Xkcd statistics dating


My Stats coursework requires a hypothesis of me before I can take it any further.We need to use more than two factors (the ideal is three) in the actual investigation.



Well, you just stomped with both feet right into the middle of a massive controversy in statistics!Correlations, given a proper model and proper methods, do yield valid causal inference.It's all about finding the right instrumental variables.We can talk up randomized controlled double blind trials all we want, but as soon as kids start dropping dead from massive organ failure after coming down with a sniffle, we're pretty damn quick to start looking for causes of those effects with whatever observational data we can find (Reye's syndrome and acetaminophen)., do yield valid causal inference.

We can talk up randomized controlled double blind trials all we want, but as soon as kids start dropping dead from massive organ failure after coming down with a sniffle, we're pretty damn quick to start looking for causes of those effects with whatever observational data we can find (Reye's syndrome and acetaminophen).

Given the data and the description from the OP it doesn't seem likely that the method of collection was anything other than "I made all of my students fill out this survey".