Sampling Error

Never shoulda tried the warm shrimp at the mini-bar.

Sampling error is an error made in the sampling process that makes our measurements and/or results pretty much garbage. Sampling errors lead to samples that aren’t representative of the population as a whole. If we try to use the sample results to make a decision about the population, we’re probably buying a ticket on the Titanic.

What do we mean by “representative of the population”? Well, let’s say we’re trying to gain some insight into how smartphone app developers (and the stocks for their companies) have performed over the last five years. We probably can’t get info on every app developer, so we grab a sample, but our sample is from a list of iPhone developers. And we've only chosen 3 of them from which to draw our big fat conclusion. That n size sample of 3 is just not representative enough of the zillions of developers out there trying to strike it rich.

We’re ignoring the Android developers, and any results we get will only be predictive for iPhone apps rather than smartphone apps as a whole. Whatever our sample tells us to do or think, it's likely rife with error. We needed a way bigger n sample to make the data representatively meaningful. And now we're going to um, get rid of that bad shrimp.