R package for discrete option test analysis

library("devtools") # if not available: install.packages("devtools")
install_github("m-Py/DOTA2")

# load the package via 
library("DOTA2")

The function get_response_table can be used to read test data from an Unipark-exported JSON string:

## `testData` is a vector of Unipark-exported JSON strings; `id` is a
## vector of unique test-taker IDs. 

test_dat <- get_response_table(testData, id)

The by_option parameter controls whether response data is retrieved on the level of an item or on the level of each option in each item. It defaults to FALSE, i.e. in the default case we retrieve response data on item level. If by_option is TRUE we retrieve response data for each response option.

item_data <- get_response_table(testData, id, by_option = FALSE)
option_data <- get_response_table(testData, id, by_option = TRUE)

get_response_table returns a data.frame in long format.

If by_option is FALSE, each row represents an item, the test-taker who responded to the item is represented in the column id. The column item_id contains the ID representing the item. Additionally (as of the time of this writing), the following columns are stored:

• correct: was the item answered correctly
• option_id_selected: the ID of the response option that was selected
• position_solution: Where was the solution shown (1 means it was the option at the top right below the stem)
• item_position: Informs about the position of the item in the presentation of all test items (1 means it was the first item that was presented in the test)
• response_time: the response time in ms

If the test is a DOMC test, the following information is also stored:

• false_alarm: was a distractor incorrectly accepted
• miss: was the solution incorrectly rejected
• n_options_seen: how many response options did the test-taker see

These columns have a different meaning in a DOMC test:

• option_id_selected can be 0, meaning that no option was accepted
• position_solution: which option was the solution in the sequential presentation order

If by_option is TRUE, each row represents an option. The test-taker who responded to the option is represented in the column id. The column item_id contains the ID representing the item the option belongs to. The column option_id contains the ID of the response option. Additionally (as of the time of this writing), the following columns are stored:

• selected: was the option selected (1/0)
• is_solution: is the option a solution (1/0)
• option_position: the position of the option in the presentation order

For DOMC tests, the following additional data is stored:

• seen: was this option presented to the test-taker (1/0)
• hit_by_option: did a hit occur when processing this option (1/0/NA)
  • 1 means this option was the solution and was accepted
  • 0 means this option was the solution and was rejected
  • NA means this option was not a solution or was not seen
• false_alarm_by_option: did a false alarm occur when processing this option (1/0/NA)
  • 1 means this option was a distractor and was accepted
  • 0 means this option was a distractor and was rejected
  • NA means this option was not a distractor or was not processed
• response_time_by_option: the response time for the option in ms

Option and item response data can be joined if that is desired (it does however make the interpretation of column names more ambiguous):

merge(item_data, option_data, by = c("id", "item_id"))