The aim of this project is to analyse the conceptual skills that predict KS2 and 3 science achievement. Research on cognitive development has identified key understandings that students must attain in order to make progress in learning mathematics and science. Some of these key ideas are common to mathematics and science learning. For example, ratio and proportion are important in both domains. In mathematics, ratio and proportion are crucial for understanding linear functions; in science, some measure of quantities, such as density, velocity and temperature, are based on ratio and proportion. Similarly, the understanding of relationships between variables is of huge importance in mathematics as well as in science. Mathematical modelling relies entirely on one's understanding of how the relations between variables are represented mathematically. Scientific theories also involve models that require understanding, for example, whether a quantity increases or decreases when another increases.

Other concepts may be more important for scientific than for mathematical learning. Empirical sciences depend on experimentation, and thus understanding what makes a good experiment is an essential part of scientific learning and discovery. In contrast, mathematical experiments are more like simulations than experimentation. The idea that all variables must be controlled, except for one, in order for an experiment to produce unambiguous results is more important in science than in mathematics. Research on cognitive development supports the idea that some mathematical concepts are independent of this aspect of scientific reasoning. Cousins et al (1993) carried out a factor analytic study of measures of young people's understanding of proportionality, probabilities and control of variables in empirical situations. They found that the first two measures loaded on one factor whereas the third loaded on a separate factor.

In brief, some conceptual skills are common to mathematics and science learning whereas others are more important for science learning. This analysis leads to two hypotheses regarding the prediction of children's science achievement.

The first hypothesis is that children's science learning will be based on (at least) two conceptual skills: one that is quantitative in nature and includes reasoning about relations between variables and mathematical concepts such as ratio and proportion, and a second non-quantitative skill that is related to the understanding of the need to control variables. Thus, measures of each of these conceptual skills should make significant and independent contributions to the prediction of achievement in KS2 and 3 Science, after controlling for general intelligence.

The second hypothesis is that children's understanding of the control of variables will be a strong predictor of KS2 and 3 science achievement but not of KS2 and 3 mathematics and English achievement.

Method

The ALSPAC dataset includes:

* a measure of general intelligence, obtained when the children were in Year 4;

* measures of children's understanding of the relations between quantities, including their understanding of proportionality, obtained when the children were in years 6 and 8;

* a measure of the children's understanding of the control of variables, obtained when the children were in year 6.

These measures will be used as predictors of the children's attainment in KS2 and 3 Science, and also KS2 and 3 Mathematics and English, in order to test both predictions. We will use regression analyses and structural equation models to test the predictions. We will use fixed order regression analyses, with the children's age at time of testing and their general intelligence entered as first and second steps, respectively. The third and fourth steps will be the children's mathematical reasoning and understanding of control of variables. These measures will be entered separately, each being entered once as the third and once as the fourth step, in order to test whether each one makes a contribution to the prediction of KS2 and 3 science, after controlling for the other one. The outcome measures will be the KS2 and 3 science, mathematics and English achievement. We expect that the understanding of control of variables makes a significant contribution to predicting KS2 and 3 science but not mathematics and English achievement. The data will also be scrutinised to see to what extent other variable (working memory, SES and reading ability) account for variance in KS2 and 3 science achievement and these variables will be controlled for, when necessary.

The analysis of the cognitive data will be complemented by an analysis of how much children say that they like science and mathematics, in order to see whether children's liking contributes to the prediction of achievement above and beyond an analysis of their cognitive skills.

The results will contribute to the understanding of the cognitive skills that support science learning. If both measures are strong predictors of science achievement, the implication is that both aspects of scientific concepts should be considered when new science topics are introduced in the classroom.