In markdly/conquestr: Conquest Version 2 and R

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

This vignette is intended to walkthrough the use of existing conquestr functions.

# install.packages("devtools")
devtools::install_github("markdly/conquestr")

# setup a temporary directory for the examples
mydir <- tempdir()

A typical analysis using ConQuest software involves an item analysis based on student responses to an assessment the students attempted (e.g. mathematics test). conquestr comes with an inbuilt dataset (short_test) which contains multiple-choice responses to 9 questions. The test was attempted r nrow(conquestr::short_test) times.

library(dplyr)
library(conquestr)
df <- as.data.frame(short_test)
head(df)

The pid column contains a unique identifier for each test attempt. The responses to the 9 questions are contained in columns q1 to q9. Values indicate the answer option chosen with NA values indicating a missing response. For example, for test attempt 10001, the 5th answer option was chosen for q2.

For these examples the pid isn't necessary. At this stage we are interested in item responses only.

df <- select(df, -pid)

The answer key contains the correct answer for each question

answer_key <- dplyr::arrange(short_test_score_key, item)
answer_key

The next step is to take the data we want to analyse and convert it into a format ConQuest can use. We can do this by using the cqc_data function which writes out a dataframe of responses in fixed-width text format to a filename provided. To keep everything together we'll call this analysis foo.

data_specs <- cqc_data(df, file.path(mydir, "foo.txt"))
data_specs

For this example, cqc_data has written df to the file "foo.txt" in fixed-width format and returns details about the columns in foo.txt.

Labels for items can be created using cqc_label.

foo_labels <- cqc_label(names(df))
writeLines(foo_labels, file.path(mydir, "foo.lab")) 

ConQuest command syntax can be generated using cqc_syntax. It requires a name argument which is used to provide a default link between label, data, and syntax files. The resp_cols argument is used to define the ConQuest responses columns which are required by the ConQuest format statment. The cqc_resp_cols function can work with the output from the cqc_data function to supply an appropriate argument to resp_cols. Just providing the required arguments generates some default syntax which is compatible with ConQuest:

foo_resp_cols <- cqc_resp_cols(data_specs$position)
foo_syntax <- cqc_syntax(name = "foo", resp_cols = foo_resp_cols)
writeLines(unlist(foo_syntax))

The default syntax file may not be useful in the current form. The commands included in the syntax file, and how they are 'linked' can be customised using the optional cmds and lookup_vals arguments.

For example, let's say we are only interested in generating an item analysis and are not concerned with generating a show file, log file or system file. To do this we can customise the result of cqc_cmds() which is the default value provided to the cmds argument in cqc_syntax.

foo_cmds <- cqc_cmds() 
foo_cmds <- foo_cmds[! names(foo_cmds) %in% c("show", "logfile", "put")]

Also, we should add a ConQuest score key given the data are raw responses and add a ConQuest codes command so that our item analysis has useful output. To make these changes we can use cqc_defaults() as a starting point for us to customise. The changes can then be passed to the lookup_vals argument in cqc_syntax.

foo_lookup <- cqc_defaults()
foo_lookup$codes <- "12345"
foo_lookup$key <- paste(answer_key$resp, collapse = "")

Now we can regenerate our syntax this time using our custom changes along with a custom filename:

foo_syntax <- cqc_syntax("foo", foo_resp_cols, filename = "myfile", cmds = foo_cmds, lookup_vals = foo_lookup)
writeLines(unlist(foo_syntax)) 

How cmds and lookup_vals relate

The lookup_vals are used to fill in the 'placeholders' (e.g. {title}) specified by the cmds argument . If a placeholder is set to NULL then the command is not included in the generated syntax.

After the syntax file is written out,

writeLines(unlist(foo_syntax), file.path(mydir, "foo.cqc")) 

the foo analysis syntax, label and data files can be used by ConQuest. After ConQuest has been run, the ConQuest output can be imported.

For example, to read in the ConQuest item analysis file we can use cq_itanal (see the readme at https://github.com/markdly/conquestr for a worked example)

markdly/conquestr documentation built on May 29, 2019, 5:40 a.m.