dyad_data.RData

`dyad_data.RData`

The dataset for the eight core tutorials. Generated by scripts/01_simulate_data.R.

Download dyad_data.RData{.btn .btn-primary download}

Object	Format	Dimensions	Description
`ddl`	long	200 rows × 10 columns	One row per person
`ddw`	wide	100 rows × 10 columns	One row per dyad

The two objects represent the same 100 dyads.

Variables

Person-level (in long format `ddl`)

Variable	Type	Description
`dyad_id`	integer	Dyad identifier (1–100)
`person_id`	integer	Person within dyad (1 or 2)
`gender`	factor (`male`, `female`)	Distinguishing variable
`wnc`	numeric	Work–nonwork conflict (centred)
`partner_wnc`	numeric	Partner’s WNC
`recovery`	numeric	Recovery from work (centred)
`partner_recovery`	numeric	Partner’s recovery
`satisfaction`	numeric	Relationship satisfaction
`has_children`	numeric (0/1)	Dyad-level covariate
`dual_earner`	numeric (0/1)	Dyad-level covariate

Wide-format equivalents (in `ddw`)

Variable	Meaning
`wnc_a`, `wnc_p`	WNC for the actor / partner
`recovery_a`, `recovery_p`	Recovery for the actor / partner
`satisfaction_a`, `satisfaction_p`	Satisfaction for the actor / partner
`has_children`, `dual_earner`	Same as in long format

By convention, in ddw, _a corresponds to male and _p corresponds to female for the scripts that analyse distinguishable dyads.

Data-generating parameters

The simulation uses the following true values, set in the script’s header. Your estimates should be close to these.

Fixed effects

Parameter	Value	Meaning
`a_wnc`	−0.30	Actor WNC effect
`p_wnc`	−0.15	Partner WNC effect
`a_rec`	+0.25	Actor recovery effect
`p_rec`	+0.10	Partner recovery effect
`c_child`	−0.10	Children effect
`c_dual`	+0.20	Dual-earner effect
`int_child_pwnc`	+0.10	Children × partner WNC
`alpha_m`	5.00	Male intercept
`alpha_f`	4.85	Female intercept

Variance components

Parameter	Value	Meaning
`rho_wnc`	0.40	Within-dyad WNC correlation
`rho_rec`	0.35	Within-dyad recovery correlation
`rho_sat`	0.30	Within-dyad satisfaction residual correlation

Design choices

Equal slopes across gender, different intercepts. This makes indistinguishability an empirical question, not a property of the data. The k-patterns page discusses why this matters.
Within-dyad correlations induced by correlated residuals (correlate_errors() in the script). The dyad members are correlated for both substantive and methodological reasons.
Children as a moderator. The int_child_pwnc = 0.10 parameter is the headline effect in the Hahn et al. (2014) replication tutorial. It represents the finding that children buffer the partner crossover effect.

Reproducibility

set.seed(2026) is set at the top of the simulation script. Re-running scripts/01_simulate_data.R will always produce the same RData file.

Loading

load("data/dyad_data.RData")
ls()                 # "ddl"  "ddw"
head(ddl, 3)
head(ddw, 3)

--- title: "dyad_data.RData" --- # `dyad_data.RData` The dataset for the eight core tutorials. Generated by [`scripts/01_simulate_data.R`](https://github.com/michaelides/apim/blob/master/scripts/01_simulate_data.R). [Download dyad_data.RData](https://github.com/michaelides/apim/raw/master/data/dyad_data.RData){.btn .btn-primary download} ## Contents | Object | Format | Dimensions | Description | |---|---|---|---| | `ddl` | long | 200 rows × 10 columns | One row per person | | `ddw` | wide | 100 rows × 10 columns | One row per dyad | The two objects represent the same 100 dyads. ## Variables ### Person-level (in long format `ddl`) | Variable | Type | Description | |---|---|---| | `dyad_id` | integer | Dyad identifier (1–100) | | `person_id` | integer | Person within dyad (1 or 2) | | `gender` | factor (`male`, `female`) | Distinguishing variable | | `wnc` | numeric | Work–nonwork conflict (centred) | | `partner_wnc` | numeric | Partner's WNC | | `recovery` | numeric | Recovery from work (centred) | | `partner_recovery` | numeric | Partner's recovery | | `satisfaction` | numeric | Relationship satisfaction | | `has_children` | numeric (0/1) | Dyad-level covariate | | `dual_earner` | numeric (0/1) | Dyad-level covariate | ### Wide-format equivalents (in `ddw`) | Variable | Meaning | |---|---| | `wnc_a`, `wnc_p` | WNC for the actor / partner | | `recovery_a`, `recovery_p` | Recovery for the actor / partner | | `satisfaction_a`, `satisfaction_p` | Satisfaction for the actor / partner | | `has_children`, `dual_earner` | Same as in long format | By convention, in `ddw`, `_a` corresponds to `male` and `_p` corresponds to `female` for the scripts that analyse distinguishable dyads. ## Data-generating parameters The simulation uses the following true values, set in the script's header. Your estimates should be close to these. ### Fixed effects | Parameter | Value | Meaning | |---|---|---| | `a_wnc` | −0.30 | Actor WNC effect | | `p_wnc` | −0.15 | Partner WNC effect | | `a_rec` | +0.25 | Actor recovery effect | | `p_rec` | +0.10 | Partner recovery effect | | `c_child` | −0.10 | Children effect | | `c_dual` | +0.20 | Dual-earner effect | | `int_child_pwnc` | +0.10 | Children × partner WNC | | `alpha_m` | 5.00 | Male intercept | | `alpha_f` | 4.85 | Female intercept | ### Variance components | Parameter | Value | Meaning | |---|---|---| | `rho_wnc` | 0.40 | Within-dyad WNC correlation | | `rho_rec` | 0.35 | Within-dyad recovery correlation | | `rho_sat` | 0.30 | Within-dyad satisfaction residual correlation | ## Design choices - **Equal slopes across gender, different intercepts.** This makes indistinguishability an *empirical* question, not a property of the data. The [k-patterns](../foundations/kpatterns.html) page discusses why this matters. - **Within-dyad correlations induced by correlated residuals** (`correlate_errors()` in the script). The dyad members are correlated for both substantive and methodological reasons. - **Children as a moderator.** The `int_child_pwnc = 0.10` parameter is the headline effect in the [Hahn et al. (2014) replication tutorial](../tutorials/moderated/hahn2014.html). It represents the finding that children buffer the partner crossover effect. ## Reproducibility `set.seed(2026)` is set at the top of the simulation script. Re-running `scripts/01_simulate_data.R` will always produce the same `RData` file. ## Loading ```r load("data/dyad_data.RData") ls() # "ddl" "ddw" head(ddl, 3) head(ddw, 3) ```