cabat: CA-BAT

Description Usage Arguments Note

View source: R/cabat.R

Description

This function defines a CA-BAT module for incorporation into a psychTestR timeline. Use this function if you want to include the CA-BAT in a battery of other tests, or if you want to add custom psychTestR pages to your test timeline. For demoing the CA-BAT, consider using demo_cabat(). For a standalone implementation of the CA-BAT, consider using standalone_cabat().

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
cabat(
  num_items = 25L,
  take_training = TRUE,
  label = "BAT",
  feedback = cabat.feedback.no_score(),
  item_bank_audio = "https://media.gold-msi.org/test_materials/BAT/v1/audio",
  practice_items = "https://media.gold-msi.org/test_materials/BAT/v1/practice-items",
  next_item.criterion = "bOpt",
  next_item.estimator = "BM",
  next_item.prior_dist = "norm",
  next_item.prior_par = c(0, 1),
  final_ability.estimator = "WL",
  constrain_answers = FALSE,
  fix_first_item = FALSE,
  dict = cabat::cabat_dict
)

Arguments

num_items

(Integer scalar) Number of items in the test.

take_training

(Logical scalar) Whether to include the training phase.

label

(Character scalar) Label to give the CA-BAT results in the output file.

feedback

Defines the feedback to give the participant at the end of the test. By default no feedback is given. This can be a timeline segment (as created by new_timeline), a test element (as created by e.g. page), or a list of test elements. The following built-in choices are available see function-level documentation for details):

  • cabat.feedback.no_score

  • cabat.feedback.simple_score

  • cat.feedback.graph

  • cat.feedback.irt

  • cat.feedback.iq

item_bank_audio

(Character scalar) File path to the directory hosting the item bank audio (typically a publicly accessible web directory).

practice_items

(Character scalar) File path to the directory hosting the practice items audio (typically a publicly accessible web directory).

next_item.criterion

(Character scalar) Criterion for selecting successive items in the adaptive test. See the criterion argument in nextItem for possible values. "bOpt" corresponds to the setting used in the original CA-BAT paper.

next_item.estimator

(Character scalar) Ability estimation method used for selecting successive items in the adaptive test. See the method argument in thetaEst for possible values. "BM", Bayes modal, corresponds to the setting used in the original CA-BAT paper. "WL", weighted likelihood, corresponds to the default setting used in versions <= 0.3.0 of this package.

next_item.prior_dist

(Character scalar) The type of prior distribution to use when calculating ability estimates for item selection. Ignored if next_item.estimator is not a Bayesian method. Defaults to "norm" for a normal distribution. See the priorDist argument in thetaEst for possible values.

next_item.prior_par

(Numeric vector, length 2) Parameters for the prior distribution; see the priorPar argument in thetaEst for details. Ignored if next_item.estimator is not a Bayesian method. The dfeault is c(0, 1).

final_ability.estimator

Estimation method used for the final ability estimate. See the method argument in thetaEst for possible values. The default is "WL", weighted likelihood, which corresponds to the setting used in the original CA-BAT paper. If a Bayesian method is chosen, its prior distribution will be defined by the next_item.prior_dist and next_item.prior_par arguments.

constrain_answers

(Logical scalar) If TRUE, then item selection will be constrained so that the correct answers are distributed as evenly as possible over the course of the test. We recommend leaving this option disabled.

fix_first_item

(Logical scalar) If TRUE, then the first test item will be constrained to a fixed pair of possible items, following the procedure used in the original CA-BAT paper. We recommend leaving this option disabled.

dict

The psychTestR dictionary used for internationalisation.

Note

Versions <= 0.3.0 of this package experimented with weighted likelihood ability estimation for item selection. Pilot testing found this approach to be problematic, tending to overpenalise the participant for early mistakes. Current versions of the package therefore revert to Bayes modal ability estimation for item selection, consistent with the original CA-BAT paper.


pmcharrison/cabat documentation built on Feb. 11, 2021, 1:44 a.m.