Checked against primary sources 2026-08-24

The wrong options are not random. They are your likely mistakes, written down.

A distractor is an error somebody expected you to make. Learn which errors get written down and the option list starts telling you things. Learn the popular shortcuts instead and one item in twenty goes the wrong way.

What the wrong options are for

An item is a stem, one keyed answer and a set of wrong options, and the wrong ones are not padding. The standard public guidance on writing them is blunt about their job.

The correct answer should always be the most correct of the answer options, but the distractors should be plausible enough to entice test-takers who do not know the correct answer.

NBME Item-Writing Guide, chapter 5

The companion rule is about the shape of the set rather than any one option in it: all options should be homogeneous and plausible, so that nothing about the way they are written points at the answer.

Two consequences follow, and the second one is where most exam advice goes wrong.

First, the options are diagnostic. If you work an item and your figure is sitting there in the list, that is weak evidence rather than strong, because a competent writer put your likely mistakes in the list on purpose and one of them may be the mistake you just made.

Second, an option that looks like nothing you would have produced is not filler waiting to be struck out. On a well-built item there is no filler. Our own bank is full of conceptual items where the keyed answer is the option that is not a number anybody computed, and a candidate eliminating on strangeness removes it first.

One thing nobody outside PSI can tell you: how a Texas item is built, how many options it carries, or where its distractors come from. None of that is published. What follows is the general construction, plus what is countable in a bank we wrote ourselves and can show you.

Eight families, from a bank where every wrong option carries a reason

Our practice bank runs to 178 items, four options each, and every one of the 534 wrong options has a recorded reason explaining what error produces it. Sorting those reasons gives eight families that keep reappearing, and most of them are specific to a code exam rather than to multiple choice in general.

• Stopped early. The chain has two steps and the option is the value after one of them. Ambient correction applied without the conductor-count adjustment, or a derated figure that was never taken back to the termination rating to see which of the two governs.
• Wrong operation on the right numbers. Two factors added instead of multiplied. Two limits averaged, which no rule permits. A one-way length where the calculation wants the round trip, which lands exactly at half.
• The right answer to the next question along. The branch-circuit figure where the feeder and service figure was asked. The nameplate current where the code names the table value. The primary where the stem said secondary.
• The superseded answer. Correct in the previous edition and printed as though nothing had moved. Load calculations answered as Article 220. A box fill count using the older allowance for grounding conductors.
• The scope answer. A real rule, correctly stated, that does not reach this case. It applies to dwellings and the stem is commercial. It reaches services and the stem is a feeder. It covers damp locations and the stem is the inside of an outdoor raceway, which is a wet one. It is written for a separate structure and the stem is a subpanel down the hall.
• The habit answer. A multiplier or a step you have applied a hundred times in a place the code does not ask for it. The 125 percent continuous multiplier belongs to conductor and device sizing, and the load calculation article rules itself out of that habit at 120.5(E), which makes the option applying it anyway the most attractive wrong answer on the paper.
• The intuition answer. Physically plausible and flatly wrong, which is usually why the rule exists. An option saying fault current returns through the earth is the standing example, and 250.4(A)(5) is where the code settles it.
• The over-read answer. The rule read harder than it is written, so a permission narrows into a ban. An item in our bank asks whether a lighting outlet can join a 20-ampere circuit already serving two bathrooms. One wrong option reaches the same verdict as the keyed answer, by reading the allowance more strictly than it is written, and a candidate who takes it is right about the answer and wrong about the rule.

Those are families in our bank, written against the published subject outline. They are not a description of PSI's item pool, and nobody who tells you otherwise has seen it either.

Three folk rules, tested against 178 items

The advice that circulates in this trade is mostly about the surface of the option list. It is testable, so here it is tested.

Strike any option containing always, never, all, none, every or only

In our bank those words appear in 82 of the 534 wrong options, 15.4 percent, and in 9 of the 178 keyed answers, 5.1 percent. So the lean is real. An absolute is about three times more likely in a wrong option than in the right one.

The cost is where it stops being a rule. Applied as elimination it removes the keyed answer on roughly one item in twenty, and seventy percent of twenty four scored items leaves a journeyman about seven wrong answers on the whole calculations portion. One in twenty against a budget of seven is not a tiebreaker, it is a hole.

The code writes absolutes because some rules are absolute. Among the keyed answers in our bank carrying one of those words: that every conductor in a shared raceway is insulated for the highest voltage present, and that a multiwire branch circuit gets a means of disconnecting every ungrounded conductor together. Both read like traps and both are the answer.

Widen the word list to include a bare "no" and the keyed figure roughly doubles, to 19 of 178, because a yes-or-no item makes "No" the answer about as often as not. The rule is that sensitive to how you write it down.

Pick the longest option

The keyed answer is strictly the longest option on 49 of our 178 items and tied for longest on 23 more. On 106 items it is neither. Four options, right about a quarter of the time: that is what picking at random does.

There is a cue in the aggregate, and it is useless per item. Our keyed options average 43.8 characters against 36.3 for the wrong ones, because a right answer often carries the condition that makes it right. Item writers are told to review an option set for exactly that and even it out.

There is always an all of the above

Zero of 178 in our bank. The standard guidance treats none-of-the-above as a flaw and suggests replacing it with an option that is specific, and our bank carries neither that form nor its twin. A rule that waits for one of them to appear does nothing at all on a paper where it never does.

The pattern behind all three is worth more than any of them. Every folk rule is an exploit of an item flaw, and the published guidance for writing items exists to remove flaws. It names the candidate who answers from test-taking skill rather than from the content, and puts the writer's aim as removing the cues that reward exactly that. So the value of a surface rule decays as an item pool gets better, and from your chair you have no way of telling how good this one is.

The check that does work, and what it costs you

When you have computed a figure and it appears among the options, spend a few seconds asking which specific error produces each of the others.

If you can account for every option you did not choose, your confidence should go up, because you have found the item's design and you are not standing in it. If two of them look like they came from nowhere, stop and re-read the stem, because the likeliest explanation is that you answered a question it did not ask and the two strange options are answers to the one it did.

Five seconds an item is affordable and worth pricing anyway. Across 59 knowledge items it is about five minutes out of 130. Across 26 calculations items it is a little over two out of 110. Spend it on the items where you produced a number, not on the ones you read straight out of the book.

The hardest option to strike is a true sentence

Scope items are the ones that separate candidates who know the material from candidates who know the material and read carefully. Every option is a correct statement of some rule. Only one of them is a correct statement of a rule that reaches the case in the stem.

The tell is in the stem rather than in the options. Before you compare them, name what kind of installation you are in: dwelling or not, service or feeder or branch circuit, indoors or out, above or below whatever threshold the article uses. An option that ignores the qualifier you just named is answering a neighboring question.

That habit costs one reading of the stem and it is the cheapest accuracy available on the knowledge portion, where a scope error looks exactly like a knowledge gap on a score report and is not one.

One family is unusually loaded this autumn

The superseded answer is the family to watch between now and the end of the year, because Texas moves the examinations to the 2026 edition on September 1, 2026 and an option that was keyed in the previous edition is the most plausible wrong option there is. Nobody has to invent it.

Load calculations are the clean case. They sat in Article 220 in every edition before this one and they are Article 120 now, so an option naming 220 is a right answer to the same question asked last year. Wiring devices came out of Article 404 in the same edition, which does the same thing to a wall switch item.

If you learned this material from anything published before September 2026, you learned some of these answers as correct. That is not a knowledge gap either, and it responds to a different fix: check the numbers you are most confident about, because confidence is what stops you looking them up.

What this does not do

It does not answer an item you do not understand. A well-built option set is built so that test-taking skill on its own does not get there, and the people who write the guidance say that is the point of the exercise.

What it does is convert material you already know into more marks than it would otherwise produce, and catch your own slips in the seconds before you commit to one. On a portion that absorbs about seven wrong answers, catching two of your own is most of the difference.

What this page cites

• NBME Item-Writing Guide, Constructing Written Test Questions Read 24 August 2026. Source for the plausibility and homogeneity rules quoted above, for absolute terms and a length-cued correct option being treated as technical item flaws, and for its account of the candidate who answers from test-taking skill rather than from the content. Published by the National Board of Medical Examiners, which writes items for a different profession and to the same standard. source
• PSI candidate information bulletin, TDLR electrical examinations Dated 9 July 2026, read 24 August 2026. Source for the item counts and time limits behind the five seconds an item arithmetic, and for the fact that nothing about item construction or option counts is published. source
• NEC 120.5(E) Percent Multiplier, at the front of the 2026 load calculation article. The subsection behind the habit answer family. Transcribed in this repository from NFPA free access on 23 August 2026. source
• NEC 250.4(A)(5) Performance requirements for the effective ground-fault current path on a grounded system. The subsection behind the intuition answer family. source
• 16 TAC 73.100 Texas adoption of the 2026 National Electrical Code effective 1 September 2026, which is what loads the superseded answer family this autumn. source