The variables x x\,x and y y\,y are recorded for a set of 15 observations. Two different codings are used.
Coding A: p=log10xp = \log_{10} xp=log10x and q=log10yq = \log_{10} yq=log10y. The product moment correlation coefficient between p p\,p and q q\,q is 0.921, and the regression line of q q\,q on p p\,p is q=0.664+1.48pq = 0.664 + 1.48pq=0.664+1.48p.
Coding B: u=xu = xu=x and v=log10yv = \log_{10} yv=log10y. The product moment correlation coefficient between u u\,u and v v\,v is 0.998, and the regression line of v v\,v on u u\,u is v=0.802+0.0913uv = 0.802 + 0.0913uv=0.802+0.0913u.
State, giving a reason, which of the models y=axby = ax^{b}y=axb and y=kcxy = kc^{x}y=kcx is better supported by these data.
Using the better-supported model, find the values of its two constants. Give your answers to 3 significant figures.
Explain why the value 1.48 in Coding A's regression line cannot be interpreted as the increase in y y\,y produced by an increase of 1 in xxx.
157 exam-style questions on AQA A Level Maths 2.2.2 Scatter diagrams and correlation. Each one has a worked solution and a mark scheme showing where the marks go.