Why the mean score sits on the pass mark

Search for the Texas insurance exam pass mark and you will be told 70%. Ask where TDI said so and the trail goes cold, because the Department did not say it. What TDI publishes is this.

Why a scaled score is not a percentage

Pearson does not report how many of the 130 questions you answered correctly. It reports a scaled score, which is what you get after equating: the process that places raw scores from different examination forms onto one common reporting scale, so a candidate who happened to sit a slightly harder form is not punished for it.

The practical consequence is that no number of questions is the pass mark. A scaled 70 is a point on a scale, and the number of questions it corresponds to moves from form to form. Anyone telling you the exact number of correct answers you need is telling you something Pearson does not fix in advance either.

The arithmetic, step by step

Three published figures, and they constrain each other tightly.

Step one: a cut at 70 fits the mean

The mean scaled score is 69.75. If the cut were 70, the average candidate would fall a quarter of a point short of it. On a symmetric distribution that would put about half of candidates above the line. TDI reports 57.7%, which is nearly eight points more than half.

That gap is not a problem for the theory. It is a description of the population. If the median candidate sits above the mean, the distribution has a long left tail, which is exactly what you get when a chunk of people book a $49 appointment with no preparation, score badly, and drag the average down without being anywhere near the cut. Texas requires no pre-licensing course, so that tail has an obvious source.

Step two: a cut at 60 does not fit anything

Now try 60. That sits just under ten points below the mean, which is roughly two thirds of a standard deviation on a spread of 14.51. On any broadly bell-shaped distribution, two thirds of a standard deviation below the mean leaves about three quarters of candidates above the line.

TDI publishes 57.7%. Three quarters is not 57.7%, and no amount of skew closes a gap that wide in that direction. A cut at 60 is ruled out by the Department's own numbers.

Step three: the round number in the neighborhood

So the cut is somewhere in the high sixties. Within that neighborhood, 70 is where a reporting scale conventionally puts a cut, and a published mean of 69.75 landing that close to it is a hint rather than a coincidence. Scaled score reporting is usually built so the cut sits on a round number.

That is our conclusion. A cut near a scaled 70, consistent with both published figures, derived and labeled as derived on every page of this site that uses it.

Where this could be wrong

We would not defend 70 to the decimal point, and here is the honest limit of the method. The stronger the left tail, the further below 70 the true cut could sit while still producing 57.7%. A cut a couple of points lower is compatible with everything TDI published. We can rule out 60. We cannot rule out the high sixties generally, and if the Department published the cut tomorrow and it were not exactly 70 we would not be surprised.

The second limit is the normality assumption. Real scaled score distributions on licensing examinations are rarely perfectly bell-shaped, and TDI publishes a mean and a standard deviation rather than a distribution. We are reasoning from two moments, not from the shape.

What to actually do with this

Very little, and that is the point. You cannot aim at a scaled score, because you never see one until the screen shows it at the test center. What you can aim at is a margin.

Our readiness target against our own mock papers is 78%, and it is deliberately set above where the evidence puts the cut so that a candidate who hits it clears a harder form with room to spare. That number is ours, it applies to our papers rather than to Pearson's, and it is a preparation target rather than a pass mark. Read signs you are ready to sit for what else has to be true besides the score.