Should we dispense with the D word?

By Kevin Wheldall, Anne Castles and Mandy Nayton

[Note: This article appeared in a more abbreviated form in The Conversation:

https://theconversation.com/should-we-do-away-with-dyslexia-24027]

The word dyslexia seems to arouse strong emotions, both for and against, and has a chequered history. Viewed by some as a device to spare the blushes of middle class parents whose children struggle to learn to read, it is seen by others as a credible explanation for the reading difficulties their child has been experiencing.

In their recently published book, The Dyslexia Debate (http://www.cambridge.org/us/academic/subjects/psychology/educational-psychology/dyslexia-debate), Joe Elliott and Elena Grigorenko open up a can of worms. But it is a can that needed opening, and we applaud these authors for bringing this important issue up for discussion and debate. Elliott and Grigorenko argue that the term “dyslexia” should be abandoned – that it is an imprecise and unhelpful label that does nothing to assist the individuals to whom it is applied.

So what do Elliott and Grigorenko say, and why do they say it?

Two key points need to be made before proceeding. First of all, this debate is largely about terminology. Elliott and Grigorenko are not denying the reality of children’s reading difficulties, or that these difficulties need to be identified and treated as early as possible. What they are arguing is that giving the label of “dyslexia” to children who experience such problems is not helpful.

Secondly, it is important to keep in mind that reading ability falls on a continuum in the population; it is normally distributed like height or weight. This means that deciding whether a child does or does not have a reading difficulty will always involve applying a somewhat arbitrary cut-off. In this sense, a diagnosis of dyslexia is similar to a diagnosis of obesity, but quite different from a diagnosis of, say, measles where it is clear when someone has it and when they don’t.

With these points in mind, let’s look more closely at Elliott and Grigorenko’s two main arguments:

1.    There is no agreement about how to diagnose dyslexia

Elliott and Grigorenko’s first key point is that applying the label of dyslexia is intrinsically unscientific because there is no universally agreed set of criteria for its diagnosis. What one clinician might call dyslexia, another does not. Some apply the label to any child who struggles with learning to read; others apply it only when the reading difficulty is accompanied by strengths in other intellectual domains; still others when the reading difficulty is associated with particular cognitive “markers” such as phonological or visual deficits. And even within these different definitions, there is variability associated with where the cut-off for an impairment is applied: consequently, estimates of the prevalence of dyslexia may range from 3% - 20% of the population.

It is undoubtedly true that the term “dyslexia” has been used in a wide variety of contexts over the years, and that this has led to considerable confusion. We think that three particular factors have contributed to the problem:

Firstly, there has been a failure to distinguish between research and clinical uses of the term. Researchers often select samples of “dyslexics” with very specific profiles. They do so in order to answer particular research questions, or to control for factors in which they are not interested in a certain context. For example, they may select their sample to have average or above average IQ, so that this factor does not influence their results. But the fact that the use of these specific selection criteria can be valuable in a research context does not mean that they should necessarily inform a clinical diagnosis of dyslexia. Similarly, researchers may decide that, for their experimental purposes, ‘dyslexia’ will be defined very generously as those students scoring below one standard deviation from the mean. But this does not necessarily mean that this (relatively large) proportion of the population (about 16%) should be regarded as dyslexic and hence in need of special treatment and/or afforded educational dispensations.

Secondly, there has been a tendency to conflate symptoms and causes within definitions of dyslexia. Sometimes the condition is defined purely in terms of the presenting problem – a reading accuracy and/or fluency difficulty  – with the diagnostician remaining agnostic as to its underlying cause. In other cases, the definition incorporates a particular theoretical perspective as to why the reading difficulty arose in the first place.  The most prominent and certainly widely-supported of these theories centres on underlying language-based phonological deficits, but there are also a host of other neural, visual and attentional theories.  Given the wide range of theories as to the causes of dyslexia, it is not surprising that this has generated an equally wide range of definitions.

Thirdly, the term is widely used, and very frequently misused, by non-experts in the field and by the mainstream media. The label is particularly popular with promoters of dyslexia “cures” of dubious merit, including such things as: nutritional supplements, exercise regimes, coloured glasses and ‘high frequency’ ear plugs. The briefest of Google searches confirms this fact - and this, of course, only serves to add to the confusion.

But, these points aside, is it the case that there is no generally agreed set of criteria for the diagnosis of dyslexia? We think that this may be overstating the case. Amongst experts in the field, there is in fact a substantial degree of consensus about what is meant by the term and how it should be defined in a clinical context. Dyslexia is widely viewed as a severe and enduring reading difficulty that persists despite high quality instruction and evidence-based intervention. This is enshrined in documents such as the Rose report (http://webarchive.nationalarchives.gov.uk/20130401151715/https://www.education.gov.uk/publications/eOrderingDownload/0201-2006pdf-EN-01.pdf)

in the UK  and the report of the Australian Dyslexia Working Party (2010) (http://www.dyslexiaaustralia.com.au/DYSWP.pdf).

The definition of dyslexia above is effectively couched within the widely-accepted “Response to Intervention” model of remediation (RTI) (http://www.musec.mq.edu.au/community_outreach/musec_briefings#17):

a model that  seeks to match the amount and nature of reading intervention to the support needs of the child. In doing so, it provides relatively straightforward means of distinguishing between children who are performing poorly on reading measures because they have not received the appropriate instruction for some reason and children who are performing poorly because they have severe and enduring learning difficulties that require sustained intervention (whom we might classify as having “dyslexia” or some other label).

The RTI model works as follows: All children receive a systematic program of scientific evidence-based reading instruction right from the outset of schooling, ie a program of instruction that is predicated upon what reading scientists have found to be critical and most effective in terms of teaching children to learn to read. This is referred to as Tier 1 instruction. The reading progress of all students is closely monitored, and those children whose performance, following instruction, is poor on curriculum-based measures, regardless of the reason, are identified as low-progress readers and are provided with more intensive small-group reading instruction. This is referred to as Tier 2 and typically involves about 20% of the age cohort.  (In some socially disadvantaged areas, of course, this figure will be much higher.) The progress of Tier 2 children is again closely monitored and those who continue to have low scores, that is, who are not “responding to Tier 2 intervention”, are provided with even more intensive one-on-one intensive remediation and support. This is Tier 3 instruction.

Children who still struggle following a period of intensive Tier 3 intervention or who progress at an unusually slow rate, are viewed as having, a severe and enduring reading difficulty that persists despite high quality instruction and evidence-based intervention. This has been estimated to be about 3% of the population of students. Thus, without needing to look to underlying causes, or to associated deficits, a small subgroup of children can be identified who can reliably and consistently be given a diagnosis of “dyslexia” on the basis of their need for continuing intensive literacy instruction. However, it is generally viewed as important to gather additional clinical information from appropriate professionals at this point to contribute to the development of effective individual education plans (IEPs) with recommendations for both remediation and accommodations.

There are two key features of this definition. First, it makes a distinction between children who are struggling with reading because they have not had the appropriate instruction (perhaps because they have not attended school regularly, or have had poor teaching) and those who are struggling despite having had sufficient opportunities. Secondly, it focuses the diagnosis at the level of reading itself. Scientists have come a long way in developing precise and detailed theories of the reading process, and these have been translated into valid and reliable clinical assessment tools. It is our view that, armed with these tools, well-trained clinical experts are in a stronger position to provide a scientifically-informed diagnosis of dyslexia than Elliott and Grigorenko suggest. Whether the actual term ‘dyslexia’ is used or some other label is employed, this small group exists and can be identified.

It is only fair to note, however, that while RTI does have the potential to: improve reading outcomes significantly and will, in addition, identify students at risk of reading failure far earlier, create a more equitable system and should result in better ongoing monitoring and assessment of all students, we are only seeing a minority of schools implementing anything even vaguely resembling RTI at this stage in Australia. This makes the use of it, at this point, as the mainstay identification model challenging, to say the least. This is not to say, however, that it should not be an important goal. It also means that until RTI is established, students will not necessarily be picked up as falling behind until later in primary school or even into secondary school. This means that the transition through the three tiers is not as straightforward as it is in the early years. Identified students will simply move straight into a Tier 2 or even Tier 3 intervention. And finally, while it is our firm view that the RTI model has many advantages, it is viewed by some as being limited in that it does not individualise, it does not diagnose, and it does not classify.

2.    Reading interventions will be the same whether dyslexia is diagnosed or not.

The second key point made by Elliott and Grigorenko is that a diagnosis of dyslexia does not have any implications for treatment. Again, they are broadly correct. Let us suppose that two 8-year old girls present at a clinic, both of whom can barely read. One girl has frequently been ill and has attended school only intermittently over the past two years. As a result she has fallen behind in learning to read. The other girl has a history of struggling to learn to read in spite of having been provided with additional instruction. The latter child might well be diagnosed with “dyslexia”; the former most likely not. But the programs of intervention put in place for each child would most likely differ very little in both content and method. A large body of evidence tells us that the most effective thing we can do for each of these children is to provide them with systematic, intensive evidence-based reading remediation, targeted at the gaps in their reading skills.

So does this mean, as Elliott and Grigorenko argue, that a diagnosis of dyslexia (or some other label) is unnecessary and redundant? Again, we feel that this may overstate things. Although the nature of the treatments provided to these children will be similar, the evidence tells us that their length and intensity are likely to be quite different. Our first girl should respond quickly and well once she receives the required intervention; our second girl, by definition based on the RTI model, is likely to need sustained and ongoing support. She may need one-on-one assistance and, ultimately, compensatory technology.

More generally, the fact that the treatment is similar across the spectrum of a condition does not mean that there is no justification for giving a label to those at the extreme end. In the same way that the diagnosis of “obesity” serves to identify the most severe and at-risk cases along the continuum of weight, and the diagnosis of “hypertension” serves to identify the most severe and at-risk cases along the continuum of blood pressure, a diagnosis of dyslexia can identify the most severe and at-risk cases along the continuum of reading. Such labels focus attention and resources where they are most urgently needed. Where we might disagree is in the term we use to describe such children and their condition.

There remain the quibbles about terminology. Many researchers and clinicians do not like the term “dyslexia” because it medicalises the condition and invokes a disease model that is not appropriate. They prefer a term like “reading disability” or “reading impairment”. Others, however, counter that the use of a medical term like dyslexia attracts attention and resources to a problem that may not be so forthcoming if these other terms were used. And as Dorothy Bishop notes, this concern seems to be borne out in the case of the much less visible condition of “Specific Language Impairment” (http://deevybee.blogspot.com.au/2010/12/whats-in-name.html). Finally, not to be underestimated is the positive effect that a diagnosis of dyslexia has on the parents and children involved, validating their concerns and often providing a much-needed boost to self-esteem. For these reasons, there is considerable divergence in the field as to which label should be preferred (and indeed divergence even amongst we three authors!)

Elliott and Grigorenko argue that the most important thing is that all children who encounter literacy difficulties receive the help that they need. We could not agree more. At the end of the day, what we choose to call the small proportion of students with severe and persistent reading problems is not nearly so important as ensuring that they receive the support that they need. In the past, passionate advocates for dyslexia may have gilded the lily by claiming that up to 20% of children are dyslexic (and some still do). This is unhelpful and ultimately counter-productive since government agencies will baulk at the expense of providing special treatment and dispensations for such a large proportion of the population. Nor is such expenditure warranted when the reading difficulties of the vast majority of low-progress readers may readily be attended to in school with effective small group Tier 2 and individualised Tier 3 interventions.

Kevin Wheldall is an Emeritus Professor of Macquarie University and Director of the MultiLit Research Unit. You can follow him on Twitter (@KevinWheldall) where he comments on reading and education (and anything else that takes his fancy).

Email: kevin.wheldall@pecas.com.au.

Professor Anne Castles is Head of the Department of Cognitive Science at Macquarie University and Deputy Director of the ARC Centre of Excellence in Cognition and its Disorders. You can follow her on Twitter (@annecastles), where she comments on issues related to reading difficulties and other cognitive disorders.

Mandy Nayton is currently the Executive Officer of DSF Literacy and Clinical Services in Perth, Western Australia as well as President of AUSPELD, The Australian Federation of SPELD Associations. She is an educational and developmental psychologist and qualified teacher with extensive experience in the field of literacy education. Mandy is an Adjunct Research Fellow at Curtin University’s School of Psychology and Speech Pathology, and a member of the Federal Government’s Schools Disability Advisory Council.

Rhubarb, rhubarb, rhubarb

 It was one of those supermarket moments. The young man at the checkout looked at the long red stalks topped with exuberant green foliage and asked my wife what it was. My heart went out to him. To have reached your teenage years without knowing the sensuous delights of rhubarb crumble was almost unbearable – far superior to the traditional teenage delights of sex ‘n’ drugs ‘n’ rock ‘n’ roll, but then I am well over sixty and he is, perhaps, sixteen. And perhaps he just did not recognise rhubarb in its raw state and, in any case, I digress.

            My wife telling him the name of the product did not seem to help him to locate it on his checkout touch screen. We waited. In desperation, I hissed “Tell him it’s r-H-u …”. Problem (finally) solved.

   As if I needed any reminding, this served to illustrate yet again the importance of factors other than decoding (phonically or otherwise) in reading and, indeed, spelling. To be literate, to be able to read and write (and spell) with reasonable proficiency, it is not enough to be able to decode or encode words and text. You actually have to know what you are reading or writing about. You need knowledge.

This sounds painfully obvious on the surface: how could it be otherwise? But a sole preoccupation with decoding and encoding, especially when teaching low-progress readers, can sometimes distract one from the equally important task of teaching … well, stuff. It is all very well to be able to read aloud “The rodent consumed the gorgonzola” but, effectively, barking at print is all that will be achieved unless you know what a rodent is, what consume means and that gorgonzola is a type of cheese.

The problem is, of course, that learning quite a bit of this ‘stuff’ actually comes about via reading. Consequently, if you are poor at reading, you will tend to read far less and to learn less general knowledge. Moreover, many of the words and terms that we learn do not crop up frequently in everyday conversation; we only encounter them in books. Hence, low-progress readers experience what Keith Stanovich calls ‘the Matthew Effect’ whereby the rich get richer and the poor get poorer; those who can read well, read more and learn more; those who read poorly, read less and hence learn less.

All of this leads to the rather obvious point that we need to teach ‘stuff’ to low-progress readers as well as how to decode effectively and efficiently. In recent years, there has been a noticeable and welcome shift among those interested in teaching reading effectively to focus on all five of the five ‘big ideas’ of reading instruction; to give equal prominence to vocabulary and comprehension as to phonemic awareness, phonics and fluency. We also need to focus on the acquisition of general knowledge about the world and to find ways of bringing low-progress readers up to speed with what typically developing readers are already likely to know. To contemplate a world without rhubarb in it is just too sad.

Small Bangs for Big Bucks: The long term efficacy of Reading Recovery

‘Best Evidence in Brief’ is a fortnightly email newsletter “brought to you by the Johns Hopkins School of Education's Center for Research and Reform in Education and the University of York's Institute for Effective Education”, both led by Bob Slavin. In the latest issue (January 30, 2013 http://tinyurl.com/a9bn66a), there is an interesting item entitled 'Lasting effects from Reading Recovery' citing a recent report (dated December, 2012) by Jane Hurry from the Institute of Education, University of London (the British home of Reading Recovery). Hurry’s report is entitled ‘The impact of Reading Recovery five years after intervention’ (http://tinyurl.com/avhubv9).

This is how ‘Best Evidence in Brief ‘ reports the study (with a link to the report):

“A recent report for the Every Child a Reader Trust looks at the impact of Reading Recovery five years after intervention. The program is known to have impressive effects in the short term, but less is known about its long-term effectiveness. This study suggests that the substantial gains which result from receiving Reading Recovery in Year 1 (the UK equivalent of kindergarten) continue to the end of primary school.

At the end of Year 6 (the UK equivalent of 5th grade), the study followed up 77 children who had received Reading Recovery five years earlier, 127 comparison children, and 50 children in Reading Recovery schools who had not received Reading Recovery. Findings showed that children who had received Reading Recovery made significantly greater progress in English than the comparison children by the end of Year 6. The 50 comparison children in the Reading Recovery schools were also significantly out-performing the comparison group in non-Reading Recovery schools on the reading test.”

It would be reasonable to conclude from this summary (published by two research centres apparently devoted to championing evidence-based practice) that Reading Recovery had already been shown, unequivocally, to be effective in the short term and that now there was convincing evidence for its efficacy in the longer term too, at the end of Year 6, in fact.

Many reading scientists are less convinced of the Efficacy of Reading Recovery in the short term; see, for example, Reynolds and Wheldall, 2007: http://tinyurl.com/bggyl46

But let us concentrate, for the present purposes, on the reporting of longer term effects as reported by Hurry and summarised by Best Evidence in Brief. This is how Hurry, herself, summarises her findings in her report:

“Reading Recovery is part of the Every Child a Reader strategy to enable children to make a good start in reading. Reading Recovery is well known to have impressive effects in the shorter term but less is known about its long term effectiveness. The present study followed up at the end of Year 6: 127 comparison children, 77 children who had received Reading Recovery five years earlier and 50 children in Reading Recovery schools who had not receive Reading Recovery. The children who had received Reading Recovery had made significantly greater progress in English than the comparison children by the end of Year 6, achieving on average a National Curriculum Level of 4b compared with a borderline between Level 3 and 4 in the comparison group. Comparison children in the Reading Recovery schools were also significantly outperforming the comparison in non Reading Recovery schools on the reading test. 78% of Reading Recovery children achieved Level 4 in English compared with 62% in the comparison group in non Reading Recovery schools and 64% for the comparison children in Reading Recovery schools. There was a tendency for Reading Recovery children to be receiving less SEN provision than children in the other two groups, but this only reached statistical significance for those on School Action Plus or a Statement. This suggests that the substantial gains which result from receiving Reading Recovery in Year 1 continue to deliver a significant advantage for those children at the end of the primary phase, providing a surer footing for transition to secondary school.” (P. 3)

If you read the details in the actual report, however, the facts for the end of Year 6 results are reported by the author (Hurry) as follows:

“The Reading Recovery children were still doing significantly better in reading (β=.191,  p<.005), effect size (Cohen’s d) = .39) and writing (β=.162, p<.013, effect size (Cohen’s d) = .33) than the comparison children in non Reading Recovery schools. They were also scoring significantly higher on their maths test (β=.154, p<.036, effect size (Cohen’s d) = .31). However, they were not significantly better than the comparison children from the Reading Recovery schools on any of the measures (reading, writing or maths). Indeed the comparison children from Reading Recovery schools, ie. those that were poor readers at six but did not receive the programme, were also doing significantly better in reading than the comparison children from non RR schools (β=.222, p<.002, effect size (Cohen’s d) = .24).” (p.12) (present author’s emphasis)

So, let’s be clear about this. The children who received Reading Recovery did not perform significantly better than the comparison students from the same school who had not received Reading Recovery. (Note: no details of statistical significance testing or effect sizes are reported for these comparisons.)

But the Conclusions section of the report states (and this is the only conclusion iterated):

“These findings indicate that effects of Reading Recovery are still apparent at the end of Year 6 and that even the children who attended Reading Recovery schools but were not offered the programme benefited somewhat from the ECaR programme.” (P. 22)

It is a source of some consternation to reflect on the fact that neither the report’s author (Hurry) nor the writers of ‘Best Evidence in Brief’ appear to have considered the (to me) obvious alternative conclusion: that there is no need to actually take part in Reading Recovery; merely attending a Reading Recovery school appears to be sufficient!

Another interpretation of these data is that they provide no evidence for the long term efficacy of Reading Recovery because those children in the school who did receive Reading Recovery performed no better than those who did not. And both of these groups performed better than the comparison children in the non-Reading Recovery schools. In other words, there is no discernable effect for the program per se, only for differences between schools. Moreover, even the significant differences between the two groups in the Reading Recovery Schools and the non-Reading Recovery school are accompanied by only small effect sizes, all of which are below Hattie’s hinge value of 0.4

Considering the huge expense involved in one to one Reading Recovery tutoring, these are very small bangs for very big bucks.

PIRLS before Swine: Or why Australia sucks at reading

John Lennon was renowned for his sharp, and oft times acidic, wit. When asked if Ringo was the best drummer in the world, he responded that Ringo was not even the best drummer in the Beatles! I was reminded of this when reviewing the latest (2011) results from the Progress in International Reading Literacy Study (or PIRLS). Not only were Australian students not the best readers in the world, they were not even the best readers among the English speaking nations surveyed. They were, in fact, the worst. We can take small comfort from the fact that New Zealand performed only marginally (and not significantly) better than Australia.

The PIRLS project essentially assesses reading comprehension by requiring students to read selected texts and then to answer questions about the material read. Year 4 students are assessed because this typically marks the point of transition from learning to read to reading to learn. (Note that 2011 was the first time that Australia had taken part in PIRLS.)

Overall, 45 countries were included in the study (excluding those countries who tested older or younger readers or who took part for their own internal benchmarking purposes, and whose results are not reported). Australia came 27^th in the league table of countries (with a mean score of 527), below all other English speaking countries and significantly lower than 21 other countries overall, including all other English speaking countries except New Zealand (mean score 531).

To put this in perspective let’s look at how some of these other English-speaking countries performed. Singapore (mean score 567), for example, came 4^th, one of the four top performing countries significantly above the others. Northern Ireland came fifth (mean score 558) and the United States came 6^th (mean score 556) (compared with 14th out of 40 in 2006). England came 11th (compared with 15th out of 40 in 2006) and Canada (mean score 548) came 12^th. (Note the improvements in performance from 2006 to 2011 by both England and the United States.)

As well as reporting mean scores by country, PIRLS also provides details of performance against four benchmarks: Advanced, High, Intermediate, and Low (and those who fail even to qualify for Low ie Below Low). In Australia 7% of students failed even to meet the Low benchmark and a further 17% met only the Low benchmark. Here are comparison percentage figures for other English speaking countries of interest: 

                        Below Low    Low     Combined              

Northern Ireland		3	10	13
Singapore		3	10	13
Canada		2	12	14
United States		2	12	14
England		5	12	17
New Zealand		8	17	25
(Australia)		7	17	24

In summary, Australia and New Zealand have over twice as many students failing to meet even the minimal Low standard as Northern Ireland, Singapore, Canada and the United States; and over one and three quarters times as many low-performing students overall (Below Low and Low combined). England falls in the middle of these two groups of countries.

These results may have come as a shock to many educationists who had been blithely arguing that there was no literacy crisis in Australia. But they provided simple confirmation for Australian reading scientists who had been warning of this problem for some time and had argued (remarkably accurately, as it turns out) that a quarter of Australia’s students could be regarded as low-progress readers. In 2004, a group of Australian reading scientists and clinicians wrote an open letter to the then Federal Minister of Education, Brendan Nelson, arguing the need for reform regarding the way reading is taught in Australia and the need for literacy teaching to be based on the available scientific evidence. The subsequent National Inquiry into the Teaching of Reading reported at the end of 2005, essentially reiterating these concerns and stating clearly what needed to be done to improve reading standards in Australia. In short, the Report was subsequently simply ignored.

Moreover, following the implementation of the National Assessments Program – Literacy and Numeracy (or NAPLAN) from 2008, we were subsequently assured (annually) that all was well on the reading front. As recently as in 2012, we were reassured that for the performance of Year 3 students in Reading only 4.4% were in Band 1, having failed to meet the National Minimum Standard, and only a further 9.4% were in Band 2 ie at the national Minimum Standard (a combined total of 13.8% of students). (Note that the NAPLAN measure of reading is similar to the reading comprehension measure employed by PIRLS.)

Clearly, we have been deluding ourselves by measuring student reading performance against unrealistically low benchmarks that do not withstand international scrutiny. NAPLAN, as a reading performance measure, needs to be recalibrated against international standards so that Bands 1 and 2 combined ‘capture’ the bottom performing 25% (not the current 14%) of students. (These low-progress readers should subsequently be earmarked for immediate additional instructional support.) Moreover, it should not be beyond the wit of the NAPLAN methodologists to tie the NAPLAN scale to the metric employed by PIRLS so that progress towards achieving the international standards could readily be monitored

Finally, it is interesting to note that two English-speaking countries that have begun to take reading instruction seriously in recent years, and who have urged that reading instruction be based on the best available scientific evidence, namely the US and England, have both improved their international standing substantially in the PIRLS league table from 2006 to 2011. Similarly, the two English-speaking countries that have performed so poorly, namely Australia and New Zealand, are those that have clung most tenaciously to the discredited ‘philosophy’ of whole language literacy instruction. Can this be simple coincidence? I think not.

Reference note:

A summary of Australia’s performance in PIRLS 2011 is provided in:

Thompson, S., Hillman, K., Wernert, N., Schmid, M., Buckley, S., & Munene, A. (2012). Highlights from TIMMS & PIRLS 2011from Australia's perspective. Melbourne: Australian Council for Educational Research.

http://www.acer.edu.au/documents/TIMSS-PIRLS_Australian-Highlights.pdf