ShinyPsych

• Shiny is a web application framework for R.
• From the website: Turn your analyses into interactive web applications. No HTML, CSS, or JavaScript knowledge required.
• Shiny applications are called "Apps" (aka, experiment)
• Shiny Example Gallery: http://shiny.rstudio.com/gallery/

• Combination of standard R code, and special Shiny R functions that create and manage interactive elements.
• Shiny has lots of 'widgets' programmed http://shiny.rstudio.com/gallery/widget-gallery.html
• You can copy widget code, and assign the results to an R object.

sliderInput(inputId = "prob.est", 
            label = "What is the probability of X?", 
            min = 0, 
            max = 100)

Becomes

• Combination of standard R code, and special Shiny R functions that create and manage interactive elements.

h3("I don't have to do a lot of thinking")
radioButtons(inputId = "REI_q1", 
             label = "1 = Disagree, 5 = Agree",
             choices = list(1:5))

Becomes

• Once a participant completes an input, the result is stored as an r object you can then use any way you'd like

# Store result in prob.est
sliderInput(inputId = "prob.est", 
            label = "What is the probability of X?", 
            min = 0, 
            max = 100)

# Use prob.est in other code

true.value <- 33
error <- abs(input$prob.est - true.value)

h3("Your estimate was off by ", error)

• You don't program explicit pages, rather, you update what's in the display based on some criteria:

# Definition of welcome page
if(CurrentValues$page == "welcome") {

  h3("Welcome to the experiment"),
   textInput(inputId = "workerid", 
                    label = "Please enter a unique ID that no one else would use", 
                    value = "", 
                    placeholder = "e.g.; Cat57Door"),
  # This displays the action putton Next.
  actionButton(inputId = "gt_Instructions", 
               label = "Continue to Instructions")

}

observeEvent(input$gt_Instructions, {CurrentValues$page <- "instructions"})

• You don't program explicit pages, rather, you update what's in the display based on some criteria:

# Definition of instructions page
if(CurrentValues$page == "instructions") {

  h3("Instructions"),
  p("In this experiment, you will....."),
  p("On the next page, you will play a practice game"),
  # This displays the action putton Next.
  actionButton(inputId = "gt_Practice", 
               label = "Continue to Practice")
}

Go to ShinySurvey

• Randomizing stimuli order is no problem.
• You can write 'pseudo-loops' to sequentially display questions or other stimuli automatically and then store sequential responses.

rei.survey <- list(
  "text" = c("I don't like to have to do a lot of thinking",
             "I try to avoid situations that require thinking in depth about something",
             "I prefer to do something that challenges my thinking abilities rather than something that requires little thought",
             "I prefer complex to simple problems",
             "Thinking hard and for a long time about something gives me little satisfaction",
             "I trust my initial feelings about people",
             "I believe in trusting my hunches",
             "My initial impressions of people are almost always right",
             "When it comes to trusting people, I can usually rely on my 'gut feelings'",
             "I can usually feel when a person is right or wrong even if I can't explain how I know"))

• You can write 'pseudo-loops' to sequentially display questions or other stimuli automatically and then store sequential responses.

# P3) REI
if (CurrentValues$page == "rei") {
    list(
      h3(paste(rei.survey$text[current.question])),
      em(paste(rei.survey$description)),
      radioButtons(inputId = "rei_survey", 
                  choices = rei.survey$labels),
      actionButton(inputId = "nextq", label = "Continue"))
    }

observeEvent(input$nextq, {
    CurrentValues$question.num <- CurrentValues$question.num + 1
    CurrentValues$rei <- c(CurrentValues$rei, input$rei_survey)
  }})

Notes

• Data can automatically be uploaded to Dropbox or emailed (I'm pretty sure).
• Group experiments are possible! (Thanks Kevin Trutmann!)

• In order for participants to use your App, it has to be hosted on a Shiny Server.
• You can run up to 5 apps for free at RStudio. However, you are limited to 25 active hours per month
• 3 options for advanced hosting: RStudio Standard ($1,100 / year), Amazon Web Services, or in-house ($4,500 / year).
• If there is enough interest, our IT will set up an in-house server

Pros

• You can integrate all of your data management using R code within the experiment.
  • Great for dynamic experiments where stimuli depend on responses.
• Automatically export data in exactly the format you want to Dropbox.

Cons

• Simple things you get for free in Survey software require extra code.
  • Required responses, checking for invalid responses.
• Crashing is, unfortunately, somewhat common.
  • No more than 20 simultaneous users.
• Response time measures are almost certainly not very precise
  • >500ms is probably ok, <500ms is probably not

• We are currently programming and documenting standard psychology experiments in Shiny and distributing them at http://www.github.com/ndphillips/ShinyPsych.

• Check out the code, play around, let us know what works and what doesn't, and contribute!