The problem with reading ages

Let's begin with a familiar scenario of a pupil struggling with reading. 

Amelia is in year 7 but she is finding the leap to secondary school tricky. She has a good group of friends, but she is finding the curriculum harder to access than she did in year 6. She has her reading test to do in the computer room later and she is nervous about the prospect. 

Amelia is one pupil amongst hundreds of thousands who finds the jump to secondary school demanding. Also, she undertakes a familiar trajectory of standardised reading testing early in her time at school. 

Reading age tests are an important and valuable baseline that can identify pupils like Amelia who may be struggling visibly or beneath the radar. For many such pupils, they can become defined by their reading age. 

Amelia is 11 years and six months old in actual chronological terms. But her reading assessment gives her a reading age of 9 years and 2 months. The yawning gap of more than two years draws immediate concerns. 

Amelia may be assigned to a popular intervention in the school to address the issue. As a result, between September and December, she undertakes useful additional support once a week. 

Tested again, Amelia may make a jump to a reading age of 10 years and 2 months. Teachers and colleagues are understandably both happy and relieved by her quick progress.

But are there limits to what we can read into the reading age jump?

Baselines matter, but reading ages may be misread 

Let me be clear: I think standardised reading age assessments are valuable and for some pupils they can offer a crucial baseline assessment that can begin turnaround that could prove all-important for Amelia. We should not rely on guesswork. 

The problem is not the reading assessment per se, but the common misreading of the reading age. 

Reading ages feel so intuitive that they stick out for leaders, teachers and parents. The more useful elements of the assessment can get relegated. The SAS (standard age score or the standardised score) is more useful as it compares where pupils sit in relation to one another across a national sample, with the average score across the whole sample being 100.

If we return to Amelia, her SAS score is 78, which is in the 'below average' range for pupils across a large national sample (around the bottom 7 or 8% nationally). This may be less catchy than the reading age, but it is less likely to be misread. .

If Amelia was tested aged 8, with the same SAS score, her reading age gap would be something like 15 months. But then when she is aged 11, an equivalent gap of 28 months would be understandably judged as her reading is getting worse. For reasons such as this, reading ages can be easily misinterpreted by busy teachers and leaders.

What to look for with standardised reading assessments

Let's return to Amelia and the positive jump in her reading age by December. Was it the intervention that made the difference? Was it great teaching? Or was it something else?

Look again at her SAS. In September it was 78; in December it is 82. The headline story is a whole year of reading progress in a single term. The SAS tells a quieter story: four points, which may well sit within the test’s confidence band. Amelia may be making steady progress, but the numbers alone can’t tell us that she has leapt forward.

In the September assessment, Amelia may have been uncomfortable in the computer room and misread a couple of questions. At the bottom end of the scale for the test, a couple of answers can make a big difference - a few marks can be tantamount to a whole year in reading age terms! Doing better in December is simply more likely for Amelia because it went so badly in September. Put simply, pupils chosen for intervention because they scored lowest are, as a group, likely to score higher next time regardless (what statisticians describe as 'regression to the mean').

When it comes to the December (re)assessment, Amelia may now be more comfortable in secondary school, and more knowledgeable about the test format. She gets more answers right on the day. This 'practice effect' is a plausible reason for improvement when tests become more familiar to pupils.

No test is perfect. Doing a given test on Monday and then a different version of reading test on Tuesday, with different questions, could see a measurement showing months of difference and it be expected. For a school, 6 months could be a success story or a major worry, but this could actually be just as expected 'measurement error' of 4 or 5 for their SAS that is baked into the reading tests.

Finally, different tests test different components of reading in slightly different ways. One standardised test may focus more on phonics and word reading, another sentence level completion. Reading is a tricky and multifaceted thing. If Amelia does one assessment in year 7 and then a different reading test in year 8, from a different supplier, it is not comparing the exact same thing like we might assume.

Five things to do with reading test data

Schools should use high-quality standardised assessments to help understand how pupils like Amelia read compared to their peers. But we need to get beyond the powerful simplicity of the reading age and use the data carefully. We can consider the following actions: 

1. Focus on the SAS score over the reading age and be careful about over-reading that too! The SAS score is more useful as a comparison and less likely to be misinterpreted. If you need to share the reading age - include the SAS too, and explain how they work together. Remember: a difference of 5 SAS points may be measurement error (so pay attention to the ‘confidence band’ for the given reading test). 
2. Use more than one measure of reading progress. Can a sample of pupils be checked against teacher judgement alongside the assessment? Is there more than one assessment available that can be corroborated? Can areas such as ‘reading fluency’ assessments be manageably compared to corroborate the evidence.
3. If you are using a reading assessment to judge an intervention, try to compare with similar pupils. Can you compare similar pupils who did and did not get a reading intervention? It is data that should be treated with caution, but it does help you to triangulate potential impact. 
4. Move from baselining to diagnosing the issue. Many standardised reading assessments do a sound job of identifying struggling pupils, but are less sensitive to helping you understand why. Teachers and leaders can fill in that gap. Listening to pupils read can identify decoding or fluency gaps. Or you can notice a difference with questions to see whether knowledge or vocabulary gaps are playing out for pupils like Amelia. 
5. Be wary of sharing with pupils and parents. Reading ages are intuitive, but they can be alarming too. I’d be reluctant to share with both pupils and parents without very careful communications. It would need to be focused on what the school is doing and why - not just notions of a reading age.

Reading ages are seemingly intuitive and straight-forward, but they can be prone to being misread or more likely over-interpreted. We need to avoid this problem with a careful understanding of the data and tentative use of it too.

Related reading

• I've written a practical blog for teachers on 'Building understanding about literacy assessment' HERE.

I have a Teachology masterclass one-off on 'Literacy and SEND'. An important part of my SHARE model is effective assessment - such as the content of this blog. If you are interested in attending in person in Birmingham, or online, it is on Friday 4th December. FIND OUT MORE HERE.