Writing and reporting: reporting rules for every test

Guide

On this page

Every calculator on the site ends with an APA-style table, a report sentence in English and a Word file – and every calculator page has a "Reporting the results" section with the books' examples for that test. This page brings together the rules above the individual test: what the results section must contain according to APA's reporting standards (American Psychological Association [APA], 2020, Section 3.7), when to write a sentence and when a table (section 6.40), how numbers are written (section 6.36), what the structure of tables and figures is (sections 7.8-7.28), and one table summarizing, for each of the 17 calculators, what must be in the sentence. Field (2024, section 1.8.2) points out that the rules "are no substitute for common sense": their role is for everyone to be consistent, not to lay down a law for every situation – and that for a thesis or a paper, always also check the guidelines of the supervisor or the journal.

1. The results section according to APA's reporting standards

The reporting standards for quantitative research (JARS; APA, 2020, Section 3.7) open with one principle: "Report the data in sufficient detail to justify the conclusions… Mention all relevant results, regardless of whether hypotheses were supported, including results that run counter to expectation; include small effect sizes (or statistically nonsignificant findings) when theory predicts large ones. Do not hide uncomfortable results by omission". Hence what the results section contains, in this order:

Before the hypotheses – data diagnostics. The findings of the data check (outliers, assumptions, missing values) are reported in the method section, before the results related to the support of the hypotheses; and the decisions derived from them – exclusions, transformations, imputation – are stated. Missing values: "it is critical to report the frequency or percentages of missing data, along with any empirical evidence and/or theoretical arguments for the causes" – and the calculators print in every table the N analyzed out of the N loaded and the deletion rule (pairwise or listwise, as in SPSS). Data diagnostics are done on the outliers and assumptions, normality check and missing values pages.

The order of the hypotheses. Primary, secondary and exploratory hypotheses – in the same order in which they were presented in the method section (section 3.6).

What an inferential test must include. "For inferential tests (e.g., t, F, chi-square) – the value of the test statistic, the degrees of freedom, the probability of obtaining a value as extreme as the one obtained (exact p value), and the size and direction of the effect. When point estimates are provided (e.g., means, regression coefficients) – always include an associated measure of precision (e.g., standard error)" (section 3.7). The "sufficient set" of statistics includes at least the group sizes, the means (or frequencies) and the standard deviations of each cell – so that a reader can reproduce the analysis and include it in a meta-analysis.

Confidence intervals. "Usually the best reporting strategy", because they combine location and precision and significance can be derived from them; one confidence level (95% or 99%) for the whole work, set in advance – and "wherever possible, base discussion and interpretation of results on point and interval estimates" (section 3.7). APA's emphasis: the significance test "is only a starting point", and an effect size with a confidence interval is "the minimum expectation for all APA journals".

Software, settings and problems. For complex methods (SEM, factor analysis, hierarchical models) – the details of the model, the software and the package (for example "lavaan in R"; without a bibliographic reference for the software), estimation problems (nonconvergence), problems in diagnostics or assumptions "that may affect the validity of the findings" (section 3.7). For the site: "The computation was performed with calculator X of the Professor Stat website (version N)" – and for cross-checking, the SPSS syntax the calculator generates.

2. Sentence, table or figure

The APA rule (2020, section 6.40), which Field (2024, section 1.8.3) also cites: "three or fewer numbers – try a sentence; four to twenty – try a table; more than twenty – try a figure" – and above all, "choose the mode of presentation that optimizes the reader's understanding". Field adds that the first rule is the most important of all: there are countless situations in which a figure for 4-20 values will show the pattern better than a table, and figures of more than 20 numbers that are "a complete mess". Two accompanying rules: not the same numbers twice – "if you have presented descriptive statistics in a table or figure, do not repeat them in the text", but mention the table and highlight in the text the data that help interpretation (section 6.43; Field: "a waste of space"); and reference by number – "as shown in Table 1…", "Figure 2 shows…", never "the table above" or "the figure on page 32" (section 7.5), and alongside the reference – what to look for in the table.

In practice, in a typical thesis: the sample description – one table in the method section (frequencies page, Table 7.2 in the APA manual); a single correlation, one t-test or one chi-square – a sentence; a correlation matrix, multiple regression, a series of t-tests or an analysis of variance with follow-up tests – a table with text alongside that highlights the picture; a distribution, an interaction or a relationship between two quantitative variables – a figure.

3. Rules of form – how each number is written

Zero before the decimal point (APA, 2020, Section 6.36): write 0 before the decimal point when the statistic can exceed 1, and omit it when it cannot – "correlations, proportions, and levels of statistical significance". Table 1 sorts the symbols that appear in the site's output.

How many digits. "Round as much as possible while considering prospective use and statistical precision… it is often preferable to round to two decimal places or to rescale" (section 6.36). Correlations, proportions and t, F and chi-square – to two decimal places; means and standard deviations of whole-number scale data (questionnaire items) – to one decimal place, "because as group measures they are more stable than individual scores"; exact p to two or three decimal places, and below .001 – "p < .001". The tradition of p < .05, p < .01 "suited the time when only tables of critical values were available" – in the text write the exact p; "p <" is allowed in tables when exact p values would be cumbersome (section 7.14). Field (section 1.8.3): "not always two decimal places – as many as are needed in the given situation", and the example of 0.0425 meters versus 4.25 centimeters. The site's convention: the tables print two decimal places for measures and statistics, three for p and for effect-size measures that cannot exceed 1, like SPSS – the correlation matrix stays at three decimal places for cross-checking, and the "Copy table for submission" button rounds to two.

Confidence interval (section 6.43): in the format 95% CI [LL, UL]; in every report – the confidence level; when it recurs in a series at the same level – without repeating "95% CI"; after a point estimate – without repeating the units: "M = 30.5 cm, 99% CI [18.0, 43.0]". In tables (section 7.16) – in square brackets in the same cell, or in two columns LL / UL; Tables 7.16 and 7.17 in the manual are the two examples.

Symbols and abbreviations (section 6.44): statistical symbols – M, SD, F, t, df, p, N, n, OR, r – and Greek letters (α, β, χ²) are not defined; other abbreviations – CI, ANOVA, CFA, RMSEA, SEM, AIC, BIC – are defined at first use. N is the whole sample and n a subgroup (Field, 2024, Section 1.8.3). Statistical symbols in italics (section 6.22) – and in Hebrew, see section 7.

Spacing and the minus sign (section 6.45): "space mathematical copy as you would space words" – a + b = c, not a+b=c; a space on both sides of a subtraction minus, and none after the minus of a negative value (−8.25); and the minus sign is a different character from the hyphen – "longer and slightly higher". In the site's output negative values are printed with the ordinary keyboard hyphen-minus; anyone submitting to a journal replaces it with − (in Word: Insert › Symbol, "Minus Sign").

A mean without spread is half the information. Field (2024, section 1.8.3): when reporting a mean, also report a standard deviation – "M = 95, SD = 56.79", or the standard deviation in parentheses when saying it in the text; and when reporting a median – the range or IQR ("Mdn = 98, IQR = 63"). And consistency: whoever chose M uses it throughout the report.

4. Tables

A table "should be an integral part of the text but designed so that it can be understood in isolation" (APA, 2020, Section 7.8) – and so every table in the site's output carries a number, a title and a note. The components (section 7.9): a bold number above the table, in the order of mention in the text (section 7.10); a short, explanatory title in italics below it (section 7.11); headings for every column – including the left-hand stub column, which holds the variables or groups; subcategories indented and not in an additional column (section 7.12); the body; and notes below (section 7.14): a general note – "Note." followed by explanations of abbreviations and symbols; specific notes with superscript letters; and a probability note – what the asterisks mark ("*p < .05. **p < .01. ***p < .001."). Rules (section 7.17): a line above the table, below the headings and at its bottom only – "no vertical rules and no borders around every cell", and no shading; to emphasize a cell – a note, or bold explained in the general note. The site's output tables are built this way, and the "Copy table to Word" button transfers the table with this structure.

Sample tables in the manual (section 7.21), worth knowing as needed: Table 7.2 – sample characteristics (n and % for each category; in the % column without the percent sign – section 7.12); Table 7.10 – a correlation matrix (lower triangle, numbered variables, asterisks with a probability note); Tables 7.15-7.18 – regression (with or without confidence intervals, hierarchical); Tables 7.19 and 7.21 – model comparison and fit indices. The frequencies, correlations, regression and path analysis pages each show the corresponding table.

5. Figures

A chart is a "figure" (Figure), with the same principles: a bold number in the order of mention (section 7.24), a title in italics below the number (section 7.25), the names of the variables and the units on the axes, a legend when there are several series (section 7.27), and a figure note (section 7.28) that explains "units of measurement, symbols, and abbreviations not included in the legend… the use of shading, color, and any design element that carries meaning", and in a chart with error bars – what they represent (standard deviation, standard error, confidence limits) and "it is also helpful to include sample sizes". Every chart in the site's output carries a number, an italic title and such a figure note – the site's convention, set in the master document following section 7.28 – and the scatterplot page has a complete sample figure. A documented limitation: the charts (SVG) are not included in the Word file the calculator downloads – copy them as an image from the browser.

6. What to report for each test – Table 2

Table 2 summarizes, for each calculator on the site, what the report sentence must include according to the sources in the "Reporting the results" section of that page, the format of the statistics (in Latin letters, as it is also printed inside the report sentence), and the effect-size measure. The last column says what the output already gives – and that is what to copy from.

7. Hebrew, English and copying to Word

The Hebrew sentence. The words in Hebrew, the statistics in Latin letters – and symbols are not translated: M and p, never translated into other letters. The technical difficulty is directionality: a sequence like "t(21.54) = −1.71, p = .101" inside a Hebrew sentence falls apart if each character aligns according to its neighbors. The site's convention, according to which the sentences in the output are built: every statistical sequence sits in a single left-to-right segment (span.en), including its parentheses, and on mobile it breaks only at commas – and so copying to Word keeps the correct order. Anyone typing it themselves in Word: type the whole sequence in one go without switching language in the middle – or paste from the calculator.

Italics. APA (2020, section 6.22) requires italics for statistical symbols – M, SD, t, p – but not for Greek letters (α, β, χ²) and not for letters that are abbreviations (CI, RT). In the site's output and in the Word file the symbols are not italicized, as in SPSS output; before submission italicize them in Word (find and replace with formatting, or by hand). Hebrew has no true italic, and in any case the rule applies only to the Latin symbols.

The English sentence. Every calculator also prints a sentence in English – for an abstract, an article or a paper in English; the variable names appear in it exactly as they are in the file (so a file in English gives a clean English sentence). The wording follows the templates of Field and Pallant: a finding, then the statistics in parentheses or after a comma.

Copying. Three buttons in every output: "Copy" next to the report sentence, "Copy table to Word" under every table (the table goes in as a Word table with horizontal lines only), and "Download full output (Word)" – a docx file with all the tables and sentences and a source line. The order in the results chapter: the sentence, followed by the table when there is more than fits in the sentence – and the table number in your paper follows the order in which you mention it, not the table number in the output.

Punctuation. A site convention that also applies to what you copy: a short dash (–, en dash) with spaces for a clause, a regular hyphen for ranges (pp. 31-32) and for compound words (well-known), straight quotation marks; the long dash (em dash) is not used in Hebrew. In Word, which automatically replaces two hyphens with a long dash – check after pasting.

8. When the sources disagree – "both and"

The three books and the APA do not always say the same thing, and in such cases the site follows a "both and" approach: it prints both and says which is which – so the reader chooses, and reports explicitly what she chose. The main cases:

The effect size for nonparametric tests. Field (2024, sections 7.4.5, 7.5.6, 7.6.7) – r_rb, the rank-biserial correlation; Pallant (2020, chapter 16) – r = z/√N with Cohen's conventions. Two different yardsticks that cannot be converted into each other; for the Wilcoxon test Pallant defines N as the number of observations (twice the number of pairs). The three pages print both; the sentence states which one is used.

The effect size for t. Field – Cohen's d; Pallant – eta squared; for paired samples Field suggests d corrected to d_z. The pages print d (or d_z) in the sentence and the other measures in the table note.

N in the parentheses. Gravetter and Wallnau (2017) and Pallant write "χ²(1, N = 200)" and "H = 3.785 (2, N = 15)"; Field – "χ²(1) = 25.36" and "H(3) = 8.66". The pages print N.

Adjusted p or not. In the pairwise comparisons after the Kruskal-Wallis test, Field reads Adj. Sig. only, Pallant scans Sig.; the site convention – the adjusted one, and both in the table.

Exact or asymptotic p. Field (Box 7.2): in a small sample, "basically anything below 50", the exact one; the pages compute it up to 50 per group (a site convention) and print both.

Two digits or three. APA – two for correlations; SPSS and R – three. The matrix in three for cross-checking, the "table for submission" in two.

Common mistakes

"p = .000". It is not zero but an SPSS rounding – write p < .001 (APA, 2020, Section 6.36; Gravetter & Wallnau, 2017, p. 287).

"p < .05" when the exact p is known. The example of Gravetter and Wallnau: p = .017, not p < .05 (p. 287); "p <" – only in tables (APA, Section 7.14).

Significance without an effect size and without a confidence interval. "The minimum expectation" (APA, Section 3.7). Field's example (2024, section 10.10.1): "not significant, p = .101; however, it represented an effect of d = 0.70" – both facts in the same sentence.

Omitting what did not come out. "Do not hide uncomfortable results by omitting them" (section 3.7) – every hypothesis that was tested is reported.

Means for a test on ranks. For the Mann-Whitney, Wilcoxon and Kruskal-Wallis tests, report medians (Field, Sections 7.4.6, 7.5.7, 7.6.8; Pallant, Chapter 16).

Copying an SPSS table as is. A correlation matrix that shows every correlation twice, tables with borders around every cell, "Sig. (2-tailed)" – the table for submission is a lower triangle with asterisks and a significance note (the correlations page; APA, Table 7.10).

A zero before the decimal point in p or r, and no zero in t or d. Table 1.

"The table above". Refer to tables by number, and say what to look for in them (APA, Section 7.5).

The same numbers in the table and in the text. "A waste of space" (Field, Section 1.8.3) – in the text, highlight; do not repeat.

A mean without a standard deviation, a median without a range. "Half the information" (Field, Section 1.8.3).

Forgetting N. In the sample description table – in the title or in the note; in a test – in the parentheses or in the sentence; in the n of a subgroup (Field; Gravetter & Wallnau).

Defining M and SD, and not defining CI and RMSEA. Exactly the opposite (APA, Section 6.44).

A statistical string reversed in Hebrew. "p = .101" that appears as "101. = p" – the whole run in LTR (section 7).

Checklist – before submission

References

American Psychological Association. (2020). Publication manual of the American Psychological Association (7th ed.). https://doi.org/10.1037/0000165-000

Field, A. (2024). Discovering statistics using IBM SPSS statistics (6th ed.). SAGE.

Gravetter, F. J., & Wallnau, L. B. (2017). Statistics for the behavioral sciences (10th ed.). Cengage Learning.

Pallant, J. (2020). SPSS survival manual: A step by step guide to data analysis using IBM SPSS (7th ed.). Open University Press.

Wilkinson and the Task Force on Statistical Inference (1999) – as cited in APA (2020, section 3.7). Per the site rule, the reference list includes only sources that were read in the original: APA (2020) – sections 3.7, 6.22, 6.36, 6.40-6.45, 7.5, 7.8-7.17, 7.21-7.28; Field (2024) – sections 1.8.2-1.8.3 and the reporting examples in the test sections (10.10.1, 7.4.6, 7.5.7, 7.6.8 and others, as quoted on the calculator pages); Gravetter and Wallnau (2017) and Pallant (2020) – by page and chapter, as quoted in the "Reporting the results" section of each calculator page.