Section 9 — Integrative exercise

What you are practising: combining everything you have learned to answer a single, fully-specified research question.

Goal

Pick one of the three outcomes. Build a kitchen-sink multilevel model and the matching wide-format SEM. Report a 4–6 sentence interpretation paragraph.

Tasks

Choose an outcome. Pick whichever of engagement, performance, creativity is most interesting to you.
Build a kitchen-sink multilevel model. Include:
- All three actor predictors
- All three partner predictors
- All three dyad-level moderators
- The focal crossover-by-moderator interaction for your chosen outcome
- A random intercept for dyad_id
Build the corresponding wide-format SEM. Include the manual product indicator for the focal interaction and the computed simple-slope parameters at three levels of the moderator (low, mean, high).
Report:
- The actor and partner effect estimates with standard errors
- The focal interaction coefficient from both approaches
- The three simple slopes (low, mean, high levels of the moderator)
- The SEM fit indices (chi-square, df, p-value, CFI, TLI, RMSEA, SRMR)
Write a 4–6 sentence interpretation paragraph covering:
- The substantive size and direction of the partner crossover effect
- Whether the moderator meaningfully changes that effect
- A plain-language interpretation
- One limitation or follow-up question

Reflection prompt

If you had to present this in a 5-minute conference talk, which figure and which one-sentence result would you lead with?

Hints & solution outline

No single correct answer. This section is graded on the reasoning and the articulation, not the exact numbers.

Reference tutorials: Section 2 (full APIM), Section 4 (gender moderation), Section 8 (focal interaction).

For the figure. A two-panel plot is the standard. Left panel: the partner crossover effect at low and high levels of the moderator (slope plot). Right panel: the predicted outcome as a function of the partner predictor, with one line per moderator level. The interactions::interact_plot() function produces a version of the right panel automatically.

For the lead sentence. A strong lead sentence states the direction and magnitude of the focal effect, names the moderator, and gives the practical takeaway. Example structure: “A one-SD increase in [partner predictor] is associated with a [direction] change of [magnitude] units in [outcome], and this effect is [stronger/weaker/unchanged] among [dyads with X].”

What to record. The full report (numbers, fit indices, interpretation paragraph) is the deliverable. The reflection prompt asks for self-critique: which figure and which sentence would you lead with? Why that one?

--- title: "Section 9 — Integrative exercise" --- # Section 9 — Integrative exercise **What you are practising:** combining everything you have learned to answer a single, fully-specified research question. ::: {.callout-note} ## Goal Pick one of the three outcomes. Build a kitchen-sink multilevel model and the matching wide-format SEM. Report a 4–6 sentence interpretation paragraph. ::: ## Tasks 1. **Choose an outcome.** Pick whichever of `engagement`, `performance`, `creativity` is most interesting to you. 2. **Build a kitchen-sink multilevel model.** Include: - All three actor predictors - All three partner predictors - All three dyad-level moderators - The focal crossover-by-moderator interaction for your chosen outcome - A random intercept for `dyad_id` 3. **Build the corresponding wide-format SEM.** Include the manual product indicator for the focal interaction and the computed simple-slope parameters at three levels of the moderator (low, mean, high). 4. **Report:** - The actor and partner effect estimates with standard errors - The focal interaction coefficient from both approaches - The three simple slopes (low, mean, high levels of the moderator) - The SEM fit indices (chi-square, df, p-value, CFI, TLI, RMSEA, SRMR) 5. **Write a 4–6 sentence interpretation paragraph** covering: - The substantive size and direction of the partner crossover effect - Whether the moderator meaningfully changes that effect - A plain-language interpretation - One limitation or follow-up question ## Reflection prompt If you had to present this in a 5-minute conference talk, which figure and which one-sentence result would you lead with? ::: {.callout-tip collapse="true"} ## Hints & solution outline **No single correct answer.** This section is graded on the *reasoning* and the *articulation*, not the exact numbers. **Reference tutorials:** [Section 2](../../tutorials/indistinguishable/mlm.html) (full APIM), [Section 4](../../tutorials/distinguishable/mlm-moderator.html) (gender moderation), [Section 8](../../tutorials/moderated/hahn2014.html) (focal interaction). **For the figure.** A two-panel plot is the standard. Left panel: the partner crossover effect at low and high levels of the moderator (slope plot). Right panel: the predicted outcome as a function of the partner predictor, with one line per moderator level. The `interactions::interact_plot()` function produces a version of the right panel automatically. **For the lead sentence.** A strong lead sentence states the *direction* and *magnitude* of the focal effect, names the moderator, and gives the practical takeaway. Example structure: *"A one-SD increase in [partner predictor] is associated with a [direction] change of [magnitude] units in [outcome], and this effect is [stronger/weaker/unchanged] among [dyads with X]."* **What to record.** The full report (numbers, fit indices, interpretation paragraph) is the deliverable. The reflection prompt asks for self-critique: which figure and which sentence would you lead with? Why that one? :::