Assessment Report Glossary: Percentiles, Standard Scores and Every Other Term, Decoded

The report arrives as a PDF, fifteen or twenty pages long, and you open it the moment it lands. By page four you're swimming: your daughter is "at the 37th percentile" on something called a Fluid Reasoning Index, her standard score is 95 with a "95% confidence interval of 89-101", and a footnote mentions that a "statistically significant discrepancy" between two index scores "had a base rate of 14.2%". You read the sentence three times. You love this child. You have no idea whether to be worried.

You're not alone, and it isn't your fault. Psychoeducational reports are written in the technical dialect psychologists use with each other, and even good clinicians sometimes forget to translate. The good news: the entire vocabulary rests on a handful of ideas, and once you hold those, the rest of the report opens up.

This is the glossary to keep beside the report. It's organised so you can read it straight through, or jump to the term that's blocking you.

The Big Idea Behind Every Number: Comparison to Same-Age Peers

Nothing in a psychoeducational report is a mark out of 100, and nothing is graded against a pass line. Every score answers one question: how did this child perform compared with other children the same age?

To make that comparison, every major test is normed, given to a large, carefully balanced sample of children (the normative sample or norm group) before publication. Your child's raw performance is converted into scores that locate them within that sample. Three terms cover the machinery:

• Raw score, the simple count of what your child did: words read, puzzles solved, digits recalled. Raw scores are meaningless on their own (twelve correct could be brilliant at six and worrying at twelve), which is why reports barely mention them.
• Norms / norm-referenced, scored by comparison to the normative sample of same-age children. Almost everything in a psychologist's report is norm-referenced.
• Age-equivalent score, a translation like "reading age 7 years 2 months". These feel intuitive but are statistically the flimsiest numbers in any report, and most psychologists now avoid them. If your report includes them, treat them as rough colour, never as the headline.

Standard Scores, Scaled Scores and the Bell Curve

Ability in any large group of children falls into a bell curve (a normal distribution): most children cluster near the middle, with fewer and fewer at each extreme. Standard scores are simply addresses on that curve.

• Standard score, the workhorse of every report. The scale is set so the mean (average) is 100 and the standard deviation, the ruler for measuring distance from average, is 15. So 100 is dead centre, 85 is one standard deviation below, 115 one above. About two-thirds of all children score between 85 and 115.
• Average range, on tests like the WISC-V, standard scores of 90-109 are labelled Average. The descriptive labels run from Extremely Low (69 and below) through Very Low, Low Average, Average, High Average, Very High, to Extremely High (130 and above). The exact label names vary slightly between tests; the bands are near-identical.
• Scaled score, the same idea on a smaller ruler, used for individual subtests: mean of 10, typical range 1-19, with 8-12 broadly average. If you see a small number like 7 next to "Digit Span", it's a scaled score, not a catastrophe, 7 is low average.
• T-score, a third ruler, common on behaviour rating scales (questionnaires you or the teacher completed): mean of 50, standard deviation of 10. On behaviour measures, higher T-scores usually mean more reported problems, often 65+ flags concern, which trips up parents used to "higher is better".

The single most useful habit: whenever you meet a number, ask which ruler it's on. 95 is average on the standard-score ruler and extremely high on a T-score ruler.

Percentiles: The Most Misread Number in Any Report

A percentile rank tells you the percentage of same-age children your child scored as well as or better than. The 75th percentile means they outperformed about 75 in 100 peers. It is not the percentage of questions answered correctly, a child at the 50th percentile got an exactly average result, not "half marks".

Two things about percentiles catch parents out:

• The 50th percentile is good. It's the precise middle of typical. Years of school-test thinking make 50 sound like scraping a pass; here it means "as expected for age".
• Percentiles bunch in the middle. Because most children score near average, a few standard-score points around 100 move the percentile a long way (standard scores 100 to 105 jumps you from the 50th to the 63rd percentile), while the same gap out at the tails barely moves it. Never measure progress or differences in "percentile points".

For orientation: standard score 85 ≈ 16th percentile, 90 ≈ 25th, 100 = 50th, 110 ≈ 75th, 115 ≈ 84th.

Composites, Indexes and the Architecture of the Report

Cognitive reports are built in layers, and the vocabulary marks the floors.

• Subtest, one specific task (repeating digits, building block designs, defining words). The smallest unit, and the least reliable on its own.
• Index score / composite score, several related subtests combined into one standard score measuring a broad ability. Composites are more reliable than any single subtest, which is why interpretation should rest on them. On the WISC-V, the most widely used children's cognitive test, there are five primary indexes: Verbal Comprehension, Visual Spatial, Fluid Reasoning, Working Memory and Processing Speed. Our plain-English WISC-V guide walks through what each one actually measures.
• FSIQ (Full Scale IQ), the global composite summarising overall performance. Useful shorthand when a child's abilities are even; actively misleading shorthand when they're not. Which brings us to:
• Discrepancy / statistically significant difference, a gap between two of a child's own scores (say, high Verbal Comprehension but low Processing Speed) big enough that it's unlikely to be measurement noise. "Statistically significant" means real, not necessarily rare,
• Base rate tells you the rare part: the percentage of children in the norm sample who showed a gap that large. A significant discrepancy with a base rate of 25% is real but common; one with a base rate of 2% is genuinely unusual and usually drives the report's conclusions.
• Achievement test, the academic counterpart to the cognitive test (reading, spelling, writing, maths), such as the WIAT or Woodcock-Johnson, scored on the same 100/15 ruler so cognitive and academic results can be compared directly.

Precision and Honesty: Confidence Intervals, Reliability, Validity

• Confidence interval (CI), the range (e.g. "FSIQ 92, 95% CI 87-98") within which the child's true score almost certainly lies. All measurement carries error; the CI quantifies it. Read the range, not just the point score, especially before comparing two scores whose ranges overlap.
• Standard error of measurement (SEM), the statistic the confidence interval is built from. You can safely let this one wash over you.
• Reliability, how consistently a test measures (would the score be similar next Tuesday?).
• Validity, whether the test measures what it claims to. In reports, "this is considered a valid assessment of his current functioning" is the psychologist confirming nothing on the day, illness, refusal, inattention, undermined the results. If validity is questioned, everything else carries an asterisk.
• Qualitative observations, the psychologist's notes on how your child worked: anxious checking, fading attention after an hour, clever compensations. Often the most valuable paragraphs in the report; numbers say what happened, observations say why it might have.

Terms Around the Edges of the Report

• Standardised administration, the test was given exactly by the manual, same wording and timing for every child, so the norms hold.
• Criterion-referenced, scored against a fixed benchmark of skills ("can decode single-syllable words") rather than against peers. Some school screeners and structured screening tools, including GiraffeLens's, report criterion-referenced ranges precisely because they describe skills without claiming an IQ-style ranking.
• Clinical range / borderline range, on behaviour questionnaires, score bands suggesting concern levels. "Clinical" here means "warrants attention", not "diagnosed".
• Cognitive vs behavioural measures, performance tests (what the child did) versus rating scales (what adults reported). Reports juxtapose them deliberately, since each catches what the other misses.
• Recommendations, the closing section translating the numbers into actions for school and home. For day-to-day life, this is the part that matters most; if it's thin or generic, you're entitled to ask the psychologist for specifics.
• Provisional / differential diagnosis, "provisional" flags a likely diagnosis awaiting confirmation; "differential" lists the alternatives considered and ruled out. Formal diagnosis itself is the province of a registered or licensed psychologist (or medical specialist), no report jargon, screening tool, or glossary changes that.

Misreadings That Cause Needless Panic (or False Comfort)

A few patterns come up again and again when parents read reports cold. Worth naming them directly:

• Treating one low subtest as a finding. Single subtests are noisy. A scaled score of 6 amid otherwise average scores might reflect a real weakness, or a lapse of attention, a misheard instruction, or simple bad luck. Psychologists interpret patterns across subtests and composites, not isolated dips; you should too.
• Reading "below average" as "below normal". Low Average (standard scores 80-89) describes roughly one child in six. It's a description of position on the curve, not a disorder. Context decides whether it matters: an 85 in Processing Speed alongside a 120 in Verbal Comprehension tells a very different story from an 85 across the board.
• Comparing scores across different reports without checking the test. A standard score from a WISC-V and one from a school screener aren't automatically comparable, different tests, different norm samples, sometimes different rulers entirely. When tracking change over time, compare like with like, ideally the same instrument.
• Expecting scores to be identical on retest. Scores wobble within their confidence intervals, and some tasks show practice effects, modest gains from having seen similar material before, which is one reason psychologists avoid re-administering the same cognitive test within a year or two.
• Assuming a "normal" report means nothing is wrong. Tests sample abilities under quiet, one-to-one, highly structured conditions, the best possible environment for many children who struggle in noisy classrooms. A clean profile with ongoing real-world difficulty is a finding in itself, and worth raising at the feedback session rather than swallowing in silence.

How to Reread the Report Now

Armed with the vocabulary, try this sequence. First, skim past the subtest tables and read the summary and recommendations, the psychologist's own translation. Second, look at the composite/index scores with their confidence intervals and percentiles, and sketch your child's profile: where are the peaks, where are the dips, and how big are the gaps between them? Third, read the qualitative observations for the story behind the dips. Last, write down every sentence you still can't parse and bring the list to your feedback session, asking a psychologist to explain their report is not an imposition; it's the job.

And keep one idea above all of it: these numbers describe how your child performed on particular tasks, on particular days, compared with age peers. They are a map of strengths and weaknesses drawn to guide support. They are not a verdict, not a ceiling, and not the measure of a child you know in far higher resolution than any bell curve ever will.