Test Scores: What do they mean?

Unless you have a background in statistics, interpreting your child’s language or speech test results can be a challenge. The purpose of this article is to explain what test scores mean in plain English.

To keep things simple, we’ve focused on the bare essentials. If you want to know more about important concepts like standard deviations, confidence levels and z-scores, we’ve included links to some good “statistics 101” videos and other resources at the end.

What’s a norm-referenced, standardised test?

Norm-referenced, standardised tests allow speech-language pathologists to compare your child’s performance on a particular day to that of other children of the same age. “Standardisation” means that the test was itself tested several times on lots of children in exactly the same way (i.e. a standard way), to ensure it works.

Norm-referenced tests allow us to report how a child performed compared to a population of other children the same age who have completed the test. Ideally, this involves the test being given to:

Common norm-referenced, standardised tests include the CELF-4, the new CELF-5, and the PPVT for language, and the DEAP and Goldman-Fristoe for speech. Other tests are used to assess phonological awareness, reading, spelling and writing.

Norm-referenced, standardised tests: what they do (and what they don’t do)

Norm-referenced, standardised tests results tell us whether, on the day tested, your child differed significantly from other children of the same age on the test. Well-designed tests, like those mentioned above, are reliable and valid. This means they measure what they say they measure and produce consistent, stable results. For these reasons, government and non-government organisations often require standardised tests be conducted on children to determine eligibility for funding or other supported assistance.

Norm-referenced, standardised test results don’t tell us how your child functions in the real world or how communication problems may affect his or her quality of life and participation. Test scores can be affected by conditions like ADHD and anxiety disorders, and factors like how tired, shy, ill, or distracted the child was on the testing date. If English isn’t your child’s first language, or if your child has a disability, the test results need to be interpreted with extreme care.

Norm-referenced, standardised tests tell us nothing about your concerns, goals or priorities. For this reason, speech pathologists should never rely solely on test results to diagnose or treat your child. Other specialised tools, language/speech sampling, interviews and observations over time provide us with essential additional information for treatment.

The assessment results table

After your child has been tested, your speech pathologist should provide you with a report. Test scores are reported in the assessment results summary, which usually looks something like this:

Core and Index scores (and what they measure)	Sum of Scaled Scores	Standard Score	Standard Score Confidence Interval (90% Level)	Percentile rank	What the score suggests
Core Language Score (CLS) (overall language development)	16	63	59 to 67	1	Severe disorder.
Receptive Language Index (RLI) (listening and comprehension)	14	66	59 to 73	1	Severe disorder.
Expressive Language Index (ELI) (expression using words and sentences)	13	65	60 to 70	1	Severe disorder.
Language Content Index (LCI) (vocabulary and semantics)	15	70	64 to 76	2	Severe disorder.
Language Structure Index (LSI) (morphology and syntax)	18	67	62 to 72	1	Severe disorder.
Working memory index (WMI)	10	71	62 to 80	3	Moderate delay.

Note: This table summarises a (fictional) child’s test results on the CELF-4. A standard score of 86-115 is within the normal range: the average standard score for each index is 100.

The shaded area represents the normal range.

This child’s assessment results suggest a severe receptive and expressive language disorder requiring urgent treatment.

What is a “confidence interval”?

Children can have an off day on the date of their assessment. So can speech pathologists. We all make mistakes. A good speech pathology report will include not just the scaled scores, but confidence intervals, too. A 90% confidence interval (like the one quoted in the table above) gives you the range of scores that you can be 90% sure contains the child’s “true” score. (That of course means there is a 10% chance, the true range is not within the range.)

Sub-test results

Behind the index results, it’s useful to look for patterns of strengths and weaknesses to help identify therapy priorities for children with communication disorders. Sub-test results provide more information about your child’s performance, and are often presented in a table like this:

Subtests	Scaled Score*	Scaled Score Confidence interval (90% Level)	Percentile Rank	Comment
Concepts & Following Directions	3	2 to 4	1	Below normal range
Word Structure	5	3 to 7	5	Below normal range
Recalling Sentences	3	2 to 4	1	Below normal range
Formulated Sentences	5	4 to 6	5	Below normal range
Word Classes 1 – Receptive	6	4 to 8	9	Below normal range
Word Classes 1 – Expressive	6	4 to 8	9	Below normal range
Sentence Structure	5	3 to 7	5	Below normal range
Expressive Vocabulary	6	4 to 8	9	Below normal range
Number repetition – forwards	4	2 to 6	2	Below normal range
Number repetition – backwards	6	4 to 8	9	Below normal range
Number repetition – total	3	1 to 5	1	Below normal range
Familiar sequences 1	7	5 to 9	16	Borderline

Note: This table summarises our fictional client’s results on each of the relevant subtests of the CELF-4. A scaled score of between 7 and 13 is within the normal range.

The shaded area represents the normal range.

This child scored below normal limits on almost all of the sub-tests, although she had areas of relative strength and weakness.

Your speech pathologist will give you information about what each of the subtests assesses. (This information is often presented in an Appendix to the assessment report.)

Scores and the normal curve

For most common norm-referenced, standardised tests, the number of people tested is so large that the scores of the people taking it form a bell-shaped or “normal” curve when plotted on a graph. This fact allows us to measure your child’s performance against children of the same age by taking your child’s raw scores and translating them into standard or scaled scores and percentiles.

A normal curve looks like this:

What types of scores are usually reported and what do they mean?

Sometimes (we don’t know why), speech pathologists report raw scores. These are simply the number of items your child answered correctly on the test. They don’t mean anything.

To report something useful, speech pathologists convert your child’s raw scores into standard scores and percentiles. To get a standard score, we use a scale, usually in a table buried at the back of the test manual. The scale sets the average score (or mean) for the test at a round number. For example, in the table above, the Receptive Language Index is based on a scale where the average is 100. If a child achieves a standard score of less than 100, then the student is said to have performed below the average. If a child scores above 100, the student is said to have performed above the average (see the bottom line of the normal curve above.)

Using a similar process, speech pathologists convert your child’s standard scores into percentiles. Percentiles tell you the percentage of scores that were lower than your child’s score for children of the same age. For example, if your child obtains a percentile rank of 70, 70 per cent of children the same age tested scored below your child’s score (see the second bottom line of the normal curve above).

So which standard scores and percentiles are within “normal limits”?

Using a scale where the average standard score is 100 (as in the normal curve above), a standard score of anywhere between 86-115 is considered “within normal limits”. Scores within these ranges are considered “normal”. As you can see:

When do standard scores suggest an “impairment”?

Confusingly, different tests use different terms to describe levels or degrees of language or speech problems. As a rule of thumb, on a scale where 100 is the average (like the CELF-4):

• a standard score of 70 or below suggests a severe impairment warranting urgent treatment;
• a standard score of 70-77 suggests a moderate impairment;
• a standard score of 78-85 suggests a mild impairment; and
• a standard score of 86 or more suggests no impairment: treatment is not needed and should not be provided.

On the normal curve diagram (above), you can see the percentile range equivalent for each of these standard score ranges.

It’s your speech pathologist’s job to make sure you understand your child’s test results. If you don’t understand the results – or anything else in the report – just ask!

For some useful videos and information on standardised tests, normal curves and basic statistics, please check out the links below:

• http://www.linguisystems.com/pdf/testingguide.pdf
• http://www.nasponline.org/communications/spawareness/testscores.pdf
• http://pareonline.net/getvn.asp?v=1&n=1
• https://www.khanacademy.org/math/probability/statistics-inferential/normal_distribution/v/introduction-to-the-normal-distribution

 

Statistics 101: A Tour of the Normal Distribution