Appendix B: about the research

Research methodology

This Report is based on a large-scale quantitative internet survey and desk research conducted by Quadrangle on behalf of RAC.

For the internet survey, Quadrangle interviewed 1,040 British motorists (defined as currently having a valid driving license and driving at least once a month). The survey was completed during August and September 2008.

The sample was nationally representative on age, gender, socio-economic groups and region.

The number of company car drivers was boosted, and as such weighting factors have been applied to restore the overall sample to be nationally representative of households with company cars. The sample was also weighted for market representation of new versus second-hand car ownership.

In addition, Quadrangle also undertook four focus groups with motorists, in order to explore in-depth the issues arising from the quantitative survey findings. The groups were undertaken to generate understanding of the differences between younger and older motorists and urban or rural dwelling drivers.

Quadrangle also undertook desk research, using all previous available RAC Reports on Motoring from 1988 – 2008*, and data in the public domain published by the Department for Transport and the Office for National Statistics.

Statistical reliability

Any figure taken from a sample can never be taken as a precise indication of the actual figures for the total population being sampled. The figures shown are an estimate, within a small margin of error, of the actual figures. The error margin varies with the sample size – the larger the sample is, the lower the error will be. It also varies with the proportions answering so the error is lower for a 90/10 result than for a 50/50 result.

In order to illustrate the use of varying sample sizes and their affect on the statistical significance of results, the table below outlines the degree of statistical error broadly associated with different sample sizes from the car drivers' survey. For example, from a sample of 1,000, if 50% answered in a particular way, we would be 95% confident that the true range is between 47% and 53%.

Sample size    Percentage error 90/10 50/50 result   
2,000 +/-2 +/-3
1,000 +/-2 +/-3
800 +/-2 +/-3
600 +/-2 +/-4
400 +/-3 +/-5
200 +/-4 +/-7
100 +/-5 +/-10

Appendix A: about this Report