Declare a DMSA design: covariates, dependence, and permutation blocks

Usage

dmsa_design(
  focal,
  fixed,
  random = NULL,
  exchangeable = NULL,
  forbid = character(),
  label = NULL
)

Arguments

focal: Character. The focal term(s) whose pooled aligned effect is tested. May be a main effect ("IC_c"), an interaction written with a colon ("time:BSI_c"), or several terms, in which case the first is the tested one and the rest are mutually-adjusted co-focal terms.
fixed: Character vector of fixed-effect covariate columns. These are the build's contract, in full - no defaults are supplied, because a default is exactly what goes stale.
random: Grouping factors for random intercepts: a character vector (c("chip","cID"), expanded to (1|chip) + (1|cID)) or a one-sided formula (~ (1|chip) + (1|cID)). NULL declares independent observations - which dmsa_check_design() will challenge if it finds repeated block IDs.
exchangeable: Character. The column defining the outermost independent unit for permutation: couples share a family, repeated measures share a person, and a person sits inside a family - so the exchangeable unit is the family in all three cases. Rows inside a block move together and keep their order, which is what makes the same mechanism valid cross-sectionally and longitudinally.
forbid: Character vector of columns that must NOT enter the model (the build's never-list). Supplying these turns a documentation rule into an error.
label: Optional name for printing.

Value

An object of class dmsa_design.

Examples

dmsa_design(focal = "IC_c",
            fixed = c("sex_c", "age", "Fib", "ctrlSV3", "ctrlSV5"),
            random = c("chip", "cID"), exchangeable = "cID",
            forbid = c("ctrlSV4", "plate"))
#> dmsa design
#>   focal      IC_c
#>   fixed       sex_c, age, Fib, ctrlSV3, ctrlSV5 
#>   random      (1|chip) + (1|cID) 
#>   exchange    cID 
#>   never       ctrlSV4, plate