xamplify sucks

Sample Size and Xamplify

We all, or at least X% of us with 3-4% margin of error, have heard of those polls/surveys which use responses from about 1000 people to extrapolate and make claims about the entire population of the USA of about 280 million.

It surprises many of us that such a small sample can give rise to such accurate predictions about the entire population but the theory behind this is well-established. However, a point to keep in mind is that using this methodology we can make accurate predictions about the population proportions only and not about a particular individual.

Xamplify's claim was that it could accurately predict the psychological profile ( psychometrics in their terminology) of a person who has not taken the survey, based on the responses of about 1000 survey-takers. It used demographic and transactional data to create equations (using linear regression) which predicted these psychologic profiles. Besides a large group of wealthy, well-respected and influential backers, Xamplify had David Donoho, world's most cited mathematician of the past decade, a statistics professor at Stanford University, listed as the architect and auditor of its statistical methodologies. This fact should give even the most-hardened skeptics a long pause.

If Xamplify's claims stopped here, it would have been serious enough, but now Xamplify was claiming that it could accurately predict which individuals were engaging in illicit activities like money laundering. For such claims which could potentially destroy many innocent lives, Xamplify better have some solid proof that its methodology worked. Did its methodology work?

Want to know more?. Try the link below/

Home >Technology> Statistics