Introduction

Tools for working with the Qualtrics API that prioritize portability, convenience, and security.

Right now, with qualtr, you can:

• keep configuration details encrypted on disk
• render surveys as elegant pdf documents
• download response data with variable labels and appropriate factor levels
• tabulate labelled summaries suitable for further analysis, or formatted for reporting

In future, I’m hoping to integrate:

• contact list retrieval and management
• support for coding likert scale batteries
• EDA oriented plotting functions

This package was inspired by Jasper Ginn’s qualtRics package, and a good deal of the functionality overlaps.

Getting started

Install the latest version of qualtr

# devtools::install_github("context-dependent/qualtr")

library(qualtr)
library(tidyverse)

Encrypt configuration details in package root

Other packages ask you to store sensitive configuration details in unencrypted yaml files. While this offers the obvious convenience of not requiring a password, it is insufficiently secure for our needs. encrypt_conf asks for your api_token, datacentre url, and key. It saves these details locally in an encrypted .rds file in the package root folder. Run the following chunk in the console, replacing the capitalized parameters with your account’s details

encrypt_conf(
  api_token = API_TOKEN,
  # base_url = BASE_URL, 
  key = PASSWORD
)

Confirm your saved configuration details by decrypting them

decrypt_conf is used internally to decrypt your configuration details, stored locally in .conf.rds in the package root. You can call it at the top level to confirm that the credentials are successfully stored.

decrypt_conf()

Finding surveys

list_surveys returns a list of all surveys associated with your API key.

list_surveys()

By default, more recently edited surveys are displayed first, but if you pass a search term to list_surveys, they will be ordered by their proximity to that.

skills_catalyst_surveys <- list_surveys("Skills Catalyst")

skills_catalyst_surveys

Sometimes, for reasons unknown, list_surveys does not capture some surveys. In these cases, you can navigate to the project in the Qualtrics interface and copy the survey ID from the url.

For the functions in the package that retrieve a survey’s questions or responses, you can specify the survey either by id, passing it as a quoted string, or by row in the last result of list_surveys. In this case, for example, you could retrieve the same survey by either passing the number 2 or “SV_0OJXn0xdyrgkl9P”. Examples of use cases presented later will make the application clear, but generally speaking, it’s ok to refer to a survey by row number if you’re working with them interactively, but better to use id if you’re coding its retrieval into a script.

Print survey questions

Download survey responses

skills_catalyst_js_exit_responses <- get_responses_v2("SV_0OJXn0xdyrgkl9P")

Tabulate scale data

Generate raw table for further analysis

satisfaction_table_raw <- 

  skills_catalyst_js_exit_responses %>% 

  qt_raw(
    vars(matches("sat"))
  )

satisfaction_table_raw

Format raw table for printing

satisfaction_table_print <- 

  satisfaction_table_raw %>% 

  qt_print()

satisfaction_table_print

Tabulate by group

satisfaction_by_employment_table <- 

  skills_catalyst_js_exit_responses %>% 

    rename(employed = attach_4) %>% 
    group_by(employed) %>% 
    qt_raw(vars(matches("sat"))) %>% 
    qt_print()

satisfaction_by_employment_table

Treating likert responses as numeric

Recode values

skills_catalyst_js_exit_responses %>% 

  scr_num(

    # .vars specifies the variables you want to recode using a tidyselect query
    # wrapped in a vars function
    .vars = vars(matches("sat_1")),

    # .rev specifies, among the variables selected in .vars, which will be reverse coded
    # by default, any variable that includes "_r" is reverse coded
    # .rev = vars(matches("_r")),
    scale = "agree"
  ) %>% 

  select(
    matches("sat_1")
  )

Scoring scales

score_scale() returns a tibble with the items, total, and average score per item. The .vars, .rev, and scale arguments work the same way.

skills_catalyst_js_exit_responses %>% 

  score_scale(

    var_name = "sat_1",
    .vars = vars(matches("sat_1")), 
    .rev = vars(matches("_r")), 
    scale = "agree"

  )

Printing surveys

skills_catalyst_js_exit_questions <- get_survey("SV_0OJXn0xdyrgkl9P")

print_survey(skills_catalyst_js_exit_questions, "prints/2019-02-27_scjs_exit.tex")