Why All The Studies Are Right

[Latrinsorm]

Suppose we have six studies investigating the link between X and Y. They do this by calculating an odds ratio: split the population into two groups, those that have been exposed to X and those that have not. Divide the rate of Y in exposed groups by the rate of Y in unexposed groups, and you have an odds ratio. If this odds ratio is not different from the baseline to a statistically significant degree, then you cannot reject the null hypothesis, which states that there is no link between X and Y (odds ratio = 1). Here is a graph:

http://i.imgur.com/WNvT0sN.png

The first three studies reject the null hypothesis, the second three can't. But what if instead of the null hypothesis (odds ratio = 1) we test the hypothesis that the odds ratio is 1.5?

http://i.imgur.com/gPchRac.png

As we can see, none of the studies can reject that hypothesis. That is Why All the Studies Are Right, and that's why marijuana causes schizophrenia.

.

Values and sources:
1.3 ± .2, http://www.ncbi.nlm.nih.gov/pubmed/12446534
2.8 + 3.7 - 1.6, http://aje.oxfordjournals.org/content/156/4/319.full
1.7 + 0.8 - 0.6, http://www.bmj.com/content/early/2003/12/31/bmj.38267.664086.63
1.5 ± 1.5, http://psychcentral.com/news/2013/12/10/harvard-marijuana-doesnt-cause-schizophrenia/63148.html
2.5 ± 1.8, http://www.apa.org/pubs/journals/releases/adb-adb0000103.pdf
2.2 ± 1.2, http://www.jpain.org/article/S1526-5900(15)00837-8/fulltext#sec2.2

[Gelston]

Shut up, Latrin.

[Warriorbird]

I can't sell alcohol any more. It leads to this.