Declare independent clusters for model-based change inference

Clustering describes dependence under a declared repeated-sampling model. It does not establish probability sampling or identify an effect of deliberation.

Usage

cluster_inference(clusters, target, rationale)

Arguments

clusters: Character column names, resolved first in derived outcomes (for example group_id), then assignments, then people. Assignment fields join by event, episode and person; people fields join by event and person. Multiple columns identify one composite cluster, not multiway clustering.
target: Population or repeated-process target of inference.
rationale: Why these clusters can be treated as independent.

Value

A specification for audit_config(cluster = ...).

Examples

cluster_inference("group_id", "Comparable small groups", "Independent groups")
#> $clusters
#> [1] "group_id"
#> 
#> $target
#> [1] "Comparable small groups"
#> 
#> $rationale
#> [1] "Independent groups"
#> 
#> attr(,"class")
#> [1] "cluster_inference"