Skip to contents

Results dashboard for sharing

Source: R/dashboard.R
mv_dashboard.Rd

Builds a small, read-only 'shiny' dashboard of a fitted model for people who do not use R: teachers, clients, committees or students. Where the workbench is for doing an analysis, the dashboard is for showing one. Pick the panels you want, give it a title, and run it, or save it as a ready-to-deploy app or a single offline HTML file with export_dashboard().

Usage

mv_dashboard(
  x,
  panels = NULL,
  title = "Assessment results",
  subtitle = NULL,
  logo = NULL,
  dif = NULL,
  rt = NULL,
  ids = NULL,
  method = "EAP",
  cards = NULL,
  alpha = 0.01
)

Arguments

x

a model from mirt::mirt(), an mv_model_selection object from select_model() (its recommended model is shown), or a dashboard object saved by export_dashboard().

panels

character vector of panels to show, in order. The default shows "summary", "items", "curves", "information", "wright" and "scores", plus "dif" and "rt" when those results are supplied.

title, subtitle

text for the dashboard header.

optional path to a PNG, JPEG or SVG image shown in the header.

dif

optional mv_dif object from dif().

rt

optional mv_rt object from fit_rt().

ids

optional vector of person identifiers for the "persons" panel, one per row of the data the model was fitted to. Defaults to row numbers.

method

scoring method for person scores, passed to mirt::fscores().

cards

logical; prepare an item health card for every item (shown when you select an item). Defaults to TRUE for unidimensional tests of up to 40 items, because each card takes a moment to draw.

alpha

significance level for item misfit flags.

Value

A 'shiny' app object. Printing it (or calling the function at the console) opens the dashboard; shiny::runApp() accepts it too. The summary object is stored in its "mv_dashboard" attribute.

Details

Everything the dashboard shows is computed once when it is built and kept in a summary object; the app never refits or rescores anything. That summary holds no item responses. Individual person scores are included only when you ask for the "persons" panel, so check that you may share them before you deploy a dashboard that has it.

Available panels:

  • "summary": headline numbers (respondents, items, reliability, flagged items) and the plain-language interpretation from interpret().

  • "items": the item table with traffic-light status; select a row to see that item's health card (item_card()).

  • "curves": trace curves for chosen items, with a slider that reads off the probabilities at any ability level.

  • "information": test information and standard error, plus item information curves.

  • "wright": the Wright map.

  • "scores": the distribution of ability estimates and the sum-score conversion table (no individual scores).

  • "persons": a searchable table of every person's score, standard error and percentile.

  • "dif": differential item functioning results (needs dif).

  • "rt": response-time results (needs rt).

"curves", "information", "wright", "scores" and "persons" need a unidimensional model and are dropped, with a message, for other models.

See also

export_dashboard() to save a dashboard as a Shiny app folder or a single offline HTML file, mv_report() for a static HTML report.

Examples

# \donttest{
mod <- mirt(Science, 1, verbose = FALSE)
app <- mv_dashboard(mod, title = "Science attitudes survey",
                    panels = c("summary", "items", "curves"))
#> Drawing 4 item health cards...
if (interactive()) app
# }