Most days at Glenorie Public School, on the rural outskirts of Sydney, small groups of the school's youngest students pore over picture books. They are working hard to make up ground and lift their reading skills to the level of their classmates.
Many are in year 1 and were struggling to read in kindergarten. But, with specialised help, they are quickly catching up.
Katie Harmon, the school's assistant principal and specialist learning and support teacher, has been using a reading program called MiniLit to improve the literacy levels at Glenorie. In just over four years, reading at the school has improved out of sight.
"We can target small groups of children for as long as they need to be on the program and what we have found is that for those children with reading difficulties, 80 per cent show very significant growth," she says. "And another 10 to 15 per cent still make good, sound growth".
MiniLit is an evidence-based phonics program used in schools across NSW. Most teachers believe phonics works, and now they have access to the proof.
For the first time, the NSW Department of Education has teamed with a new not-for-profit education group, Evidence for Learning, to pull together more than 10,000 pieces of research from around the world to show teachers what works - and what doesn't - in the classroom.
Using three clear measures - the average cost, the strength of the evidence and the average number of months it adds to a student's learning - teachers, school leaders and parents can make informed decisions about how to improve schools.
Dubbed a toolkit for teachers, the online summary of educational research outlines what approaches work across 34 areas ranging from homework, class sizes and feedback to outdoor learning, phonics and behaviour interventions.
It looks at whether repeating a year is worth it, the usefulness of extending school time and the role of digital technology in the classroom.
Jenny Donovan is the executive director of the department's Centre for Education Statistics and Evaluation, which has worked with Evidence for Learning to develop the toolkit. She says it isn't about "telling teachers where they went wrong".
"Instead of academic research, this is about connecting directly to the work that is being used in schools," she says. "What we have done is give teachers access to a really useful resource that will provide tools and data and give them the chance to really drill down into classroom management.
"It is very targeted, very timely and very detailed."
The power of phonics
One of the 34 areas that Donovan says is likely to be of interest to teachers and parents is phonics, or learning to read new words by sounding them out.
When it comes to phonics, the research shows it is very cost-effective to teach - $1500 per teacher and $60 per student - and the evidence that it works is strong. It also adds as much as four months' to a student's learning.
"There have been a number of studies, reviews and meta-analyses that have consistently found that the systematic teaching of phonics is beneficial," the summary of phonics in the toolkit says.
But phonics can also be controversial. Late last year, a report by Jennifer Buckingham from the Centre for Independent Studies recommended a phonics screening check for all year 1 students. Britain introduced a similar phonics screening program for its first-graders in 2012.
The federal education minister Simon Birmingham endorsed Buckingham's research but the NSW Teachers Federation said the screening test was "anti-teacher", because it was based on not trusting teachers to do their jobs properly.
Donovan says there is "still a reading war happening" but the "overwhelming evidence" indicated that phonics was crucial to children in their early years.
"But it isn't enough to just have a program that nods to phonics," Donovan says.
Are devices all bad?
Another area of interest, Donovan says, is the use of digital technology. "Expenditure is estimated at $600 per student for initial equipment costs and a further $1000 per class ($40 per student) for professional development and technical support," the toolkit says.
Donovan says for this moderate cost, there is strong evidence that learning gains can be made through the use of technology.
"As technology becomes more ubiquitous and we begin the transition to NAPLAN online, it is helpful to gain some reassurance that digital technology can indeed improve learning if it is used appropriately," Donovan says.
Is homework worth it?
The benefits of homework, for both primary and high school students, is also analysed in the toolkit. And while the costs of teachers sending revision home may be low, extra work outside the classroom does not always make a huge difference.
For primary-aged children, whose homework is typically "reading or practising spelling and number facts", the evidence suggests that it only has a small impact on a student's achievement.
"It is certainly the case that schools whose students do homework tend to be more successful. However it is less clear that the homework is the reason why they are successful," the research says.
"There is some evidence that when homework is used as a short and focused intervention it can be effective in improving students' achievement, but this is limited for primary-age students."
When students get to high school, it is a different story. While in primary school it may only add an extra two months progress, in high school it is five months.
"Evidence also suggests that how homework relates to learning during normal school time is important. In the most effective examples homework was an integral part of learning, rather than an add-on," the research says.
And less may be more. "Studies imply that there is an optimum amount of homework of between one and two hours per school day (slightly longer for older students), with effects diminishing as the time that students spend on homework increases."
Does size matter?
One of the most politically sensitive topics for parents - class size - is also reviewed as part of the research. In NSW, the statewide average number of children in kindergarten is 20, 21 in year 1 and 22 in year 2. The department says classes should not exceed 30 students in years 3 to 6.
But despite the push to keep class numbers down, the research says it is an expensive measure for not a large gain.
"Intuitively, it seems obvious that reducing the number of students in a class will improve the quality of teaching and learning, for example by increasing the amount of high quality feedback or one to one attention learners receive," the research summary in the toolkit says.
"However, overall the evidence does not show particularly large or clear effects, until class size is reduced to under 20 or even below 15."
Matt Deeble is the director of Evidence for Learning, which is supported by Social Ventures Australia and the Commonwealth Bank.
He describes his organisation as "the Choice magazine for education".
"We know that we want students to get at least one year's worth of learning each year but how cool would it be for them to get 18 months or even 24 months in one year," Deeble says.
"This is about helping people make better decisions in education. So often we know teachers get told of a great reading program, for example, or they make a decision based on what the school down the road is doing," he says.
"But this is helping teachers strengthen their professional judgment and making evidence available when they are making their decisions."
Back at Glenorie and Katie Harmon says the reason MiniLit, and a similar reading program for older students call MultiLit, has been so successful is because they are researched-based programs proven to work.
"We can really make a difference to their reading by being able to take small groups out of their classroom three to four times a week," she says.
"For many of them who were struggling half way through kindergarten, they are back to where they should be at the end of year 1, reading a quality piece of literature. The results really have been outstanding."