What follows is a complete Results section, first paragraph to last, written as it would appear in a real manuscript. Below it, the same text is annotated sentence by sentence: why each element sits where it does, what decision lies behind it, and what happens if you drop it.
If what you want is the general notation guide — italics, decimals, how each statistic is written — that is in how to report results in APA 7. This is the other thing: the whole template and the running order.
The example study
A two-arm trial: 64 university students with anxiety symptoms randomised to an eight-session emotion regulation intervention or to a waiting list. Primary measure: anxiety (STAI-T). Measured at pretest, posttest and three-month follow-up.
The section, in full
Preliminary analyses
"Of the 64 participants randomised, 58 completed the posttest assessment (90.6%) and 54 the follow-up (84.4%). Attrition did not differ between conditions, χ²(1, N = 64) = 0.42, p = .517, nor in baseline anxiety, t(62) = 0.88, p = .383. An intention-to-treat approach was followed and missing data were handled with full information maximum likelihood.
The STAI-T distribution was inspected by group and time point. Skewness ranged from −0.42 to 0.61 and kurtosis from −0.55 to 0.38, within the acceptable range for the linear model. Levene's test indicated no heteroscedasticity at pretest, F(1, 62) = 1.14, p = .290.
Descriptive statistics
Table 1 reports means and standard deviations of the STAI-T by condition and time point. At pretest the two groups started from equivalent scores (intervention: M = 24.10, SD = 5.02; waiting list: M = 23.40, SD = 5.31), t(62) = 0.54, p = .591, d = 0.14, 95% CI [−0.35, 0.63].
Main analysis
A 2 (condition) × 3 (time) mixed ANOVA was fitted on the STAI-T. Mauchly's test indicated that sphericity was violated, W = .84, χ²(2) = 9.12, p = .010, so the Greenhouse-Geisser correction was applied (ε = .86).
The condition × time interaction was significant, F(1.72, 89.44) = 14.27, p < .001, η²p = .22. The main effect of time was also significant, F(1.72, 89.44) = 21.03, p < .001, η²p = .29, and the main effect of condition was not, F(1, 52) = 2.91, p = .094, η²p = .05.
Pairwise comparisons with Bonferroni adjustment showed that the intervention group reduced anxiety between pretest and posttest (difference = 6.20 points, 95% CI [3.90, 8.50], p < .001, d = 1.18) and maintained the improvement at follow-up relative to pretest (difference = 5.80, 95% CI [3.30, 8.30], p < .001, d = 1.09). The waiting list group did not change on any contrast (all p > .38).
Secondary analyses
The between-group effect size at posttest was d = 0.92, 95% CI [0.38, 1.45]. In the intervention group 41.4% of participants (12 of 29) fell below the clinical cut-off, compared with 6.9% (2 of 29) in the waiting list group, χ²(1, N = 58) = 9.06, p = .003, OR = 9.53, 95% CI [1.92, 47.30]."
Annotated, sentence by sentence
1. Participant flow comes first, and it comes with numbers
The first paragraph says how many came in, how many left, and whether the leavers resembled the stayers. That contrast — attrition by condition and on the baseline variable — is not decoration: it is the evidence that dropout did not bias the result. If the group that was doing worst dropped out more, the effect you report afterwards may be an artefact.
And the missing-data handling goes here, by name. "Cases with missing data were removed" is a decision, and a poor one, but at least it is a decision on the record. What does not pass is not saying.
2. Assumptions: with numbers, not adjectives
Notice it does not say "the data were normal". It gives the skewness and kurtosis ranges. That is the difference between a checkable claim and a statement of intent.
Notice too what is not there: no Shapiro-Wilk test. With 29 per group that test adds nothing — it flags trivial departures or misses important ones depending on n — and what decides is inspecting the shape. The long argument is in the Shapiro-Wilk normality test, misunderstood. If your reviewer asks for it, report it; but do not make it the argument.
3. Descriptives go in a table, and the text keeps only what gets read
The table carries all six means; the text repeats only the two pretest ones, because those are what support baseline equivalence. Repeating in prose everything already in the table is the most common writing error in this section, and it buries what matters.
The pretest contrast carries its effect size with an interval even though it is not significant. A d = 0.14 with an interval of [−0.35, 0.63] says "there is no difference and we looked with enough precision to tell"; a bare p = .591 says half as much.
4. The main analysis is announced before it is given
"A 2 × 3 mixed ANOVA was fitted" comes before any number. The reader has to know what they are looking at before they look. And Mauchly's result comes before the F, because the degrees of freedom that follow depend on it. The full detail of that correction is in repeated measures ANOVA in APA 7.
5. The interaction first, main effects after
In a design with a significant interaction, the order is neither chronological nor the software's: it is by importance. The interaction is the study's hypothesis — that the intervention changes things over time — so it goes first. Main effects follow, in two lines.
And the non-significant ones are reported too, with their numbers. A condition effect at p = .094 left out is selective omission, and in a manuscript heading for a journal with an open data policy it shows immediately.
6. Differences in scale units, not only in d
"Difference = 6.20 points, 95% CI [3.90, 8.50]" is clinical information: anyone who knows the STAI knows what six points mean. The d = 1.18 follows, for comparison with other studies. Both together, in that order, is what makes the result usable.
7. Contrasts that came to nothing get grouped
"The waiting list group did not change on any contrast (all p > .38)." Three lines of non-results compressed into half a line. It is correct and it is readable; listing them one by one would have buried the result that matters.
8. Clinical significance, at the end
The last paragraph translates the statistical effect into people: how many stopped exceeding the clinical cut-off. It is optional in APA but it is what an applied-journal editor looks for, and often it is the only thing that survives into the reader's summary.
The order APA 7 expects
One: participant flow and missing data. Two: preliminary checks and assumptions. Three: descriptives. Four: the main hypothesis test. Five: secondary or exploratory analyses, labelled as such. Six: any sensitivity analysis.
That order is not a style quirk: it reproduces the order in which a sceptical reader needs things. First they want to know whether the sample holds up, then whether the model applied, then what was there, and only then what came out.
What does NOT go in Results
Interpretation. "This suggests emotion regulation protects against anxiety" is Discussion. Results carries what happened, not what it means.
Comparisons with the literature. "In line with Smith et al. (2020)" is Discussion.
Limitations. Discussion.
Procedural detail. How the scale was administered is Method. If you find yourself explaining in Results how you did something, it was missing from Method.
The hypotheses. They belong in the Introduction. Results can refer to them by number ("hypothesis 2 was not supported"), but does not restate them.
A quick check before submitting
Does every contrast you report carry its effect size? Does every effect size carry its confidence interval? Are the df there for every statistic? Do the numbers in the text match the tables, one by one? Have you said what was done with missing data? Are the non-significant contrasts you predicted reported? Is there an interpretive sentence that belongs in the Discussion?
That last question is the one most often answered yes. The exact notation for intervals, means and bootstrap is in how to report confidence intervals and bootstrap in APA 7, and if what you need is to turn a loose statistic into its sentence, the APA 7 results formatter does it.
And if what you want is the standalone template for each test to fill in, the APA 7 Results Kit gathers all ten model paragraphs with a copy button. It opens by leaving your email in the form just below this article.
Before you submit
Results is the section where a methodological reviewer decides whether the study was done properly, and they decide it in two readings. If you want the objections before they get there, upload the manuscript to the Q1 paper reviewer: it is free, needs no sign-up, and hands each critique back pinned to the sentence that triggers it.
References
American Psychological Association. (2020). Publication manual of the American Psychological Association (7th ed.).
Appelbaum, M., Cooper, H., Kline, R. B., Mayo-Wilson, E., Nezu, A. M., & Rao, S. M. (2018). Journal article reporting standards for quantitative research in psychology: The APA Publications and Communications Board task force report. American Psychologist, 73(1), 3-25.