Multiple Choice tasks

#> Registered S3 method overwritten by 'textutils':
#>   method             from 
#>   toLatex.data.frame exams

This is just a normal multiple choice task.

Minimum version

A minimum template is automatically created when you initiate an rqti project through RStudio. Alternatively, it can be added by clicking on New file -> R Markdown -> From Template. The rqti templates end with rqti. Here we look at the templates multiplechoice (simple) and multiplechoice (complex).

The minimum you need to provide is the type: mpc (or the equivalent type: multiplechoice or type: mchoice) in the YAML section and a list with at least two elements in a section called # question:

---
type: mpc
knit: rqti::render_qtijs
---

# question

Which exercise types are supported by the R package rqti?

- *dropdown list*
- *gaps*
- programming language evaluation
- drawing

# feedback

All basic types of QTI are supported, including dropdown list. More advanced
exercises (as available in OPAL/ONYX) are not yet supported because they are LMS
specific.

Note that in this example, a feedback section was also provided. This is optional, but usually it is a good idea to give some explanation for students.

Further note that the knit parameter is set to the custom rqti knit function, which will handle the preview. Clicking the Knit button in RStudio produces:

Multiple choice task rendered in qtijs

You can also use the OPAL (set it up before, see article Working with the OPAL API) render function (knit: rqti::render_opal), which should produce:

Multiple choice task rendered in OPAL

Multiple-choice tasks function similarly to single-choice tasks, with the key difference being that multiple (or no) answers can be correct. To indicate the correct options, surround them with asterisks (using markdown for italics). If you need to use italics within the choices themselves, enclose the entire question in <em></em> tags to avoid conflicts.

For an Rmd item without an explicit points value, the default maximum is 0.5 times the number of correct options. Each selected correct option then earns 0.5 points; penalties for selected incorrect options balance the total positive credit. The default item score is floored at zero. These are the default rules, not the only available scoring methods; see scoring_scheme below for alternatives and multiple_choice_scoring() for its own help page.

More control

If you want to have more fine-grained control, consider the available attributes for the YAML section in the Rmd template multiple-choice (complex).

---
type: multiplechoice # equivalent to mpc
knit: rqti::render_qtijs # if you do not want our preview renderer, remove this
identifier: mpc001 # think twice about this id for later data analysis!
title: A meaningful title that can be displayed in the LMS
shuffle: true # random order of choices
orientation: vertical # OR horizontal
points: 2
---

# question

Which exercise types are supported by the R package rqti?

- *match pair (pair several elements)*
- *match tables (single choice or multiple choice tables)*
- drawing
- graphical match (drag selection to specific position)

# feedback

All basic types of QTI are supported. More advanced exercises (as available in
OPAL/ONYX) are not yet supported because they are LMS specific.

Which renders in OPAL as:

More complex multiple choice task rendered in OPAL

YAML attributes

type

Has to be multiplechoice or mpc or mchoice (compatible with exams package).

identifier

This is the ID of the task, useful for later data analysis of results. The default is the file name. If you are doing extensive data analysis later on, it makes sense to specify a meaningful identifier. In all other cases, the file name should be fine.

title

Title of the task. Can be displayed to students depending on the learning management system settings. The default is the file name.

shuffle

If true, randomizes the order of the choices. Defaults to true. Only in rare occasions it makes sense to have a strict order of choices (setting shuffle to false).

orientation

Should the items be displayed in vertical or horizontal mode? Default is vertical.

points

For an Rmd item, a scalar points sets the maximum for the task. If omitted, the maximum is 0.5 times the number of correct options. In the S4 helper, multipleChoice(points = ...) instead accepts a vector with one entry per option. Positive entries identify the correct options and their sum is the maximum P.

scoring_scheme

The default "standard" uses balanced scoring equivalent to Youden’s J, with negative totals floored at zero. With equally weighted correct options, C correct options and W incorrect options, both greater than zero, the awarded score is

$P\max\left(0, \frac{\#\text{correct options selected}}{C} - \frac{\#\text{incorrect options selected}}{W} \right).$

The raw normalized expression is Youden’s J. It has zero expectation under independent random selection. Flooring negative totals at zero means the awarded score does not generally have zero expectation. If the S4 item supplies explicit negative option weights, standard preserves those weights; use weights to preserve explicit zero weights as well. Unequal correct-option weights or custom penalties need not be equivalent to Youden’s J. The formula above applies to the ordinary automatically balanced default.

Use multiple_choice_scoring() to choose another multiple-response model:

Method	Meaning
`standard`	Default balanced (Youden’s J) mapping, with a zero floor; explicit weights are preserved
`weights`	Supplied option weights, including explicit zeros
`all_or_nothing`	Full credit only for selecting exactly the correct set
`plus_minus`	Equal positive and negative contributions for selected options
`subset`	Credit for selected correct options only if no incorrect option is selected
`ripkey`	Credit for selected correct options unless more than C options are selected
`true_false`	Credit for each correctly classified option, including unchecked incorrect options
`exams`	The R/exams partial-credit, weighting and score-floor settings

For example, select plus/minus scoring in R:

mc <- multipleChoice(
  choices = c("A", "B", "C"), points = c(0.5, 0, 0.5),
  scoring_scheme = multiple_choice_scoring("plus_minus")
)

Or in YAML:

scoring_scheme: plus_minus

Configurable methods accept a named mapping in YAML and named settings in R:

mc_exams <- multipleChoice(
  choices = c("A", "B", "C"), points = c(0.5, 0, 0.5),
  scoring_scheme = multiple_choice_scoring(
    "exams", partial = TRUE, negative = -0.25, rule = "false2"
  )
)

scoring_scheme:
  method: exams
  partial: true
  negative: -0.25
  rule: "false2"

weights and plus_minus accept lower_bound, a nonpositive fraction of P. all_or_nothing accepts incorrect and omitted credits. The exams method accepts all five weighting rules (false2, false, true, all, none), partial, and negative. Quote the "false" and "true" rule names in YAML so that they are parsed as strings rather than logical values. See ?multiple_choice_scoring for defaults and the full definitions.

The true_false method treats unchecked options as false decisions; a blank selection can therefore earn credit for incorrect options left unchecked. Checkboxes do not distinguish such a decision from an omission. This is a different format from asking for an explicit true/false response to every statement. The single-choice kanzow_* presets, and method 21’s statement_truth setting, belong to single_choice_scoring() and are not accepted by multiple_choice_scoring().

Feedback

Feedback can be provided with the section

# feedback (general feedback, displayed every time, without conditions)
# feedback+ (only provided if student reaches all points)
# feedback- (only provided if student does not reach all points)

List of answers as a variable

For more complex tasks the list of answers is often just available as a variable. In this case you can use the helper function mdlist to convert the vector into a markdown list:

mdlist(c("dropdown list", "programming language evaluation", "numeric gap"), 
       solutions = c(1, 3))
[1] "- *dropdown list*\n- programming language evaluation\n- *numeric gap*"

Some advice on multiple choice tasks

A multiple-choice task can always be converted into multiple single-choice questions with true/false or yes/no options. From a psychometric standpoint, both types of tasks share the same drawbacks, primarily due to the potential for guessing. Consequently, they should be avoided when possible. Generally, their psychometric properties are inferior to those of numeric or string-based gap tasks that assess similar content.

However, there are scenarios where forced-choice tasks are necessary. For example, presenting several statistical analyses and asking students to determine whether the results are statistically significant is a valuable task. Even in such cases, a multiple-choice format may not be the best option. Better alternatives include single-choice questions, dropdowns, or match tables. A multiple-choice question can often be restructured into several single-choice or dropdown questions with yes/no options. This may be cumbersome for longer multiple-choice lists, for which a match table with yes/no options can be a more convenient solution.

The main advantage of these alternative formats is that they require students to make explicit choices. In a multiple-choice task where all options are incorrect, a student could earn full points simply by not making any selections—effectively getting credit without engaging with the question. To avoid this, multiple-choice tasks must balance the distribution of correct and incorrect options, which is not necessary for single-choice, dropdown, or match table formats.

While multiple-choice tasks are supported, we strongly advise against using them whenever possible in favor of more robust alternatives.

2026-10-23