Open your completed Week 7 layout_app.R and choose Run App.

Trace each input: which outputs change? Why did we use reactive() for the shared filter instead of repeating it inside each output?

Unzip psyc40940-week-08-tabs-controls.zip into your existing module project. Keep scripts beside data/ inside week-08-tabs-controls/.

Open session-guide.html in a web browser. Read it beside your code; edit the named app scripts in RStudio and use Run App to test them.

Complete the blanks for the current checkpoint before running. The guide explains which code remains commented until later.

The connections are unchanged. Keep controls and the count outside the tabs, and preserve the output names.

Predict: if the reader changes group while viewing one tab, the plot in the other tab will changed too.

Quick check (about 5 minutes): predict which rows remain with one group, both groups, or no groups selected.

checkboxGroupInput(
  "groups", "Output groups",
  choices = c("Spoken output" = "speech",
              "Written output" = "writing"),
  selected = c("speech", "writing")
)

choices lists the available values; selected ticks both initially. input$groups holds the current selection, which may contain one, two, or no values.

Use modality %in% input$groups to keep rows matching any selected group. == compares values element by element and was suitable for the earlier single choice.

Here "groups" names the control; modality remains the data column.

Break: we will use these group choices with a range filter after the break.

sliderInput("vocab_range", "Keep vocabulary scores within",
            min = 0.80, max = 1.00,
            value = c(0.80, 1.00), step = 0.01)

Why filter? We might want to examine reaction times among participants with vocabulary scores of 0.90–0.95, rather than everyone. The range selects which participants appear; the axes still show reaction time.

The handles supply a vector: [1] is its lower value; [2] is its upper value.

selected <- filter(
  d_participants,
  modality %in% input$groups,
  vocabulary_score >= input$vocab_range[1],
  vocabulary_score <= input$vocab_range[2]
)

Each condition must hold. Vocabulary range selects participants; bin width groups their reaction times into bins.

Group and vocabulary range select participants; vocabulary score is not mapped to either plot axis. The count and both plots use the same result from selected_data().

Trace it: which input changes the histogram without changing the count or boxplot?

Open controls_app.R and follow Exercise 2A: filter groups and a range (about 12 minutes).

Complete four active blanks: checkbox function, group input name, and two range indexes. Leave the commented validation block for the next checkpoint.

• Test both groups (72), speech (33), and writing (39) with the full range.
• Set the score range to 0.90–0.95 with both groups: 19 participants remain. Check both tabs.
• Change bin width: the count and boxplot should stay fixed.

Then clear both groups or use the range 0.80–0.81. Observe the result and restore a valid selection. We will now improve the feedback.

A reader’s subgroup of interest may contain no participants. What happened, and what could they change?

• Cause: no group is ticked, or no participants match the chosen score range.
• Recovery: select a group or widen the vocabulary-score range.
• Why explain it? A blank plot could look like an app error or unfinished loading.

These checks run inside a reactive expression or a render block when its calculation is needed.

• Without validation: our empty selection produces count 0 and empty plots, with no recovery guidance.
• Without req(): an unavailable input can reach the filter and cause an R error. Our fixed slider normally supplies a default, so this is mainly a startup guard.

Shiny documentation: req() and validate() / need().

The checks are code inside one reactive expression, rather than additional reactive nodes.

• Input missing: req() stops this attempt silently.
• No rows: validation stops; each dependent output displays the recovery message.
• Rows present: return the data and calculate the outputs. Changing inputs can restore a valid selection.

Continue in the same controls_app.R and follow Exercise 2B: explain an empty selection (about 13 minutes).

• Enable the four commented lines from validate(need( to )).
• Complete its two blanks: the filtered data-frame name and your own recovery message.
• Save, stop and rerun. Test no groups, an empty range, then a valid selection.

Tell the reader what happened and what to change. Check the count and each tab: they all depend on the shared preparation.

The guide explains how to uncomment the lines and lists the expected results.

Use your completed controls_app.R to fill these boxes and add arrows (3 minutes). Copy the boxes onto paper or use a drawing tool.

1. Name each input, the shared expression and each output from the server code.
2. Add arrows from dependencies to the calculations that read them.
3. Mark req() and validate(need()) inside the shared expression.

Predict: what updates when the range changes? Does bin width rerun the filter? Tabs are not extra nodes.

• Seven nodes, six arrows: compare their names and directions with your code.
• Range changes: update selected rows, count and both plots, even though vocabulary is not on either plot axis.
• Bin width changes: redraw only the histogram; selected rows stay cached.

Use your graph to explain the empty-selection and recovery tests you completed in Exercise 2B.

If your core app works, use the remaining time for the density comparison. It extends Week 7’s multiple-group selection with a choice of display:

• Recommended: colour or separate panels? Complete optional_density_app.R to compare the same selected groups using two density views.
• Compare pooled and faceted histograms: uncomment the supplied facet_wrap() line. Which view helps you describe the overall distribution, and which helps compare groups? Update the app explanation to match.
• Add a summary tab: use the supplied tableOutput() and renderTable() example in extension_app.R. Check the count and median for each selected group.

The tasks are independent. Summary and Density are alternative third tabs: choose one, or replace it when trying the other.

How do speech and writing reaction-time distributions differ? Open optional_density_app.R and follow Optional: compare groups with colour or panels in the guide. Complete its seven blanks.

radioButtons(
  "density_view", "Compare groups using",
  choices = c("Colour" = "colour",
              "Separate panels" = "facets"),
  selected = "colour"
)

• Select both groups. Compare overlapping coloured curves with separate panels on common scales. Which shows overlap or shape more clearly?
• Groups/range: change which observations are included. Colour/panels: change how those same observations are displayed.
• Predict and test: the display control changes only the density plot. Add its connection to your graph.

Explain your preferred view in the app text. The guide explains the supplied if / else. Density shows shape, not participant counts.

Choose a control for the value your calculation needs. The same input can guide filtering or a plotting choice, as our bin-width slider already does.

These are reference options for later project work; no additional functions are required today.