Analysis of the oneBlast eNewsletter experiment
Setup
You should create a directory called 'data' in the working directory.
You can create the data files in that directory by downloading the raw data files from the secure file storage location.and then doing: Rscript -e "library(knitr);knit('datasetup.Rmd')" at the command line. If you want an .html file or a .pdf file that you can view easily, it is easier to use the "Knit to HTML" or "Knit to PDF" or "Compile PDF" button in RStudio. Alternatively, you can do Rscript -e "library(rmarkdown);render('datasetup.Rmd') to make an html file.
The subscription outcome arises from finding subscribers in the subscriptions.csv file who match members of the experimental pool in the designdata.csv file. The file createoutcomes.Rmd does this merging. In the end, roughly 1000 subscribers are not matchable with the experimental pool. We do not know whether these people were not a part of the experiment at all or whether they received emails as a part of the study but then subscribed using different names and different email addresses. We address this problem in the analysis.
HTML Version of these Results
The Google Analytics outcomes are analytized in gastats_analysis.Rmd.
HTML Version of these Results
Files
The source code is in R markdown format. To recreate our analysis you would either use the Knit to HTML button in RStudio or, after installing the rmarkdown package, do library(rmarkdown);render("oneblast_analysis.Rmd").
Plan and Process
We wrote up a plan for analysis using Google Docs to enable easy commenting and group editing. Once we had a more or less final version, we downloaded it as .docx format and then converted this to github flavored markdown so that it could be posted publically online: pandoc -f docx -t markdown_github AnalysisPlan.docx -o AnalysisPlan.md.