Pipeline stage: test blueprint and construct definition

States what the test measures and how it is built: the construct, the content cells with a target number of items in each, the item types, the intended score use and the target reliability. The blueprint is the contract later stages check against: generate_items() fills its cells, item coverage is counted against its targets, and the target reliability replaces the reliability rule used at calibration.

Usage

set_blueprint(
  p,
  cells,
  construct = NULL,
  use = NULL,
  item_types = "mc",
  length = NULL,
  population = NULL,
  grade_level = NULL,
  reliability = NULL
)

Arguments

p: a pipeline from mv_pipeline().
cells: the content cells: a data frame with a column content, an optional column cognitive (cognitive level, such as recall or application) and a column n with the target number of items per test form; or a named vector of target counts, with content areas as names.
construct: a sentence describing what the test measures.
use: the intended use of the scores, such as "pass/fail", "classification", "ranking", "diagnostic" or "growth".
item_types: item types to write: "mc" (multiple choice) and/or "open" (open-ended, scored with a rubric).
length: number of items on a form; defaults to the sum of the cell targets.
population: a description of the test takers (used when writing items).
grade_level: optional target reading grade level for item text.
reliability: optional target reliability; replaces the reliability rule.

Value

The pipeline with p$blueprint, a list holding the settings and cells, a tibble with one row per cell (cell, content, cognitive, n).

Details

The checks:

empty_cell (block): a content cell has no target count (zero or missing).
score_use (review): the intended use of the scores is not stated; decisions about cut scores, reliability and reporting depend on it.
target_reliability (info): no target reliability was given, so the reliability rule (0.80 by default) is used.
test_length (info): the stated test length differs from the sum of the cell targets.

Examples

bp <- data.frame(content = c("Algebra", "Algebra", "Geometry"),
                 cognitive = c("Recall", "Application", "Application"),
                 n = c(5, 10, 5))
p <- set_blueprint(mv_pipeline(), bp, construct = "Grade 8 mathematics",
                   use = "pass/fail", reliability = 0.85)
p$blueprint$cells
#> # A tibble: 3 × 4
#>   cell                   content  cognitive       n
#>   <chr>                  <chr>    <chr>       <dbl>
#> 1 Algebra / Recall       Algebra  Recall          5
#> 2 Algebra / Application  Algebra  Application    10
#> 3 Geometry / Application Geometry Application     5