needs into a table built to your course's manual, and drop the diagnostic material. Choosing well is part of the assessment.
Which numbers actually go in the paper?
The test statistic, its degrees of freedom, the probability, and separately a measure of how large the effect is. Four figures. The first three cannot tell you whether a difference matters, only whether it would be surprising, which is why the fourth is increasingly required rather than optional at graduate level.
What do I do when the preconditions check fails?
Report it and pick one of three routes: a test that does not require the assumption, a declared transformation of the data, or continuing while naming the violation as a limitation. All three are respectable and graduate criteria often reward the candour. Only proceeding silently as though it passed damages you.
How do I write the result as a sentence?
Four clauses. Name the test and what it compared, give the figures, state the direction of the difference in ordinary words, then say what that means for your research question. Stopping after the figures is the common error and it leaves your marker holding evidence with no interpretation attached.
Can I screenshot the output table?
Better not to. Software tables carry formatting that belongs to the program rather than to any style manual, so an image matches nothing your course asked for. Retyping the few values you need into a properly formatted table takes ten minutes and forces the selection decision that is actually being assessed.
