What a Matrix Multiple Response item looks like
The format is a table. Rows list findings, interventions, lab values, or other items from the case. Columns list categories the candidate must judge each row against. The candidate clicks one or more checkboxes per row.
Three column-set patterns are common on the exam:
- Expected / Unexpected / Unrelated for Analyze Cues items
- Improved / No Change / Worsened for Evaluate Outcomes items
- Indicated / Contraindicated / Non-Essential for Take Action items
| Finding | Expected | Unexpected | Unrelated |
|---|---|---|---|
| Row 1 | |||
| Row 2 | |||
| Row 3 | |||
| Row 4 |
Most Matrix Multiple Response items have between 4 and 8 rows, with 3 columns being the most common column count.
How Matrix Multiple Response is scored
Scoring is 0/1 dichotomous per row. To earn the point for a row, the candidate must correctly check every column that applies and leave unchecked every column that does not apply. Get one column wrong on a row, and that entire row scores zero. Get the rest of the rows right, and they score independently.
This means a 5-row Matrix item is worth up to 5 raw points, with each row scored separately. The difference from Bow-Tie is important: Bow-Tie also uses 0/1 per blank, but the Bow-Tie has exactly 5 blanks in a fixed pattern. Matrix items vary in row count by question.
| Scenario (5-row item) | Raw score | Outcome |
|---|---|---|
| All 5 rows fully correct | 5 / 5 | Maximum credit |
| 4 rows correct, 1 row with one column wrong | 4 / 5 | Partial credit across rows |
| Correct columns checked on every row, plus one extra wrong column on row 3 | 4 / 5 | Row 3 fails for over-selection |
| No rows fully correct | 0 / 5 | No credit |
Worked example
Here is a Matrix Multiple Response item modeled on the Analyze Cues subtype, which is the most common version of this format on the exam.
| Finding | Expected | Unexpected | Unrelated |
|---|---|---|---|
| Crackles in bilateral lower lobes | |||
| Bilateral pedal edema +2 | |||
| Temperature 102.6 F | |||
| Hemoglobin A1C 6.1% | |||
| Weight gain of 8 lbs over 4 days |
Three findings are expected in heart failure exacerbation: crackles, pedal edema, and rapid weight gain. The fever is unexpected and points to a secondary problem like pneumonia or infection. The A1C is unrelated, since it reflects long-term glycemic control and not the current acute presentation. Each row scores independently, so a candidate who gets 4 of these right and one wrong earns 4 out of 5 raw points on the item.
The strategic difference between Matrix and SATA
On paper, Matrix Multiple Response looks like several Select All That Apply items stacked together. The key difference is the scoring rule.
SATA on the NGN uses the +/- partial credit rule, where each correct option checked earns +1 and each wrong option checked earns -1 (floored at zero per item). Matrix Multiple Response uses strict row-level dichotomous scoring, with no partial credit within a row. This makes Matrix items unforgiving on individual rows but generous in that each row is its own scoring unit.
The practical takeaway: on SATA, marginal answers cost you only their own point. On Matrix, a single wrong checkbox on a row kills that entire row.
What students get wrong on Matrix items
- Treating columns as mutually exclusive. Some items allow more than one column per row to be correct. Always read whether the columns are mutually exclusive before assuming you should pick only one.
- Over-selecting on uncertain rows. Checking an extra "just in case" column fails the row.
- Reading the column headers wrong. Expected versus Unexpected can flip depending on the case context. A fever of 102 is Expected in sepsis but Unexpected in heart failure.
- Forgetting that rows score independently. Getting stuck on one tricky row is not worth losing time you would use to nail four easy rows.
Key takeaways
- Matrix Multiple Response uses a grid: rows are findings, columns are categories
- Each row can have one or more correct columns checked
- Scoring is 0/1 dichotomous per row, no partial credit within a row
- Rows score independently, so partial credit across rows is possible
- Most items have 4-8 rows and 3 columns
- One wrong checkbox on a row kills that row's entire point