Models behind A/B/n tests
I'm a graduate student in psychology and would like to include data from an A/B/n Google analytics experiment in my dissertation. I've already collected the data from a large, public website, and am interested in reporting the results. I need to include information about the statistics behind the analyses. Does anyone know if there are publications on the models that Google is currently using? Is there any information about the statistical tests (the Analytics help section is helpful, but only explains basic Bayesian statistics - I need more information than that, if possible!)?
Thanks for any leads -
Re: Models behind A/B/n tests
Wow! That is a topic I would have never expected to research myself. Thanks for posting. I am definitely no expert in this field but I have gathered a few links for you that might help you further with the research or discovery process:
- This is Steven L. Scott (Steven L. Scott is a statistician primarily interested in the development and application of Bayesian methods. He received his PhD from the Harvard statistics department in 1998, and served on the faculty of USC's Marshall School of Business until 2007. After a brief stint in banking he joined Google in 2008.) Maybe you can read his papers and find more.
- These are the research papers published by Google. Maybe that will help you further to analyse and answer your related needs.
- INFERRING CAUSAL IMPACT USING BAYESIAN STRUCTURAL TIME-SERIES MODELS : this is a research paper published that has the keywords you mentioned.
- Overlapping Experiment Infrastructure
- Focusing on the Long-term
I hope it helps or at least points you to the right direction!! Good luck with your dissertation.
Was my response helpful? If yes, please mark it as the ‘Best Answer.’ Learn how here.