Discover analysis requirements
Value
A data frame with analysis, layer, required tables, required fields, unit, denominator, and assumptions. Semicolon-separated entries are AND requirements; additional checks are made on each event's usable data.
Examples
analysis_requirements()[c("analysis", "tables", "assumptions")]
#> analysis
#> 1 sample
#> 2 attendance
#> 3 assignment
#> 4 participation
#> 5 exchange
#> 6 responsiveness
#> 7 arguments
#> 8 materials
#> 9 exposure
#> 10 outcomes
#> 11 group_effects
#> 12 randomization
#> 13 survey_inference
#> 14 attrition
#> 15 cluster_inference
#> tables
#> 1 people;benchmarks
#> 2 people;recruitment
#> 3 people;assignments
#> 4 people;sessions;attendance;turns;coverage
#> 5 sessions;turns;annotations
#> 6 sessions;turns;annotations;links
#> 7 sessions;turns;arguments;links
#> 8 sessions;materials;arguments
#> 9 sessions;turns;arguments;links;attendance;items;waves;responses
#> 10 people;episodes;items;waves;responses
#> 11 people;assignments;items;waves;responses
#> 12 people;assignments;items;waves;responses;randomization
#> 13 people;items;waves;responses;sampling
#> 14 people;items;waves;responses
#> 15 people;items;waves;responses
#> assumptions
#> 1 Comparable target population and measurement; marginal balance only
#> 2 Predictors measured before deliberation; AUC is descriptive
#> 3 Balance does not prove randomization
#> 4 Silence requires complete recording and identified speakers; overlap is not interruption
#> 5 Conditional on supplied coding; unannotated is unknown
#> 6 Links identify substantive responses
#> 7 Reviewed inventory supports vetted coverage, not objective truth
#> 8 Material availability precedes the session; review is attributed
#> 9 Temporal opportunity to encounter an opposing argument, not evidence of attention
#> 10 Descriptive pre/post change; P requires a substantive midpoint
#> 11 Baseline-adjusted group associations; moderator confounding checked
#> 12 Sharp null of no assignment effects; fixed-size complete or blocked randomization
#> 13 Declared probability sampling; response ignorability is not established
#> 14 Post missingness among baseline respondents; roster-only absence is separate
#> 15 Explicit independent-cluster declaration and target; model-based CR2 inference