The introduction of the early years foundation stage (EYFS) has reignited debate over how under-fives should be assessed, with academics and teachers questioning the advice that assessment should be "based on practitioners' observation of what children are doing in their day-to-day activities".
Dr Christine Merrell, of the Centre for Evaluation and Monitoring at Durham University, said that computer assessments, such as the centre's own Performance Indicators in Primary Schools (Pips), also have a role to play.
"We have grave concerns about the way the EYFS is proposing to assess young children to find out what they can do," she said.
"When you think of child-initiated activities and observing those, you are relying on a child to show what they can do, which is not necessarily the most reliable way.
"This (guidance) effectively instructs schools not to use well- established, valid and reliable measures that have been designed specifically for early years."
Professor Pat Preedy, executive principal of Sherfield School, an independent in Hampshire, said she also believed a computer-based assessment as well as observation of child-led activities was necessary to get a full overview of a child's progress. "I wouldn't want one without the other," she said.
The issue is a hot topic, with national profile data about under-fives' progress in 2008 due to be published next week. Last year, 71 per cent of five-year-olds achieved a good level in personal, social and emotional development, and 49 per cent had good communication skills.
A recent TES poll of 1,480 early years teachers also revealed that the way children are assessed is a significant issue for them.
At the end of reception year teachers have to complete a profile for each pupil, showing their progress on 13 nine-point scales.
The foundation stage guidance states that, for each scale point, no more than 20 per cent of the evidence that a child has reached a certain goal - such as being able to name the letters of the alphabet - should come from activities that are adult-led. Most of the evidence should be drawn from "knowledge of the child, observation and anecdotal assessment".
Before the foundation stage profile was introduced in 2003, baseline assessments were carried out at the beginning of reception year. There were 90 credited schemes including Pips, which, according to a 2001 survey, was used in 17 per cent of local authorities.
Some praised Pips for measuring progress while being easy to administer and enjoyed by children. But others thought it too narrow and too dependent on a child's performance on a single occasion.
The 2001 survey fed into a Qualifications and Curriculum Authority report on changes to the baseline assessment, which proposed a single national test at the end of reception, but warned that on-entry assessment was seen as vital by teachers.
The idea of a single national test was welcomed in principle, but the resulting foundation stage profile was criticised as overly bureaucratic. Ofsted reported that this was partly because heads still wanted baseline assessments done as well.
Ann Tanner, head of Whitley Park Infants in Reading, advocates using a mixture of methods. This year, all reception-aged children in Reading will be assessed on entry in three ways: a home visit; observations; and a computer programme for diagnosing language difficulties.
She said: "A home visit is the initial assessment teachers make. In nursery, every child will be assessed through observations on communication, language and literacy development, and on personal, social and emotional development. There will be an additional assessment with Language Link.
"Language Link breaks down ability, rather than saying `this child has good language achievement and this one hasn't'. It gives finer detail so teachers can plan activities to support them.
"The problem with data is when people abuse it. People talk about testing three-year-olds instead of trusting professionals, which is what we are."
Jo Goodall, an early learning adviser and school improvement partner in Reading, said that the foundation stage was more than a curriculum - it had an underlying philosophy and principles.
"Language Link is to support speech and language therapists," she said. "I don't think one simple (adult-led) activity will negate all the child- initiated activities. You have to have a balance.
"This is not number crunching. We just want to know, is there a problem here or not?"
Jan Dubiel, programme leader for the foundation stage profile at the National Assessment Agency, said: "Research strongly indicates that observation is the most consistent way of getting information about children. Testing may provide consistent statistical data, but it's not a reliable way.
"We're not saying that adults can't tell children what to do, but children achieve their highest level when they have an inner motivation. That's when they demonstrate their real knowledge."