Nutrition is the joker among the sciences. None is more important, yet none produces more uncertain or ambiguous conclusions.

Today’s paper in The Lancet on the effect of food additives on children’s behaviour is no exception. Despite attention to detail, the results are confusing. What parents want to know is whether food additives and preservatives are making their children behave badly. To this, the study answers yes – but with important caveats.

Parents also want to know which additives are responsible, and they want to see them removed from foods. To these questions the study does not have answers. Small wonder that the Food Standards Agency has thrown the decision back on to parents, suggesting if they are worried about their children’s behaviour they should consider modifying their diets.

The Southampton team responsible used two cocktails of additives and preservatives, called Mix A and Mix B, in a drink. Both included, per portion, 45 milligrams of sodium benzoate (E211) a preservative widely used in fizzy drinks, jams and fruit juices.

Mix A also included 20mg of three colourants, Sunset Yellow (E110), Carmoisine (E122) and Tartrazine (E102). Mix B had a greater level of colourants (30mg) made up of Sunset Yellow, Carmoisine, Quinoline Yellow (W104) and Allura Red (E129), and was designed to represent what an average child consumes.

The results show that in three-year-olds, Mix A had a significantly adverse effect on behaviour, which was assessed by teachers and parents. But Mix B did not.

In eight to nine-year-olds Mix B had significant effects, but Mix A did not, when all the children were included in the analysis. But if only those children who drank at least 85 per cent of the drinks were included, and for whom there was no missing data, then both Mix A and Mix B had significant effects.

The size of the effect, roughly, was an increase in hyperactivity that represented less than a tenth of that seen in children diagnosed with ADHD (attention deficit hyperactivity disorder).

The study still does not tell us which additive is responsible, or whether the guilty party is the preservative, sodium benzoate. This matters because while it may be relatively easy to dispense with colourants, preservatives are important to prevent spoiling.

It is important to remember that the effects are on the margin of statistical significance. They are also small when compared with the range of behaviour seen in children. That said, they are in line with earlier studies.

So is this the final proof that food additives cause misbehaviour? Not quite. It demonstrates an association, not cause and effect. However, it is likely to accelerate moves to reduce additives in children’s food further, and to discourage parents from allowing them to eat many of the fast foods frowned on by nutritionists. That is an outcome unlikely to do any harm.

Times recommends

• How to solve the big style questions
• Aids in Africa: ‘It is the staff who really inspire you’
• What's the best low-alcohol drink for Christmas?