Plus-Minus Scoring and the Real Cost of One Extra Selection

A wrong selection does not just fail to help

Most partial-credit systems are generous. You get credit for what you got right and nothing for what you missed. NCSBN's plus-minus model does not work that way.

Under plus-minus scoring, correct options earn points and incorrect options subtract. That extra option you added because it seemed reasonable is not neutral. It takes something back.

That is the whole lesson, and it earns a page.

Which item types use it

NCSBN's scoring FAQ names three families under the plus-minus model.

Select-all-that-apply multiple response items are the ones you will meet most. Matrix multiple response items are scored under it too, summed across columns. Multiple response grouping items are scored under it as well, summed across rows.

Everything else in the FAQ sits in a different model. Matrix multiple choice, select N multiple response, drop-down cloze, drop-down table and bowtie are all zero-one items, where an incorrect option earns zero and nothing is taken away.

That difference is the practical one, and it is worked through in why each row of a matrix item scores on its own.

Why NCSBN scores anything this way

A model that subtracts is unusual and it is not arbitrary. If a select-all item gave credit only for correct selections, the winning strategy would be to select every option on the screen, and the item would stop measuring anything at all.

Subtraction removes that strategy. What is left is an item measuring discrimination, which is the ability to tell a plausible option from a correct one, rather than familiarity with the topic in general.

Knowing why the rule exists makes it easier to live with. You are not being punished for effort. You are being asked to be specific.

The round-up-to-zero rule

NCSBN's wording is that all negative total scores are rounded up to zero. That sounds like a safety net. Read carefully, it is thinner than it looks.

It protects the rest of your exam from one disastrous item. It gives you nothing back on the item itself.

So the floor only helps once you have already reached it. Every subtraction above that floor is real, and it comes straight out of points you had earned.

A worked example of the arithmetic

Say an item lists eight options and three of them are correct.

You select three and get all three right. Three points earned, nothing subtracted, full marks for the item.

You select five, and the three correct ones are among them. Three points earned, two subtracted, and the item pays one point. More work, less score.

You select all eight. Three earned, five subtracted, a negative total, and the rule rounds it up to zero. Maximum effort, minimum result.

That pattern is worth staring at. Past a certain point, effort and score run in opposite directions here.

What this does not mean

It does not mean under-select out of fear. An option you can defend is worth a point, and leaving it out costs you that point just as surely as a wrong pick costs you one.

The rule is symmetrical. Missing a correct option and adding an incorrect one cost you exactly the same in this model.

So the target is not fewer selections. The target is accurate ones.

Why row and column independence limits the damage

The one piece of genuinely good news is structural. Matrix multiple response is summed across columns and grouping items are summed across rows, with each column or row scored independently.

That containment matters. An over-selected column cannot reach across and drain a column you handled well, and one bad row does not pull down the rows above it.

So a large plus-minus item is not one enormous bet. It is a set of small ones, and a bad section stays a bad section.

What to do differently on the screen

Have a reason for every option you select. Not a feeling that it fits, a reason you could say out loud in one sentence.

Then apply the isolation test. Read each option as though it were the only one on the screen and ask whether it would be true. Options that look right only beside their neighbors are the ones that cost you.

Do not select an option to cover yourself. On a zero-one item that habit is free and on a plus-minus item it is expensive, and the item does not always announce which one it is.

When you can defend one option and not the other, defend one.

How to practice against it

When you review a missed select-all, do not just look at the correct answer set. Sort your errors into two piles. Options you added that were wrong, and options you left out that were right.

Those two piles need different fixes. An adding problem is usually a reasoning problem, where the cues were collected and never connected. A leaving-out problem is usually a knowledge gap on one specific point.

Count them separately for a week and you will know which one you actually have.

Where the reasoning has to come from

Over-selection is rarely a strategy problem. It is usually a sign that the cues in the stem were collected but never connected, so every option still looks plausible.

That is a specific, fixable skill, and it has its own page in recognizing cues versus analyzing cues.

The third scoring model, where an item pays nothing unless both halves of a pair land, is covered in rationale scoring and the half-right answer.

For where the scoring models fit inside the exam's judgment model, start at how the next generation NCLEX measures clinical judgment.