In kthorstmann/quest:

library(quest)

The following documentation will present an example on how to use the pacakge quest. The quest pacakge was designed to take do the work that is done whenver an already published questionnaire is used on a sample.

The quest package should minimize the work that is always done when analyzing questionnaire data. It relies heavily on the packages psychand lavaan to do subsequent analysis.

The package begins its work where the collection of raw data has endet: Usually, a data set is obtained that contains the raw answers of participants (for example on a 1 to 7 type scale) and a table is presented that shows which item belongs to which scale and whether it needs to be recoded or not. This is where the work of the questpackage begins.

Example - The Big Five Short Inventory (BFI-K)

To exemplify the use of the package, consider the following example using data from the Big Five Inventory. First of all, we need to install and load the package quest.

## Install from github not yet available
library(quest)

library(quest)

The package contains a data frame bigfive , which contains $N = r nrow(bigfive)$ subjects that have completed the Big Five questionnaire, a questionnaire on the dark triad, as well as newly designed items on narcisism.

Keytable and shortkeys

The followoing table from the Big Five Inventar (21 Items) shows how each item in the questionnaire has to be coded and to which scale it belongs:

|Scale|Extraversion|Agreeableness|Conscientiousness|Neuroticism|Openness| |:-|----:|---:|---:|---:|----:| | | 1R |2R | 3 | 4 | 5 | | | 6 |7 | 8R | 9R | 10 | | | 11R |12R | 13 | 14 |15 | | | 16 |17R | 18 | 19 | 20 | | | | | | | 21 |

You can hand this structure over to the quest package using the following code (known as a shortkey):

scales <-  c("E = 1r, 6, 11R, 16",
             "A = 2r, 7, 12r, 17r",
             "C = 3, 8r, 13, 18",
             "N = 4, 9r, 14, 19",
             "O = 5, 10, 15, 20 21")

For some questionnaires (see ?shortKeys), the shortkeys are already in the package and can be requested using the shortKeys function:

shortKeys("BFI-K")

The shortkey contains the following information:

1. How many items does the questionnaire have?
2. Which item belongs to which scale?
3. Which item needs to be recoded?

Preparation of data

The second ingredient to the usage of the quest package is the data. In the package, a data frame is already provided to exemplify its use. Type bigfive to get the data frame:

data <- bigfive

The first 21 items are relevant for the BFI. All items are in the original order (as they should be). You need to extract the exact items into a separate data frame for the usage of the quest package (another option is providing the codebook from your study, see section on using a codebook).

bfik.items <- grep(pattern = "BFI_", x = names(data), value = TRUE)
bfik.data <- data[bfik.items]

Recoding & computation of scales

After having selected the data frame that contains only the items from the questionnaire in the correct order, you can use this data frame together with the shortkeys to proceed in your analysis.

itemkey.1 <- scale2list(scales[1])
itemkey.1

To extract the five scales from the questionnaires, make first a list of all the keys (nested list).

itemkeys <- purrr::map(scales, scale2list)
itemkeys[1]

The itemkeys can then be used to transform the data set to a final data frame with recoded values and scale scores.

# Extract all BFI data:
bfi.data <- data[grep(x = names(data), pattern = "BFI_", value = TRUE)]

bfi.scales <- codeScales(bfi.data, scales, recode = TRUE, return.both=TRUE)
head(bfi.scales)

This data frame can then be used for further analyses.

Measurement models

Sometimes, the use of measurement models is neccessary. Using the lavaan package can be used. To get the measurement models for your questionnaire, just use:

# Extract all BFI data:
bfi.data <- data[grep(x = names(data), pattern = "BFI_", value = TRUE)]
mModels(data = bfi.data, scales = scales, run.models = FALSE)

You can also fit the models directly, using the following code:

bfi.data <- data[grep(x = names(data), pattern = "BFI_", value = TRUE)]
mModels(data = bfi.data, scales = scales, run.models = TRUE)

You can also use the function NOT_YET_WRITTEN to take your codebook (which you should have) and let this object be generated.

kthorstmann/quest documentation built on May 20, 2019, 7:05 p.m.