Selection bias makes everyone look brilliant

Of course drivers who switched insurer saved money. The people offered a higher price stayed where they were. The people offered £2.40 less could not be arsed with the form. The final group was selected for receiving a quote attractive enough to overcome inconvenience.

Different insurers favour different combinations of mileage, location, driving history, engine, age and storage. Each company can attract the motorists its own formula prices cheaply. Several insurers can truthfully report large savings among switchers. None needs to be cheapest for most people.

The phrase doing the work is “customers who switched”. It sounds like a minor qualification and removes everybody who did not save. The large number belongs to a filtered group; the advert lets an unfiltered viewer imagine it belongs to them.

The same selection improves success rates elsewhere. A course asks graduates who completed it and answered the survey. A diet programme photographs members who remained. A dating service displays couples who stayed together. The calculation excludes the exits, refusals and failures before producing the advertised rate.

List every group entering the calculation. How many people requested a quote? How many received a lower one? How many declined, abandoned the process or never answered? Show savings across that full sequence and identify the group represented by the headline figure. The reported saving applies to people who received an attractive quote and completed the switch.

Behavioural principles

Behavioural ideas at play in this post

Short, plain-English explanations of the principles behind this post, with links to related books and examples in the archive.