Many times I have to step in colleagues, customers, and alumni conversations to explain why statistical sampling works. First think I use to ask is ‘do you really think Government ask every country inhabitant to find out if s/he is employed, or looking for a job?’; that’s it my dear readers, unemployment figures, for instance, are gathered by means of ‘sampling’.
Then, when acknowledging what I say must be true, people tend to think how important and unique they are. ‘Wait a moment, probably your statistics can handle an estimation of who is working and who don’t, and so on, for the whole of the economy taking into account your… sampling you said?; however, you won’t be able to do it with my tastes or behaviour… so market research is wrong, hein!’.
Well, sorry to tell you that you are not that unique, you are part of a group, a cluster, a tribe; your preferences, your behaviour, those are casted in our brain by our evolution as social mammals. Whatever you think it makes you ‘the one’ and exclusive on the Earth, forget it, a whole group will have those characteristics as well. The only problem will be to find it, but leave that to us operations researchers, economists, and statisticians. Humm, in fact, stratified sampling will do, and a little bit of magic (OR mixes science and art ).