Dangerous statistical exercise

EXTRAPOLATION from small samples to a whole population is always a dangerous statistical exercise and the results should always be accompanied by statements about the error band around a single number—so often it is only the latter that is cited, writes Mike Todd.

 

Whilst the calculation used in the case of estimating the number of people visiting canals and rivers may well have a central figure of 350 million, it may also have an error band of 300 million!

Hard to believe

Rather than continuing the somewhat sterile debate between CaRT and those who find the conclusions somewhat hard to believe, it would be much better for the media to engage with CaRT (using FOI if necessary) to obtain greater detail of how the figures were obtained and what lies behind the headline.

In political debate, it is common for different sides to quote statistics seemingly pointing to opposite conclusions yet all having some support. The radio programme that analyses these is often more illuminating that any other debate! The background is often that the varying figures are actually measuring different things but in the inevitable simplification process that public debate seems to require, they appear to be the same yet contradictory.

More insight

Unless we can have more insight into where the numbers came from (and I don't just mean saying that they came from an 'independent' pollster so they must be right) we are never really going to understand the current and future state of interest in using the facilities which canals and rivers offer everyone. Perhaps narrowboatworld could lead the way in seeking this information and publishing it so that we can all critique the process.