Help

Question guide

What each question format tests, how it is scored, and one tip. Scores are whole numbers from 0 to the question’s maximum.

Scoring models

0/1
Each scorable part (an option, a row, a blank, a slot) earns 1 if correct, else 0. The question’s score is the sum.
+/-
+1 for each correct selection, -1 for each incorrect selection, and the score never goes below 0. The most it can earn is the number of correct options. Grouped and table formats apply this to each row, then add the rows.
Rationale, dyad
One sentence, two blanks: 1 point only if both are correct.
Rationale, triad
One sentence, three blanks sharing an anchor, usually the condition. 2 points max: the anchor must be correct, and each of the two supporting blanks then earns 1. Anchor wrong: 0.

Question formats

Multiple Choice

What it tests
Choosing the one best answer from 4 to 6 options.
How it is scored
0/1, max 1: the correct option earns 1 point, anything else earns 0.
Tip
An assignment can shuffle the options. An option that points at others or at a position, such as “All of the above”, stays where you put it.

Extended Multiple Response

What it tests
Recognizing every finding or action that applies. Select all that apply (SATA) has 5 to 10 options with at least one correct; Select N asks for exactly N.
How it is scored
+/-. Max = the number of correct options.
Tip
Write a rationale for each option. The editor warns when every option is correct, or when an option has no rationale of its own.

Multiple Response Grouping

What it tests
Sorting findings by group: a table where each row (a body system, say) has 2 to 4 option cells, and the student selects all that apply in each row.
How it is scored
+/-, applied to each row and then summed. Max = the total number of correct cells.
Tip
Keep each row to one idea, so a wrong choice costs a point in that row and nowhere else.

Matrix Multiple Choice

What it tests
Making one call per row against a shared scale of 2 or 3 columns, such as Indicated / Contraindicated / Non-essential.
How it is scored
0/1 per row: each row earns 1 point when its one selection is correct. Max = the number of rows. An unanswered row earns 0.
Tip
Use 2 to 8 rows. Rows can be shuffled in an assignment; the columns never move, because they are a scale.

Matrix Multiple Response

What it tests
Making several calls per row: rows by columns, with any number of selections in each row (at least one per row is typical).
How it is scored
+/-, applied to each row and then summed.
Tip
Use it when a finding can belong to more than one column, such as a sign seen in two conditions. If each row has exactly one answer, use Matrix Multiple Choice instead.

Highlight Text

What it tests
Recognizing cues in a passage. You mark the phrases a student can select; the rest of the text cannot be selected.
How it is scored
+/-. Max = the number of correct phrases.
Tip
Give it at least 2 selectable phrases with at least 1 correct. The editor warns when more than 60% of them are correct.

Highlight Table

What it tests
Recognizing cues in a table, such as Assessment / Findings: the selectable phrases sit inside the cells.
How it is scored
+/-, over the whole question, or per row when “Score each row separately” is on.
Tip
The table is never shuffled, because where a finding sits is part of the answer.

Drag-and-Drop Cloze

What it tests
Completing prose from a word bank of 4 to 8 words dragged into 1 to 3 blanks. On a phone, tap a word and then tap a blank.
How it is scored
0/1 per blank: each blank earns 1 point when its word is correct.
Tip
Each word can be used once, unless you turn on “Let a word fill more than one blank”.

Drag-and-Drop Rationale

What it tests
The drop-down rationale sentence, answered from a word bank: a dyad (2 blanks) or a triad (3 blanks).
How it is scored
Rationale scoring, as for Drop-Down Rationale. Dyad: 1 point only if both are correct. Triad: 2 points max, and 0 if the anchor is wrong.
Tip
Put plausible wrong words in the word bank for both the condition and the evidence, so neither blank can be solved alone.

Ordered Response

What it tests
Putting 4 to 6 steps in order, by dragging or with up and down buttons.
How it is scored
0/1 for the whole question: 1 point only for the exact order. With “Give a point for each step in the right place” on, each step in its correct position earns 1 instead. It is off by default.
Tip
Type the steps in their correct order. Students always start from a scrambled order that is never the answer.

Bowtie

What it tests
Pulling a whole scenario together: pick the 1 potential condition from 4, 2 actions to take from 5, and 2 parameters to monitor from 5.
How it is scored
0/1 per slot, 5 slots, max 5. The order within a pair does not matter.
Tip
Make the wrong actions and parameters fit one of the wrong conditions, so the student has to commit to the condition first.
A Bowtie question: empty slots for two actions to take, one potential condition and two parameters to monitor, with the choices for each listed below them.
Five slots: two actions, one condition, two parameters.

Case studies and Trend questions

Case study

Six questions, one for each clinical judgment step, in the fixed order below. The patient record sits beside every question: a pane on a laptop, a sheet on a phone.

Build one in a quiz with Create case study. It needs exactly 6 questions, steps 1 to 6 in order, and a patient record with at least one tab.

How it is scored: Each step’s question is scored by its own format’s rule.

Step 1 of 6 of a case study, Recognize Cues: the patient record on the left and a Highlight Text question on the right.
Step 1 of a case study, with the patient record beside the question.

Trend question

Not a separate format: any question outside a case study becomes a Trend question when its patient record is charted at two or more times, which adds a time selector.

Use it to ask what changed, for example whether each finding improved, stayed the same or declined over a shift.

How it is scored: Scored by the rule of the format it uses.

A Trend question: a patient record with a time selector for 0800, 1200 and 1600, beside a matrix asking whether each finding improved, is unchanged or declined.
The time selector above the patient record makes this a Trend question.

The six clinical judgment steps

From the NCSBN Clinical Judgment Measurement Model. A case study has one question for each step, in this order. Tag any other question with the step it tests.

1. Recognize Cues

What it asks
Which information in the patient record matters, and which is noise?
Typical formats
Highlight Text or Table, select all that apply, Matrix Multiple Response.

2. Analyze Cues

What it asks
What could these cues mean? Link the findings to the conditions they point to.
Typical formats
Drop-Down Cloze or Rationale, Matrix Multiple Choice, Multiple Response Grouping.

3. Prioritize Hypotheses

What it asks
Which explanation is most likely, and which is most urgent?
Typical formats
Drop-Down Rationale, Drag-and-Drop Rationale, Select N.

4. Generate Solutions

What it asks
What outcomes are wanted, and which interventions could reach them?
Typical formats
Select all that apply, Drop-Down Table, Matrix Multiple Choice or Response.

5. Take Action

What it asks
Which interventions should the nurse carry out now?
Typical formats
Drag-and-Drop Cloze, Matrix Multiple Choice, Select N.

6. Evaluate Outcomes

What it asks
Did the interventions work? Compare what happened with what was expected.
Typical formats
Matrix Multiple Choice (improved / no change / declined), Highlight, select all that apply.