The 15 questions that do not count

You answer 145 questions. 130 of them count. Nobody tells you which fifteen do not, and no amount of staring will identify one.

The arithmetic

The pretest items are distributed roughly in proportion to the scored counts, so the general portion carries twice as many of them as the Texas portion carries in total. They sit among the scored questions and look exactly like them.

What they are for

Every new question has to be tried on real candidates before it can be trusted to count. A question that looks perfectly fair to the person who wrote it can turn out to be ambiguous, or to have two defensible answers, or to be answered correctly just as often by the weakest candidates as by the strongest. You cannot tell any of that from reading it.

So new items are seeded into live forms, unscored, and their performance is analyzed. The ones that behave well graduate into the scored pool. The ones that do not are rewritten or dropped. That is how a question bank stays alive without anybody's license depending on an untested item.

You cannot spot them, and trying is expensive

There is no marker, no visual difference and no position on the paper where they cluster. A pretest item is a normal-looking multiple-choice question about types of policy or Texas statute, written to the same content outline as the rest.

Candidates reason their way into thinking they can identify one anyway. The question was strangely worded. It covered something obscure. It did not look like anything in the outline. Every one of those is at least as likely to mean you are unprepared for that topic as it is to mean the question does not count.

And the search costs clock you cannot spare. At 1.15 minutes a question, which is our arithmetic from the published figures, thirty seconds spent wondering whether an item counts is roughly half the time you had for it.

What it changes about how you sit the exam

Almost nothing, and that is the useful conclusion. Answer all 145 as though all 145 count, because from where you sit they do.

The one thing it does change is how you should read a bad patch. If you hit a run of questions that feel harder than everything around them, part of that run may be items being trialed, and some pretest items are difficult precisely because their difficulty is what is being measured. A stretch of hard questions is weaker evidence about your performance than it feels like at the time.

It also changes the pacing arithmetic

This is where most people get it wrong, including most study sites. They divide 150 minutes by 130 questions, because 130 is the number everyone quotes, and get a comfortable-looking figure.

The clock does not care that fifteen of them are unscored. You still read them, still think about them and still click an answer, and the time comes out of the same 150 minutes. Practicing against the wrong figure is how a well-prepared candidate ends up rushing the last stretch.

Our opinion on the design

Fifteen out of 145 is a reasonable rate and the Texas exam is more transparent about it than most. Pearson publishes the count, publishes the split by portion, and states plainly that candidates are not told which items are which.

The concession, and it is a real one: the candidate carries the cost. You spend the same effort on those fifteen questions, under the same clock, and the benefit accrues to future forms rather than to you. That is a defensible arrangement and it is still worth naming rather than presenting the whole thing as a favor.

The one thing to take away

Practice at 145 questions in 150 minutes, not 130. Everything else about pretest items is interesting and changes nothing you should do.