CAT4 scores explained: SAS, stanines, percentiles and the profile

The raw score, the number of right answers, is turned into an SAS using GL's national norms. Because the comparison is with children of the same age to the month, a child born in August is not marked down for being almost a year younger than a classmate born in September.

An SAS of 115 is one standard deviation above average, which about one child in six reaches or passes. An SAS of 85 is one standard deviation below. Neither end of the scale says anything about how hard a child works or how well they will do; it describes how they reasoned on those tests, on that day.

The overall "mean CAT4 score" is the average of the four battery scores. GL's technical report notes that this mean spreads a little less widely than a single battery score, with a standard deviation of about 12.

Stanines are deliberately broad. Two children with SAS 104 and 111 share stanine 6, and most teachers find that more useful than reading meaning into a few points either way.

It is not a percentage of questions right. A child can answer 60% of the questions and sit well above the 50th percentile, because CAT4 is built so that few children answer everything. Percentile ranks bunch up in the middle: the gap between the 50th and 60th percentile is only a few SAS points, while the gap between the 90th and 99th is much bigger.

A child with a spatial bias often learns well from diagrams, models and seeing things done; one with a verbal bias from reading, discussion and writing. Most children show no bias or only a mild one. In GL's UK technical report about 11% show a mild verbal bias and about 11% a mild spatial bias; GL's documents give different figures for the moderate and extreme groups, so we do not quote them.

GL's guide on strategies for learning suggests ways teachers can use each profile. A strong spatial profile with lower verbal scores is worth knowing about, because that child's ability can be missed by tests that rely on reading.

Schools that use CAT4 in admissions do not all publish a cut-off, and those that do differ, so we do not give one. If a place depends on the result, ask the school how it is used: on its own, by battery, or alongside interviews and reports.

If your child's score was lower than you expected, it is worth remembering how much one morning can move a short timed test. Tiredness, nerves and a format met for the first time all cost points. A score is a picture of one day, not a verdict on a child.

GL describes CAT4 as one input for teachers, alongside attainment and what they see in class. If you think your child was placed too low, a calm conversation with the school, with examples of their work, is the right step. A child who froze on an unfamiliar format is one reason a result can sit below the child, and teachers know this.

For the next test, the best help is the same as ever: rest, a calm start, and a child who knows the formats. The free CAT4 practice test shows each of them once.

Our practice test reports percent correct for each test and battery and a practice band, from "Getting used to it" to "Very confident". Those bands are for spotting a format that has not clicked or a test where time ran out, not for predicting a number. The CAT4 test guide explains how the real day runs, and the levels page covers which level your child will sit.

GL Assessment says plainly that it advises against any kind of practice before CAT4 and does not endorse any practice materials, ours included, because CAT4 is not a test of learnt knowledge and coaching could make a score less accurate. We take that seriously. Our questions are for meeting the formats once or twice so nothing on the day is a shock, not for drilling a child towards a number.