alt Hacker NewsWhen they send these 30-question surveys, surely they must be aware that the people who respond are not a random sample of the customer population but a sample of the subpopulation that is willing to take a 30-question survey for them?
When they send these 30-question surveys, surely they must be aware that the people who respond are not a random sample of the customer population but a sample of the subpopulation that is willing to take a 30-question survey for them?
Simple. Your mistake is assuming that these surveys used to gather actual information.
The 30 questions satisfy all of the bikeshedding smoothbrains in the survey-design-committee. The survey itself isn't used to make informed decisions to improve the product, but entirely to justify the manager's impact and thus everybody's bonuses.