Standard Score vs. Scaled Score in Speech-Language Evaluations

Your child's speech-language evaluation comes back with a table full of numbers — 85, 7, 16th percentile, 95% CI 80–90 — and every one of them measures something different. The school team reads them as if the meaning were obvious. It isn't.

The problem is that speech-language reports mix two different scoring scales in the same table, and parents who don't know which is which draw the wrong conclusions. A score of 7 sounds terrible on one scale and perfectly average on another. Understanding the distinction between standard scores and scaled scores is the first step toward knowing whether the report actually supports what the school is telling you.

Standard Scores: The Big Picture Composites

Standard scores are the numbers you see next to labels like Core Language Score, Receptive Language Index, or Expressive Language Index. On the CELF-5 and most major speech-language batteries, standard scores use a mean of 100 and a standard deviation of 15.

That means:

• 100 is dead center — the average performance for that age group
• 86–114 is the report's "average range"
• 85 sits at the 16th percentile — 84% of same-age peers scored higher
• 70 sits at the 2nd percentile — two standard deviations below the mean

When a school tells you your child's Core Language Score is 86 and "within normal limits," that falls at the lower edge of the report's average band. A standard score of 86 means your child performed at about the 18th percentile. Whether that's adequate depends on a lot more than the label.

Standard scores are composites that combine results from multiple subtests. A child who scores brilliantly on vocabulary but falls apart on sentence repetition can land at 86 overall — the higher scores pulling up the lower ones until the composite looks unremarkable.

Scaled Scores: The Subtest Details

Scaled scores appear next to individual subtest names — Recalling Sentences, Word Structure, Following Directions, Formulated Sentences. On the CELF-5, scaled scores use a mean of 10 and a standard deviation of 3.

The average range for scaled scores is 8 to 12. A scaled score of 10 is the 50th percentile. A scaled score of 7 is below the average range, at roughly the 16th percentile — one standard deviation below the mean.

Here's what trips parents up: a scaled score of 7 and a standard score of 85 represent the same relative position on the bell curve (both are –1 SD), but they look completely different on paper. A parent scanning the report sees "85" next to the composite and thinks it's passable, then sees "4" next to Recalling Sentences and wonders if that's out of 10 or out of 100.

It's neither. A scaled score of 4 is 2 standard deviations below the mean — the 2nd percentile. That means 98% of same-age children performed better on that subtest. When a score that extreme gets averaged into the composite, the composite barely flinches. That's the trap.

Percentile Ranks: The Number Parents Understand Fastest

Percentile ranks translate both standard and scaled scores into a single question: what percentage of same-age children scored at or below this level?

• 50th percentile = average
• 25th percentile = within the report's average percentile band (18th–84th)
• 16th percentile = about one SD below the mean, just below that band
• 5th percentile = clearly below expectations
• 2nd percentile = two SDs below the mean

Percentiles are non-linear: the same standard-score change can produce a larger percentile change near the mean than at the lower tail. A percentile by itself also does not describe functional impact.

One thing percentiles do well is cut through jargon. When someone tells you your child's Recalling Sentences score is "below the mean but within the broad normative range," asking for the percentile puts a number on it. If it's the 5th percentile, 95% of peers did better. That's harder to dismiss.

Confidence Intervals: What the Score Could Actually Be

Every standardized test has measurement error. A child who takes the same test twice won't get the exact same score — fatigue, attention, the testing room's temperature, and random variation all contribute. The confidence interval acknowledges this.

A 95% confidence interval of 80–90 around a standard score of 85 represents measurement uncertainty around the obtained score. That range matters when the score sits near an eligibility cutoff. If the district uses a cutoff near 85 and the report says 85 with a CI of 80–90, ask how the team considered that uncertainty alongside the rest of the evaluation and educational impact.

Putting It Together: Reading the Table

When you look at the score table in a speech-language evaluation, here's the translation:

The most important step is comparing the composite standard score to the individual scaled scores underneath it. If the composite is 88 but two subtests are at scaled scores 4 and 5, those low subtest scores show specific weaknesses that the composite does not show; the child's profile is uneven.

What to Do With This Information

If you're heading into an IEP or eligibility meeting, don't let the team stop at the composite. Ask for the subtest-level scaled scores and their percentile equivalents. Ask what the confidence interval is on any score near a cutoff. Ask whether the composite average is masking a pattern of extreme highs and lows.

The Speech-Language Evaluation Decoder walks through every score type on the CELF-5 and related tests, translates common report sentences into plain language, and provides specific questions to ask about each subtest. It covers the discrepancy between composites and subtests in detail — including what to say when the school insists the composite is all that matters.

A standard score and a scaled score sit on different scales, measure different things, and tell different parts of the story. Reading them together — and knowing when the team is reading them selectively — is the difference between accepting a "Does Not Qualify" letter and asking the questions that change the outcome.