How to Write Up a Moderation Analysis Using Hayes PROCESS Macro Model 1 in SPSS
A section-by-section guide to the regression output and write-up assignment — what goes in each part, how to report the statistics in APA format, and where most students lose marks before they get to the hypothesis test.
You have run the PROCESS macro in SPSS. You have your output tables. Now you need to write it up — and the assignment requires you to report the conduct of the analysis, present the statistical outcome, and describe the hypothesis test. Each of these is a distinct task with a distinct structure, and collapsing them into one undifferentiated paragraph is one of the most common ways students lose marks on this assignment. This guide walks through each section of a moderation analysis write-up based on Hayes (2018) Model 1, covering what belongs in each part, how to report the numbers in APA format, and what the rubric criteria are actually testing.
This guide does not complete the assignment for you. It explains the structure, the reporting conventions, and the reasoning behind each section so you can apply those to your own SPSS output and your own variables. The examples use the Stress–Gender–Depression dataset referenced in the assignment instructions, but the framework applies to any Model 1 simple moderator analysis.
What This Guide Covers
What Hayes Model 1 Is and What It Tests
Hayes (2018) Model 1 is a simple moderation model. It tests whether the relationship between a predictor variable (X) and an outcome variable (Y) differs depending on the level of a third variable — the moderator (W). In regression terms, moderation is the presence of a statistically significant interaction between X and W in predicting Y.
The model includes three predictors in the regression equation: the predictor X, the moderator W, and the product of X and W (the interaction term, often labelled Int_1 in PROCESS output). If the interaction term is statistically significant, moderation is present — the effect of X on Y is different at different levels of W. If it is not significant, you do not have evidence of moderation, though the main effect of X on Y may still be meaningful.
Predictor (X)
The independent variable whose effect on Y you are primarily interested in. Must be continuous (interval or ratio level) for a standard Model 1 run. In the course dataset example: Stress (DASSSTRS).
Moderator (W)
The variable that may change the strength or direction of the X→Y relationship. Can be categorical (e.g., dichotomous gender) or continuous. In the course dataset example: Gender (dichotomous).
Outcome (Y)
The dependent variable being predicted. Must be continuous (interval or ratio level). In the course dataset example: Depression (DASSDEP).
Moderation and mediation are completely different hypotheses. Moderation asks whether W changes the relationship between X and Y — they act simultaneously. Mediation (Hayes Models 4, 6, and others) asks whether X first causes M, and M then causes Y — a sequential causal chain. Warner (2021, Chapter 7) is explicit on this distinction, and confusing the two in a write-up signals a fundamental misunderstanding of the analysis. Model 1 is always moderation, never mediation.
Before You Write: What to Pull From the PROCESS Output
Before writing any section, locate and record the following values from your SPSS PROCESS output. Every number you need for the write-up comes from these tables. Write them down or highlight them in the output before opening a Word document.
Section 1: Reporting the Conduct of the Analysis
The first paragraph of any results write-up names the analysis that was conducted, identifies all variables and their roles, and specifies their measurement level. This paragraph does not include any numbers from the output. Its purpose is to orient the reader to exactly what was done before they encounter the statistical results.
For a Model 1 moderation analysis, this paragraph must state: (1) that a simple moderator analysis using Hayes (2018) Model 1 was completed, (2) the name and role of each variable (X, W, Y), (3) the measurement level of each variable (dichotomous/categorical, continuous-interval), and (4) the dataset the analysis was drawn from. A conceptual diagram (Figure 1 in APA format) showing X → Y with W pointing to the X→Y path is also typically included and referenced in this paragraph.
A simple moderator analysis using Hayes (2018) Model 1 was completed using the variables [Moderator name], a [dichotomous/continuous] variable [with two groups of X and Y / measured on a continuous scale], as the moderator (W), [Predictor name], as measured by [instrument], a continuous-interval level of measurement variable, as the predictor (X), and [Outcome name], as measured by [instrument], a continuous-interval level of measurement variable, as the criterion (Y) from the [dataset name]. The relationships examined are presented in Figure 1.
Every variable needs its name, its role (X, W, or Y), and its measurement level. Do not say “depression was the dependent variable” — say “Depression, as measured by DASS-Depression, a continuous-interval level of measurement variable, as the criterion (Y).” The precision of language here is part of what is being evaluated.
A conceptual model diagram (Figure 1) showing the moderator (W) pointing to the X→Y path, with X pointing to Y, is standard for moderation write-ups. In APA 7th edition, figure titles are formatted as: Figure 1 on one line (bold), then Conceptual Model of Simple Moderator Analysis on the next line (italicised). The figure itself is typically a simple box-and-arrow diagram. Do not place the figure caption inside the figure — it goes below it, outside the diagram boundaries. Reference the figure in the text (“The relationships examined are presented in Figure 1”) before the figure appears.
Section 2: Describing Data Screening and Assumptions
After naming the analysis, the write-up transitions to a brief description of data preparation and assumption testing. This section confirms that the data are suitable for the analysis and reports any violations that were found, along with how they were handled. It does not need to be long — typically one to three sentences — but it must be present.
For a multiple regression-based moderation analysis, the core assumptions to address are: normality of residuals, linearity of relationships, absence of extreme outliers, and homogeneity of variance. If the assignment included a separate data screening assignment (as in the course structure referenced in these materials), this section can reference that prior work: “The data were screened and tests of assumptions for multiple regression were completed, as demonstrated in the [prior assignment name].”
What to Report If Assumptions Are Met
A brief statement that the tests of assumptions were completed and the data are suitable for the analysis. Reference the prior screening assignment if applicable. One to two sentences is sufficient when no violations were found.
What to Report If There Was a Violation
Name the violated assumption, describe how it was identified, explain whether data transformation was applied and why (or why not), and state what the violation may or may not affect. Warner (2021) is a citable source for the consequences of common violations such as minor non-normality of residuals.
If your residuals show minor non-normality, this is typically not grounds for data transformation in a large sample. Warner (2021) and Cohen et al. (2003) both note that violation of normality of residuals does not cause problems with significance testing in larger samples, does not suggest coefficient discrepancies should be anticipated, and may simply indicate that a better-fitting model is needed — not that the data should be transformed. A brief statement acknowledging the deviation and citing the rationale for not transforming is appropriate and defensible.
Section 3: Presenting the Statistical Outcome
This is the section that reports the actual numbers from the PROCESS output. It has two components: a narrative paragraph that states the key statistics in APA format, and the supporting SPSS output tables (Table 1 for the model summary, Table 2 for the model coefficients). Both components are required — the narrative and the tables together, not one instead of the other.
What the Narrative Paragraph Must Include
The narrative paragraph for the statistical outcome of a Model 1 moderation analysis needs to state: that the PROCESS macro was used and which version, the null hypothesis being tested, the result for the main predictive effect of X on Y (t statistic, p value, and b coefficient), and the result for the interaction term that tests moderation (t statistic, p value, and b coefficient). Then direct the reader to the tables: “See Tables 1 and 2.”
A moderator analysis using PROCESS (v.[version]; Hayes, [year]) was conducted to test the null hypothesis. There was a [significant / non-significant] predictive effect between [X variable] and [Y variable], t = [value], p [< or =] [value], b = [value]. The result of the interaction effect between [W variable] and [X variable] upon [Y variable] was [statistically significant / not statistically significant], t = [value], p = [value], b = [value]. See Tables 1 and 2.
Note: report the PROCESS version from your output header. Report b (unstandardised coefficient) not beta (standardised) unless instructed otherwise. The interaction term result is reported second, after the main effect, regardless of which is significant.
Reading the PROCESS Output Tables: What Each Number Means
Most errors in the statistical outcome section stem from misreading the PROCESS output — pulling the wrong number for the wrong statistic, or confusing the row for the interaction term (Int_1) with the rows for the main predictors. This section maps the output to what you need to report.
| PROCESS Output Element | What It Tells You | Report As |
|---|---|---|
| R (Model Summary) | Multiple correlation — overall strength of association between the set of predictors and Y | R = .XX in narrative or Table 1 |
| R-squared (Model Summary) | Proportion of variance in Y accounted for by the full model — the effect size for the overall model | R² = .XX in narrative or Table 1 |
| F, df1, df2, p (Model Summary) | Significance test for the overall model — whether the combined predictors account for significant variance in Y | F(df1, df2) = XX.XX, p [< .001 / = .XX] |
| Coeff (b) — Int_1 row | The unstandardised regression coefficient for the X×W interaction — quantifies the moderation effect | b = ±X.XXXX in narrative |
| t — Int_1 row | Test statistic for the interaction coefficient — used to determine significance of the moderation effect | t = ±X.XXXX in narrative |
| p — Int_1 row | Probability value for the interaction — compare to α = .05 to determine whether moderation is supported | p = .XXXX or p < .001 in narrative |
| LLCI / ULCI — Int_1 row | Lower and upper limits of the 95% confidence interval for the interaction coefficient | Report in Table 2; can note if CI excludes zero (if significant) |
| Coeff / t / p — X row (DASSSTRS) | Main effect of the predictor X on Y, controlling for W and the interaction | t = XX.XXXX, p [< .001], b = X.XXXX in narrative |
APA Table Format for PROCESS Output
The assignment requires presenting tabular output. In APA 7th edition, statistical tables have a specific format that differs from the raw SPSS/PROCESS output appearance. You cannot paste the PROCESS output directly — you reproduce the key information in an APA-formatted table. The two tables typically needed for a Model 1 write-up are a Model Summary table (R, R², MSE, F, df1, df2, p) and a Model Coefficients table (Constant, X variable, W variable, Int_1 — each with Coeff, SE, t, p, LLCI, ULCI).
APA Table Formatting Requirements for This Assignment
- Table number and title: “Table 1” (bold, above the table), then on the next line the title in italics — e.g., Model Summary for Moderator Analysis. No period after the title.
- Horizontal lines only: APA tables use horizontal lines at the top, below the column headers, and at the bottom. No vertical lines, no internal horizontal lines between data rows.
- Column headers: Use standard statistical notation: R, R-squared, MSE, F, df1, df2, p for Table 1; Coeff, SE, t, p, LLCI, ULCI for Table 2.
- Note below table: A Note. line (italicised “Note” followed by a period) below the table identifies what Y variable is (e.g., Note. Y variable is DASSDEP).
- Decimal places: Report to 4 decimal places for most statistics in PROCESS output (matching what PROCESS outputs); p values to 4 decimal places, or use p < .001 for very small values.
- Row labels: In Table 2, row labels identify each predictor: Constant, your X variable name (e.g., DASSSTRS), your W variable name (e.g., Gender), and Int 1 (the interaction term).
Section 4: Describing the Hypothesis Test
The final required section of the write-up describes the outcome of the hypothesis test. This is not interpretation in the context of prior literature or theoretical frameworks — that belongs in a discussion section, not a results section. The hypothesis test section explains, in plain language, what the statistical results mean for the specific research question being tested.
There are only two possible outcomes for any hypothesis test. Either the results fail to reject the null hypothesis (p ≥ α, where α = .05), meaning there was not enough evidence in the data to demonstrate the null statement is incorrect — or the results reject the null hypothesis (p < α), meaning there is sufficient evidence that the null is incorrect and the alternative hypothesis is supported.
If the Interaction Is NOT Significant (p ≥ .05)
The results failed to reject the null hypothesis, which stated that [W variable] does not significantly moderate the predictive effect between [X variable] and [Y variable]. The results suggest that although there may be a significant predictive effect between X and Y, the extent of that effect is not dependent upon [W variable].
If the Interaction IS Significant (p < .05)
The results rejected the null hypothesis. [W variable] significantly moderated the predictive effect between [X variable] and [Y variable]. The relationship between X and Y differs significantly depending on the level of [W variable]. Simple slopes analysis or a plot of the interaction should be described to explain the nature of the moderation.
The results failed to reject the null hypotheses, which stated that gender does not significantly moderate the predictive effect between Stress (X) and Depression (Y). The results suggest that although there is a significant predictive effect between Stress and Depression, the extent of the effect between Stress and Depression is not dependent upon one’s gender.
This is the model provided in the assignment guidance — note how the hypothesis test section stays within the results and does not begin interpreting the result in relation to theory or prior literature. That belongs in the Discussion. The hypothesis test section answers only: was the null rejected or not, and what does that mean for the research question?
Write “the results failed to reject the null hypothesis” — not “the null hypothesis was accepted” or “there is no moderation effect.” Failing to reject the null means the evidence was insufficient to conclude moderation is present; it does not prove moderation is absent. This is a fundamental statistical concept and using the wrong language signals a misunderstanding that costs marks. The same logic applies to significant results: “the results rejected the null hypothesis” — not “the hypothesis was proven” or “the analysis confirmed that.”
APA Format for Reporting Regression Statistics
APA 7th edition has specific conventions for reporting statistical values. These are not optional stylistic choices — they are required format, and the rubric criterion on grammar, mechanics, and formatting evaluates APA compliance directly. The following table covers the most common formatting issues in moderation write-ups.
| Statistic | Correct APA Format | Common Error |
|---|---|---|
| t statistic | t = 19.9224 (italicise the t; use 4 decimal places; include sign if negative) | t = 19.9224 (not italicised); t(1232) = 19.92 (degrees of freedom belong in parentheses only when you report them) |
| p value | p = .4155 or p < .001 (italicise; no leading zero before decimal; use 4 decimal places) | p = 0.4155 (leading zero); p<.05 (no space); P = .42 (capitalised; 2 decimal places) |
| b coefficient | b = 0.9353 (no italics for b; include sign; use 4 decimal places) | β = .94 (beta is the standardised coefficient — different thing; no leading zero lost) |
| R and R-squared | R = .8571, R² = .7345 (italicise; no leading zero; superscript for squared) | R=.86 (no space around equals); R2 = .73 (not superscripted) |
| F statistic | F(3, 1232) = 1136.3851, p < .001 (italicise F; df in parentheses; 4 decimal places) | F = 1136.39 (missing df); F(3,1232) (no space after comma in df) |
| Confidence intervals | 95% CI [0.8432, 1.0274] (square brackets; 4 decimal places; LLCI first) | CI (0.84, 1.03) (parentheses instead of square brackets; 2 decimal places) |
| Italics in text | Statistical symbols in running text are italicised: t, p, F, R, N, n, M, SD | Not italicising statistical symbols is one of the most frequent APA formatting errors in student write-ups |
Annotated Sample Write-Up: What Each Paragraph Is Doing
The following is a model write-up structure for a Model 1 moderation analysis, annotated to show what each paragraph accomplishes. This uses the course dataset variables as a demonstration — your write-up uses your own SPSS output values and your own variable names.
Paragraph 1: Conduct of Analysis
Names the analysis (Hayes Model 1 simple moderator analysis), identifies all three variables with their roles (X, W, Y), states the measurement level of each, names the dataset, and references Figure 1. Contains no statistics from the output.
Paragraph 2: Data Screening
States that data were screened and assumption tests completed. References the prior screening assignment if applicable. Notes any violations found, explains how they were handled, and cites the methodological rationale. States the data are suitable for the analysis (or explains any caveat if they are not).
Paragraph 3: Statistical Outcome — Narrative
States that PROCESS (version; Hayes, year) was used to test the null hypothesis. Reports the main effect of X on Y (t, p, b). Reports the interaction term result (t, p, b). Directs reader to Tables 1 and 2.
Table 1: Model Summary
APA-formatted table presenting R, R-squared, MSE, F, df1, df2, p for the overall model. Table number and italicised title above; Note below identifying the Y variable.
Table 2: Model Coefficients
APA-formatted table presenting Constant, X predictor, W moderator, and Int_1 rows — each with Coeff, SE, t, p, LLCI, ULCI columns. Same format conventions as Table 1.
Paragraph 4: Hypothesis Test
States the outcome of the hypothesis test — rejected or failed to reject — with reference to the specific null hypothesis (naming the variables). Provides a one-to-two sentence summary of what this means for the research question in terms the reader can understand without specialist statistics knowledge. Does not interpret in relation to theory or literature.
Where Most Write-Ups Lose Marks
Reporting the Wrong Row for the Interaction
Reporting the t and p values from the X row (main effect of the predictor) instead of the Int_1 row (the interaction term) when describing the moderation test. The interaction test is always in the Int_1 row — the main effect row tests something different.
Instead
Before writing, label your output: circle the Int_1 row and write “interaction / moderation test” next to it. Circle the X variable row and write “main effect of X.” Keep them clearly distinguished. Report from each row for the correct purpose.
Skipping the Conduct of Analysis Section
Starting the write-up directly with the numbers — “The results showed t = 19.92, p < .001” — without first naming the analysis, identifying variables, and referencing the conceptual diagram. The conduct of analysis paragraph is a required structural element, not an optional preamble.
Instead
Always open with the conduct paragraph: name the model (Hayes Model 1), name each variable with its role and measurement level, name the dataset, and reference Figure 1. Only then move to data screening and the statistical results.
Interpreting the Results in the Results Section
Connecting the statistical outcome to prior literature, offering theoretical explanations, or suggesting practical implications within the results section. “The significant effect of Stress on Depression is consistent with cognitive models of depression (Beck, 1979)…” belongs in Discussion, not Results.
Instead
The results section answers the research question factually. The hypothesis test paragraph summarises what the statistical results mean for the specific null hypothesis. Interpretation, explanation, and connection to theory are Discussion content. Keep them separate.
Pasting Raw PROCESS Output as Tables
Copying the PROCESS console output directly into the Word document and presenting it as “Table 1.” PROCESS output does not conform to APA table formatting conventions — it has different column labels, includes elements not required for the write-up, and lacks the APA table title structure.
Instead
Reproduce the relevant values in a properly formatted APA table with a bold table number, italicised title, horizontal-lines-only formatting, and a Note below. Include only the columns required: R, R-squared, MSE, F, df1, df2, p in Table 1; Coeff, SE, t, p, LLCI, ULCI in Table 2.
Using “Accept the Null” Language
Writing “the null hypothesis was accepted” or “there is no relationship between the variables” when the interaction is not significant. This misrepresents how hypothesis testing works and signals a conceptual error that the rubric’s content knowledge criterion will flag.
Instead
Write “the results failed to reject the null hypothesis.” Failing to reject the null means the evidence was insufficient — not that the null is true. This distinction is not pedantic; it is conceptually important and is part of what the assignment rubric evaluates under content knowledge.
Missing APA Formatting for Statistical Symbols
Writing t, p, F, R without italics throughout the narrative. Not italicising statistical symbols is specifically listed in the APA 7th edition manual and is evaluated in the rubric’s grammar, mechanics, and formatting criterion.
Instead
Use your word processor’s italic formatting for all statistical symbols in running text: t, p, F, R, N, M, SD. Statistical symbols in table column headers are also italicised. This is a consistent, universal APA rule — apply it throughout the entire document.
Frequently Asked Questions
Getting the Write-Up Right: What the Rubric Is Measuring
The assignment rubric has five criteria. The highest-weighted criterion specifically evaluates whether the justification of the statistical analysis, its benefits, and its conduct are “supported by industry frameworks and theoretical constructs and peer-reviewed research.” In the context of a results write-up, this means: citing Hayes (2018) for the PROCESS method, citing Warner (2021) or Cohen et al. (2003) for methodological decisions around data screening and assumption handling, and following the reporting conventions established in the APA Publication Manual. The statistical values themselves are not a substitute for methodological grounding — you need both the numbers and the citable rationale for the analytical decisions you made.
The second criterion — cohesion and organisation — evaluates whether the write-up has a clear structure that a reader can follow from the conduct of analysis through to the hypothesis test conclusion. The four-section structure described in this guide (conduct, screening, statistical outcome, hypothesis test) is that structure. Deviation from it — for example, folding the hypothesis test into the statistical outcome paragraph, or omitting the data screening section — produces a write-up that is harder to follow and easier to mark down for lack of organisation.
For direct support with this assignment — whether you need a model write-up reviewed, help interpreting your PROCESS output, or assistance formatting the APA tables correctly — our statistics assignment writing team works specifically with SPSS-based regression analyses and APA results section writing. We cover the moderation framework, the PROCESS macro output, and the write-up structure as an integrated service grounded in what the rubric requires.
Continue with: biostatistics assignment help · psychology writing services · dissertation results chapter writing · critical analysis paper writing · proofread my research paper.